diff --git "a/Llama-3-8b-hf/trainer_state.json" "b/Llama-3-8b-hf/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/Llama-3-8b-hf/trainer_state.json"
@@ -0,0 +1,27030 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.46424057978490185,
+  "eval_steps": 500,
+  "global_step": 4500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999997563060744e-05,
+      "loss": 1.3594,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999990252244153e-05,
+      "loss": 1.8721,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999978067553796e-05,
+      "loss": 1.5195,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999961008995607e-05,
+      "loss": 1.3225,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999939076577906e-05,
+      "loss": 1.1817,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999912270311376e-05,
+      "loss": 1.1352,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999988059020909e-05,
+      "loss": 1.0836,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999844036286483e-05,
+      "loss": 1.0594,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999980260856137e-05,
+      "loss": 0.9789,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999756307053947e-05,
+      "loss": 0.963,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999970513178678e-05,
+      "loss": 0.9293,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999649082784807e-05,
+      "loss": 0.9039,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999958816007535e-05,
+      "loss": 0.8917,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99995223636881e-05,
+      "loss": 0.8656,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999451693655125e-05,
+      "loss": 0.8557,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999376150010868e-05,
+      "loss": 0.8536,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999295732792146e-05,
+      "loss": 0.8173,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999210442038164e-05,
+      "loss": 0.8173,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999120277790477e-05,
+      "loss": 0.7977,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 0.8007,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998925328992175e-05,
+      "loss": 0.773,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999882054453657e-05,
+      "loss": 0.7681,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998710886777298e-05,
+      "loss": 0.7683,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998596355767805e-05,
+      "loss": 0.767,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998476951563914e-05,
+      "loss": 0.7683,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998352674223816e-05,
+      "loss": 0.7601,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998223523808092e-05,
+      "loss": 0.7464,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999808950037968e-05,
+      "loss": 0.755,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99979506040039e-05,
+      "loss": 0.7408,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9997806834748455e-05,
+      "loss": 0.7503,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9997658192683412e-05,
+      "loss": 0.745,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9997504677881224e-05,
+      "loss": 0.7402,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9997346290416703e-05,
+      "loss": 0.733,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999718303036705e-05,
+      "loss": 0.7311,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9997014897811834e-05,
+      "loss": 0.7308,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9996841892833e-05,
+      "loss": 0.71,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999666401551487e-05,
+      "loss": 0.714,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9996481265944146e-05,
+      "loss": 0.7151,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9996293644209886e-05,
+      "loss": 0.7151,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9996101150403543e-05,
+      "loss": 0.7177,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9995903784618936e-05,
+      "loss": 0.7152,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9995701546952252e-05,
+      "loss": 0.7071,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9995494437502064e-05,
+      "loss": 0.7143,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9995282456369313e-05,
+      "loss": 0.7071,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9995065603657317e-05,
+      "loss": 0.7008,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999484387947177e-05,
+      "loss": 0.7028,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999461728392073e-05,
+      "loss": 0.7172,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9994385817114644e-05,
+      "loss": 0.7071,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994149479166324e-05,
+      "loss": 0.7018,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 0.704,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999366219030611e-05,
+      "loss": 0.6851,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993411239631713e-05,
+      "loss": 0.6807,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999315541829008e-05,
+      "loss": 0.6887,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992894726405894e-05,
+      "loss": 0.6897,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999262916410621e-05,
+      "loss": 0.687,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999235873152047e-05,
+      "loss": 0.6847,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999208342878047e-05,
+      "loss": 0.6822,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991803256020393e-05,
+      "loss": 0.6813,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991518213376787e-05,
+      "loss": 0.6841,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.6815,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990933518997086e-05,
+      "loss": 0.6737,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990633867545956e-05,
+      "loss": 0.6697,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999032934678125e-05,
+      "loss": 0.6721,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990019956851384e-05,
+      "loss": 0.6836,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998970569790715e-05,
+      "loss": 0.6754,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989386570101716e-05,
+      "loss": 0.6728,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989062573590618e-05,
+      "loss": 0.6656,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9988733708531772e-05,
+      "loss": 0.6609,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998839997508546e-05,
+      "loss": 0.6821,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9988061373414342e-05,
+      "loss": 0.6593,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9987717903683447e-05,
+      "loss": 0.6699,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998736956606018e-05,
+      "loss": 0.6831,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9987016360714307e-05,
+      "loss": 0.6664,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998665828781799e-05,
+      "loss": 0.6718,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9986295347545738e-05,
+      "loss": 0.6543,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9985927540074453e-05,
+      "loss": 0.6684,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9985554865583394e-05,
+      "loss": 0.6693,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99851773242542e-05,
+      "loss": 0.6605,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9984794916270876e-05,
+      "loss": 0.6585,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9984407641819812e-05,
+      "loss": 0.6506,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998401550108975e-05,
+      "loss": 0.6551,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9983618494271825e-05,
+      "loss": 0.6596,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9983216621559525e-05,
+      "loss": 0.6507,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998280988314872e-05,
+      "loss": 0.6431,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9982398279237657e-05,
+      "loss": 0.6446,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9981981810026932e-05,
+      "loss": 0.6435,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998156047571954e-05,
+      "loss": 0.6497,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9981134276520828e-05,
+      "loss": 0.6525,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9980703212638522e-05,
+      "loss": 0.6507,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 0.6542,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997982649166588e-05,
+      "loss": 0.6474,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9979380835002846e-05,
+      "loss": 0.6428,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9978930314510826e-05,
+      "loss": 0.6533,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9978474930409396e-05,
+      "loss": 0.6479,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9978014682920503e-05,
+      "loss": 0.6454,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997754957226847e-05,
+      "loss": 0.6419,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9977079598679978e-05,
+      "loss": 0.6419,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99766047623841e-05,
+      "loss": 0.6497,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9976125063612254e-05,
+      "loss": 0.6427,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 0.6555,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9975151079578238e-05,
+      "loss": 0.6413,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9974656794790777e-05,
+      "loss": 0.632,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9974157648476768e-05,
+      "loss": 0.6371,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9973653640879486e-05,
+      "loss": 0.6315,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997314477224458e-05,
+      "loss": 0.6493,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997263104282007e-05,
+      "loss": 0.6408,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997211245285634e-05,
+      "loss": 0.6351,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997158900260614e-05,
+      "loss": 0.6219,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99710606923246e-05,
+      "loss": 0.645,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9970527522269204e-05,
+      "loss": 0.6434,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996998949269982e-05,
+      "loss": 0.6332,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9969446603878673e-05,
+      "loss": 0.6212,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996889885607036e-05,
+      "loss": 0.6306,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9968346249541848e-05,
+      "loss": 0.6315,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9967788784562474e-05,
+      "loss": 0.6271,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9967226461403934e-05,
+      "loss": 0.6325,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99666592803403e-05,
+      "loss": 0.6371,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996608724164801e-05,
+      "loss": 0.6402,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9965510345605866e-05,
+      "loss": 0.6373,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 0.6238,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996434198259908e-05,
+      "loss": 0.6347,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9963750516203887e-05,
+      "loss": 0.6399,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9963154193597728e-05,
+      "loss": 0.6286,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996255301507125e-05,
+      "loss": 0.6315,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 0.6234,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9961336091431728e-05,
+      "loss": 0.6378,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9960720346911798e-05,
+      "loss": 0.6362,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9960099747657774e-05,
+      "loss": 0.6201,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.995947429397213e-05,
+      "loss": 0.6332,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9958843986159705e-05,
+      "loss": 0.6274,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9958208824527702e-05,
+      "loss": 0.6285,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9957568809385693e-05,
+      "loss": 0.6297,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9956923941045613e-05,
+      "loss": 0.624,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.995627421982176e-05,
+      "loss": 0.6218,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99556196460308e-05,
+      "loss": 0.6261,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.995496021999177e-05,
+      "loss": 0.6302,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9954295942026065e-05,
+      "loss": 0.6234,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.995362681245744e-05,
+      "loss": 0.6306,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9952952831612027e-05,
+      "loss": 0.6229,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 0.6175,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9951590317407152e-05,
+      "loss": 0.6174,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9950901784711765e-05,
+      "loss": 0.6307,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9950208402067735e-05,
+      "loss": 0.6311,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9949510169813006e-05,
+      "loss": 0.6196,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9948807088287884e-05,
+      "loss": 0.6179,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.994809915783505e-05,
+      "loss": 0.6203,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9947386378799534e-05,
+      "loss": 0.6169,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9946668751528745e-05,
+      "loss": 0.6095,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9945946276372435e-05,
+      "loss": 0.6119,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 0.6199,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9944486783814135e-05,
+      "loss": 0.621,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.994374976712348e-05,
+      "loss": 0.6095,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.994300790396999e-05,
+      "loss": 0.6143,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9942261194715236e-05,
+      "loss": 0.6171,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9941509639723155e-05,
+      "loss": 0.612,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9940753239360047e-05,
+      "loss": 0.6144,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993999199399457e-05,
+      "loss": 0.6041,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9939225903997748e-05,
+      "loss": 0.6139,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993845496974297e-05,
+      "loss": 0.5947,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9937679191605964e-05,
+      "loss": 0.616,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993689856996485e-05,
+      "loss": 0.6077,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9936113105200085e-05,
+      "loss": 0.624,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99353227976945e-05,
+      "loss": 0.6176,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9934527647833276e-05,
+      "loss": 0.6034,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9933727656003964e-05,
+      "loss": 0.609,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993292282259647e-05,
+      "loss": 0.6232,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9932113148003057e-05,
+      "loss": 0.6039,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9931298632618355e-05,
+      "loss": 0.6126,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9930479276839347e-05,
+      "loss": 0.6069,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.992965508106537e-05,
+      "loss": 0.6133,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9928826045698138e-05,
+      "loss": 0.6053,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9927992171141707e-05,
+      "loss": 0.6169,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99271534578025e-05,
+      "loss": 0.6187,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.992630990608929e-05,
+      "loss": 0.61,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9925461516413224e-05,
+      "loss": 0.6081,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9924608289187786e-05,
+      "loss": 0.6207,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9923750224828833e-05,
+      "loss": 0.6056,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.992288732375458e-05,
+      "loss": 0.6158,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9922019586385587e-05,
+      "loss": 0.5991,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.6118,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9920269604457444e-05,
+      "loss": 0.6051,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9919387360751216e-05,
+      "loss": 0.6166,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.991850028245609e-05,
+      "loss": 0.6074,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9917608370004417e-05,
+      "loss": 0.6103,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9916711623830904e-05,
+      "loss": 0.6073,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9915810044372618e-05,
+      "loss": 0.6033,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9914903632068975e-05,
+      "loss": 0.6204,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9913992387361747e-05,
+      "loss": 0.5979,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9913076310695068e-05,
+      "loss": 0.6011,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.991215540251542e-05,
+      "loss": 0.6094,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.991122966327164e-05,
+      "loss": 0.6074,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.991029909341493e-05,
+      "loss": 0.6062,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9909363693398828e-05,
+      "loss": 0.6062,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9908423463679246e-05,
+      "loss": 0.5966,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9907478404714438e-05,
+      "loss": 0.6059,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.990652851696501e-05,
+      "loss": 0.6042,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.990557380089393e-05,
+      "loss": 0.607,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9904614256966514e-05,
+      "loss": 0.5953,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.990364988565043e-05,
+      "loss": 0.5947,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 0.6013,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.990170666273471e-05,
+      "loss": 0.5941,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9900727812082177e-05,
+      "loss": 0.6028,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.989974413593518e-05,
+      "loss": 0.6058,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.989875563477316e-05,
+      "loss": 0.6094,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.989776230907789e-05,
+      "loss": 0.5982,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.989676415933351e-05,
+      "loss": 0.5995,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.989576118602651e-05,
+      "loss": 0.6046,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9894753389645723e-05,
+      "loss": 0.6014,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9893740770682334e-05,
+      "loss": 0.5982,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9892723329629885e-05,
+      "loss": 0.5894,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9891701066984264e-05,
+      "loss": 0.6078,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9890673983243708e-05,
+      "loss": 0.6097,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9889642078908805e-05,
+      "loss": 0.5928,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9888605354482494e-05,
+      "loss": 0.6024,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.988756381047006e-05,
+      "loss": 0.6013,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.988651744737914e-05,
+      "loss": 0.5921,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9885466265719723e-05,
+      "loss": 0.5958,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9884410266004134e-05,
+      "loss": 0.5872,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.988334944874706e-05,
+      "loss": 0.6046,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.988228381446553e-05,
+      "loss": 0.5937,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.988121336367892e-05,
+      "loss": 0.5984,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9880138096908955e-05,
+      "loss": 0.5902,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9879058014679704e-05,
+      "loss": 0.5941,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.987797311751759e-05,
+      "loss": 0.5997,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 0.5924,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9875788880512183e-05,
+      "loss": 0.5958,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9874689541733455e-05,
+      "loss": 0.6016,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9873585390151003e-05,
+      "loss": 0.5903,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9872476426302983e-05,
+      "loss": 0.5882,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.987136265072988e-05,
+      "loss": 0.5935,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.987024406397454e-05,
+      "loss": 0.5913,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9869120666582153e-05,
+      "loss": 0.5942,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.986799245910024e-05,
+      "loss": 0.5932,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.986685944207868e-05,
+      "loss": 0.5955,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9865721616069695e-05,
+      "loss": 0.5844,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9864578981627844e-05,
+      "loss": 0.6025,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9863431539310033e-05,
+      "loss": 0.5971,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.986227928967551e-05,
+      "loss": 0.5994,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9861122233285873e-05,
+      "loss": 0.5994,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 0.5805,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9858793702499322e-05,
+      "loss": 0.5878,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9857622229237315e-05,
+      "loss": 0.585,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9856445951489984e-05,
+      "loss": 0.601,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.985526486983063e-05,
+      "loss": 0.6061,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9854078984834904e-05,
+      "loss": 0.5972,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.985288829708079e-05,
+      "loss": 0.5954,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9851692807148612e-05,
+      "loss": 0.5877,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9850492515621038e-05,
+      "loss": 0.5934,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.984928742308308e-05,
+      "loss": 0.5913,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.5937,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9846862837327733e-05,
+      "loss": 0.6008,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9845643345292055e-05,
+      "loss": 0.5862,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9844419054609418e-05,
+      "loss": 0.5931,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9843189965876525e-05,
+      "loss": 0.5888,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.984195607969242e-05,
+      "loss": 0.5862,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9840717396658483e-05,
+      "loss": 0.5832,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9839473917378432e-05,
+      "loss": 0.5881,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.983822564245833e-05,
+      "loss": 0.5933,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9836972572506557e-05,
+      "loss": 0.5921,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.983571470813386e-05,
+      "loss": 0.5981,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.98344520499533e-05,
+      "loss": 0.5902,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.983318459858028e-05,
+      "loss": 0.5832,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9831912354632537e-05,
+      "loss": 0.5781,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9830635318730155e-05,
+      "loss": 0.5951,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9829353491495545e-05,
+      "loss": 0.5962,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982806687355345e-05,
+      "loss": 0.5894,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982677546553095e-05,
+      "loss": 0.5827,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982547926805747e-05,
+      "loss": 0.5923,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9824178281764753e-05,
+      "loss": 0.5782,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982287250728689e-05,
+      "loss": 0.5741,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9821561945260292e-05,
+      "loss": 0.5839,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982024659632372e-05,
+      "loss": 0.5846,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9818926461118254e-05,
+      "loss": 0.5921,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.981760154028731e-05,
+      "loss": 0.5878,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9816271834476642e-05,
+      "loss": 0.5808,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.981493734433433e-05,
+      "loss": 0.5853,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.981359807051079e-05,
+      "loss": 0.5867,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.981225401365877e-05,
+      "loss": 0.5841,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.981090517443334e-05,
+      "loss": 0.5819,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9809551553491918e-05,
+      "loss": 0.5899,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9808193151494233e-05,
+      "loss": 0.5768,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9806829969102356e-05,
+      "loss": 0.5776,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9805462006980688e-05,
+      "loss": 0.5842,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.980408926579596e-05,
+      "loss": 0.5788,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9802711746217222e-05,
+      "loss": 0.5809,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9801329448915863e-05,
+      "loss": 0.5862,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9799942374565597e-05,
+      "loss": 0.5902,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.979855052384247e-05,
+      "loss": 0.5847,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9797153897424854e-05,
+      "loss": 0.5827,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.979575249599344e-05,
+      "loss": 0.5862,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9794346320231265e-05,
+      "loss": 0.5937,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9792935370823676e-05,
+      "loss": 0.5816,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9791519648458352e-05,
+      "loss": 0.5806,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.97900991538253e-05,
+      "loss": 0.5799,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9788673887616852e-05,
+      "loss": 0.579,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9787243850527663e-05,
+      "loss": 0.5861,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.978580904325472e-05,
+      "loss": 0.5903,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9784369466497333e-05,
+      "loss": 0.5822,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9782925120957123e-05,
+      "loss": 0.587,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.5809,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9780022126346413e-05,
+      "loss": 0.5758,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.977856347869079e-05,
+      "loss": 0.575,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.977710006508212e-05,
+      "loss": 0.5747,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9775631886233655e-05,
+      "loss": 0.5833,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9774158942860962e-05,
+      "loss": 0.5815,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9772681235681936e-05,
+      "loss": 0.5724,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.97711987654168e-05,
+      "loss": 0.5902,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9769711532788083e-05,
+      "loss": 0.5706,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.976821953852065e-05,
+      "loss": 0.5793,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9766722783341682e-05,
+      "loss": 0.5867,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9765221267980675e-05,
+      "loss": 0.5744,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.976371499316945e-05,
+      "loss": 0.5894,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.976220395964215e-05,
+      "loss": 0.5609,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9760688168135233e-05,
+      "loss": 0.5845,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9759167619387474e-05,
+      "loss": 0.578,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9757642314139977e-05,
+      "loss": 0.5835,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9756112253136154e-05,
+      "loss": 0.581,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9754577437121733e-05,
+      "loss": 0.582,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.975303786684477e-05,
+      "loss": 0.5813,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9751493543055634e-05,
+      "loss": 0.5739,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9749944466507007e-05,
+      "loss": 0.5735,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.974839063795389e-05,
+      "loss": 0.583,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9746832058153602e-05,
+      "loss": 0.5688,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9745268727865774e-05,
+      "loss": 0.5832,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743700647852356e-05,
+      "loss": 0.5734,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9742127818877605e-05,
+      "loss": 0.579,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.974055024170811e-05,
+      "loss": 0.5785,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9738967917112752e-05,
+      "loss": 0.5808,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9737380845862745e-05,
+      "loss": 0.5737,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9735789028731603e-05,
+      "loss": 0.5721,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9734192466495162e-05,
+      "loss": 0.5635,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9732591159931564e-05,
+      "loss": 0.5857,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9730985109821268e-05,
+      "loss": 0.5877,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972937431694704e-05,
+      "loss": 0.568,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972775878209397e-05,
+      "loss": 0.5763,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9726138506049438e-05,
+      "loss": 0.5655,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9724513489603153e-05,
+      "loss": 0.5748,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9722883733547128e-05,
+      "loss": 0.571,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9721249238675688e-05,
+      "loss": 0.5658,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9719610005785466e-05,
+      "loss": 0.5791,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.97179660356754e-05,
+      "loss": 0.5799,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.971631732914674e-05,
+      "loss": 0.5771,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9714663887003055e-05,
+      "loss": 0.575,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9713005710050203e-05,
+      "loss": 0.5744,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.971134279909636e-05,
+      "loss": 0.5677,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9709675154952017e-05,
+      "loss": 0.5742,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9708002778429957e-05,
+      "loss": 0.5655,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9706325670345276e-05,
+      "loss": 0.5827,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9704643831515377e-05,
+      "loss": 0.5748,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9702957262759964e-05,
+      "loss": 0.5711,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.970126596490106e-05,
+      "loss": 0.5754,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9699569938762975e-05,
+      "loss": 0.5739,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.969786918517233e-05,
+      "loss": 0.5772,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.969616370495806e-05,
+      "loss": 0.5625,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9694453498951392e-05,
+      "loss": 0.5634,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9692738567985853e-05,
+      "loss": 0.5776,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9691018912897285e-05,
+      "loss": 0.5704,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.968929453452383e-05,
+      "loss": 0.5587,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9687565433705926e-05,
+      "loss": 0.5867,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.5734,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.968409306811004e-05,
+      "loss": 0.5765,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9682349805024447e-05,
+      "loss": 0.5709,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.968060182287918e-05,
+      "loss": 0.5672,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.967884912252619e-05,
+      "loss": 0.5644,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9677091704819714e-05,
+      "loss": 0.5675,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.96753295706163e-05,
+      "loss": 0.5589,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9673562720774792e-05,
+      "loss": 0.5644,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.967179115615633e-05,
+      "loss": 0.5709,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9670014877624353e-05,
+      "loss": 0.5635,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9668233886044597e-05,
+      "loss": 0.5619,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9666448182285095e-05,
+      "loss": 0.5692,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9664657767216176e-05,
+      "loss": 0.5707,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.966286264171047e-05,
+      "loss": 0.5754,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9661062806642903e-05,
+      "loss": 0.5636,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.5644,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9657449011333328e-05,
+      "loss": 0.561,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9655635052852648e-05,
+      "loss": 0.5822,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.965381638833274e-05,
+      "loss": 0.5675,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9651993018660002e-05,
+      "loss": 0.5714,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9650164944723116e-05,
+      "loss": 0.5615,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9648332167413067e-05,
+      "loss": 0.5663,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9646494687623135e-05,
+      "loss": 0.5794,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9644652506248872e-05,
+      "loss": 0.5783,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.964280562418815e-05,
+      "loss": 0.563,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.96409540423411e-05,
+      "loss": 0.5661,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9639097761610174e-05,
+      "loss": 0.5629,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.96372367829001e-05,
+      "loss": 0.5593,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.963537110711789e-05,
+      "loss": 0.5719,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.963350073517285e-05,
+      "loss": 0.5732,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9631625667976584e-05,
+      "loss": 0.5645,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9629745906442973e-05,
+      "loss": 0.572,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.962786145148819e-05,
+      "loss": 0.5647,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9625972304030697e-05,
+      "loss": 0.5726,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.962407846499124e-05,
+      "loss": 0.5624,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9622179935292855e-05,
+      "loss": 0.5692,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.962027671586086e-05,
+      "loss": 0.5724,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9618368807622863e-05,
+      "loss": 0.5641,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9616456211508756e-05,
+      "loss": 0.5676,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.961453892845071e-05,
+      "loss": 0.5727,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 0.5621,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.961069030524294e-05,
+      "loss": 0.5554,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9608758966968987e-05,
+      "loss": 0.5684,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9606822945502642e-05,
+      "loss": 0.5653,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.96048822417875e-05,
+      "loss": 0.5627,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9602936856769432e-05,
+      "loss": 0.5656,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.96009867913966e-05,
+      "loss": 0.5573,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9599032046619437e-05,
+      "loss": 0.5654,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9597072623390668e-05,
+      "loss": 0.5742,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.959510852266529e-05,
+      "loss": 0.5628,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9593139745400575e-05,
+      "loss": 0.5691,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9591166292556093e-05,
+      "loss": 0.5668,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.958918816509367e-05,
+      "loss": 0.5677,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9587205363977428e-05,
+      "loss": 0.5529,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.958521789017376e-05,
+      "loss": 0.5644,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9583225744651334e-05,
+      "loss": 0.5629,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.95812289283811e-05,
+      "loss": 0.559,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9579227442336276e-05,
+      "loss": 0.5683,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9577221287492368e-05,
+      "loss": 0.5585,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.957521046482715e-05,
+      "loss": 0.5669,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.5591,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9571174819955264e-05,
+      "loss": 0.5575,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9569149999715514e-05,
+      "loss": 0.5559,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9567120515588307e-05,
+      "loss": 0.5597,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.956508636856278e-05,
+      "loss": 0.5588,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9563047559630356e-05,
+      "loss": 0.5516,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9561004089784726e-05,
+      "loss": 0.5576,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9558955960021847e-05,
+      "loss": 0.5646,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9556903171339963e-05,
+      "loss": 0.5632,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9554845724739565e-05,
+      "loss": 0.5645,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9552783621223437e-05,
+      "loss": 0.5596,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9550716861796623e-05,
+      "loss": 0.5613,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9548645447466433e-05,
+      "loss": 0.5514,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9546569379242446e-05,
+      "loss": 0.5775,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9544488658136522e-05,
+      "loss": 0.5545,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.954240328516277e-05,
+      "loss": 0.5653,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.954031326133758e-05,
+      "loss": 0.5529,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9538218587679605e-05,
+      "loss": 0.5529,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9536119265209763e-05,
+      "loss": 0.5618,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9534015294951235e-05,
+      "loss": 0.5619,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9531906677929472e-05,
+      "loss": 0.5529,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.952979341517219e-05,
+      "loss": 0.5594,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9527675507709368e-05,
+      "loss": 0.5611,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9525552956573244e-05,
+      "loss": 0.5603,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9523425762798328e-05,
+      "loss": 0.5598,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9521293927421388e-05,
+      "loss": 0.5508,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9519157451481453e-05,
+      "loss": 0.5614,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9517016336019817e-05,
+      "loss": 0.5511,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.951487058208003e-05,
+      "loss": 0.5576,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9512720190707915e-05,
+      "loss": 0.5576,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.5643,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9508405499861235e-05,
+      "loss": 0.5653,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.95062412024896e-05,
+      "loss": 0.5619,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9504072271891486e-05,
+      "loss": 0.5666,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.950189870912401e-05,
+      "loss": 0.5604,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9499720515246524e-05,
+      "loss": 0.5651,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.949753769132067e-05,
+      "loss": 0.5607,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.949535023841032e-05,
+      "loss": 0.5582,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9493158157581617e-05,
+      "loss": 0.5421,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9490961449902946e-05,
+      "loss": 0.5528,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9488760116444966e-05,
+      "loss": 0.5549,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9486554158280576e-05,
+      "loss": 0.5571,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9484343576484935e-05,
+      "loss": 0.563,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9482128372135446e-05,
+      "loss": 0.552,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9479908546311783e-05,
+      "loss": 0.5459,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.947768410009586e-05,
+      "loss": 0.5638,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.947545503457184e-05,
+      "loss": 0.5642,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9473221350826145e-05,
+      "loss": 0.5659,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9470983049947446e-05,
+      "loss": 0.5615,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.946874013302666e-05,
+      "loss": 0.5532,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9466492601156964e-05,
+      "loss": 0.5632,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9464240455433775e-05,
+      "loss": 0.5504,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.946198369695476e-05,
+      "loss": 0.5507,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.945972232681984e-05,
+      "loss": 0.5567,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.945745634613117e-05,
+      "loss": 0.5429,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.945518575599317e-05,
+      "loss": 0.5596,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9452910557512497e-05,
+      "loss": 0.5468,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.945063075179805e-05,
+      "loss": 0.5637,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9448346339960984e-05,
+      "loss": 0.5516,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.944605732311469e-05,
+      "loss": 0.5514,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.944376370237481e-05,
+      "loss": 0.5523,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.944146547885923e-05,
+      "loss": 0.5607,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9439162653688066e-05,
+      "loss": 0.5575,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9436855227983695e-05,
+      "loss": 0.5505,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9434543202870726e-05,
+      "loss": 0.5469,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.943222657947601e-05,
+      "loss": 0.5536,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9429905358928648e-05,
+      "loss": 0.5442,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9427579542359966e-05,
+      "loss": 0.5514,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9425249130903544e-05,
+      "loss": 0.5564,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.942291412569519e-05,
+      "loss": 0.5625,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.942057452787297e-05,
+      "loss": 0.5528,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9418230338577164e-05,
+      "loss": 0.5461,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9415881558950302e-05,
+      "loss": 0.5451,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9413528190137158e-05,
+      "loss": 0.5538,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9411170233284728e-05,
+      "loss": 0.5535,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9408807689542257e-05,
+      "loss": 0.5731,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9406440560061214e-05,
+      "loss": 0.5575,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9404068845995317e-05,
+      "loss": 0.5547,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9401692548500504e-05,
+      "loss": 0.5579,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9399311668734957e-05,
+      "loss": 0.5542,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.5453,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9394536167035535e-05,
+      "loss": 0.5624,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9392141547429183e-05,
+      "loss": 0.5521,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.938974235020714e-05,
+      "loss": 0.5549,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9387338576538743e-05,
+      "loss": 0.5614,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.938493022759556e-05,
+      "loss": 0.5553,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9382517304551397e-05,
+      "loss": 0.5516,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9380099808582278e-05,
+      "loss": 0.5463,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.937767774086646e-05,
+      "loss": 0.5485,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9375251102584438e-05,
+      "loss": 0.5555,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.937281989491892e-05,
+      "loss": 0.5513,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.937038411905484e-05,
+      "loss": 0.5442,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.936794377617938e-05,
+      "loss": 0.5575,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9365498867481926e-05,
+      "loss": 0.5578,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9363049394154095e-05,
+      "loss": 0.5639,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9360595357389735e-05,
+      "loss": 0.5602,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.935813675838491e-05,
+      "loss": 0.5553,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9355673598337916e-05,
+      "loss": 0.556,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.935320587844926e-05,
+      "loss": 0.5574,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9350733599921684e-05,
+      "loss": 0.5507,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9348256763960146e-05,
+      "loss": 0.5565,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9345775371771826e-05,
+      "loss": 0.5551,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9343289424566122e-05,
+      "loss": 0.5519,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9340798923554657e-05,
+      "loss": 0.5548,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.933830386995127e-05,
+      "loss": 0.5553,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9335804264972018e-05,
+      "loss": 0.5399,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9333300109835182e-05,
+      "loss": 0.5423,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9330791405761254e-05,
+      "loss": 0.5576,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9328278153972947e-05,
+      "loss": 0.5466,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.932576035569519e-05,
+      "loss": 0.5456,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9323238012155125e-05,
+      "loss": 0.5543,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.932071112458211e-05,
+      "loss": 0.5408,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9318179694207726e-05,
+      "loss": 0.5533,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.931564372226576e-05,
+      "loss": 0.5408,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9313103209992205e-05,
+      "loss": 0.5507,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9310558158625286e-05,
+      "loss": 0.545,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9308008569405424e-05,
+      "loss": 0.5623,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.930545444357526e-05,
+      "loss": 0.5544,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9302895782379648e-05,
+      "loss": 0.5485,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9300332587065644e-05,
+      "loss": 0.5501,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.545,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9295192599081747e-05,
+      "loss": 0.5517,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9292615808917027e-05,
+      "loss": 0.5434,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9290034489644247e-05,
+      "loss": 0.5518,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9287448642521513e-05,
+      "loss": 0.5555,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9284858268809135e-05,
+      "loss": 0.5432,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9282263369769633e-05,
+      "loss": 0.54,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.927966394666773e-05,
+      "loss": 0.5452,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9277060000770342e-05,
+      "loss": 0.5421,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9274451533346617e-05,
+      "loss": 0.55,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9271838545667876e-05,
+      "loss": 0.5452,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9269221039007666e-05,
+      "loss": 0.5465,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9266599014641724e-05,
+      "loss": 0.5613,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9263972473847995e-05,
+      "loss": 0.5458,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9261341417906622e-05,
+      "loss": 0.5452,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.925870584809995e-05,
+      "loss": 0.5337,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9256065765712524e-05,
+      "loss": 0.5477,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9253421172031086e-05,
+      "loss": 0.5456,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.925077206834458e-05,
+      "loss": 0.5438,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9248118455944153e-05,
+      "loss": 0.5495,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9245460336123136e-05,
+      "loss": 0.5526,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.924279771017706e-05,
+      "loss": 0.5481,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.924013057940367e-05,
+      "loss": 0.553,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.923745894510288e-05,
+      "loss": 0.5462,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9234782808576823e-05,
+      "loss": 0.5549,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.923210217112981e-05,
+      "loss": 0.5465,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9229417034068352e-05,
+      "loss": 0.5486,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.922672739870115e-05,
+      "loss": 0.5371,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9224033266339103e-05,
+      "loss": 0.5566,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9221334638295296e-05,
+      "loss": 0.561,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9218631515885007e-05,
+      "loss": 0.5431,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.921592390042571e-05,
+      "loss": 0.5444,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9213211793237056e-05,
+      "loss": 0.5563,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9210495195640895e-05,
+      "loss": 0.5592,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9207774108961273e-05,
+      "loss": 0.5516,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9205048534524405e-05,
+      "loss": 0.5448,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9202318473658707e-05,
+      "loss": 0.5448,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9199583927694775e-05,
+      "loss": 0.5466,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9196844897965393e-05,
+      "loss": 0.5378,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.919410138580553e-05,
+      "loss": 0.5388,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9191353392552346e-05,
+      "loss": 0.5478,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9188600919545176e-05,
+      "loss": 0.5466,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9185843968125543e-05,
+      "loss": 0.5523,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.918308253963715e-05,
+      "loss": 0.5518,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9180316635425883e-05,
+      "loss": 0.5452,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9177546256839814e-05,
+      "loss": 0.5381,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9174771405229187e-05,
+      "loss": 0.5408,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9171992081946436e-05,
+      "loss": 0.54,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9169208288346168e-05,
+      "loss": 0.5566,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9166420025785165e-05,
+      "loss": 0.5348,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9163627295622397e-05,
+      "loss": 0.5505,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9160830099219007e-05,
+      "loss": 0.5497,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9158028437938316e-05,
+      "loss": 0.541,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9155222313145817e-05,
+      "loss": 0.5354,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9152411726209176e-05,
+      "loss": 0.5385,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.914959667849825e-05,
+      "loss": 0.541,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.914677717138505e-05,
+      "loss": 0.5496,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9143953206243778e-05,
+      "loss": 0.5461,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.914112478445079e-05,
+      "loss": 0.536,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9138291907384632e-05,
+      "loss": 0.549,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.5493,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9132612792957808e-05,
+      "loss": 0.5507,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9129766558365076e-05,
+      "loss": 0.5297,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.912691587403503e-05,
+      "loss": 0.5474,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9124060741357065e-05,
+      "loss": 0.5406,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9121201161722732e-05,
+      "loss": 0.5474,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.911833713652576e-05,
+      "loss": 0.5513,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9115468667162038e-05,
+      "loss": 0.5307,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9112595755029625e-05,
+      "loss": 0.5505,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9109718401528742e-05,
+      "loss": 0.553,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.910683660806177e-05,
+      "loss": 0.5422,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9103950376033276e-05,
+      "loss": 0.5373,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9101059706849957e-05,
+      "loss": 0.5353,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9098164601920702e-05,
+      "loss": 0.5419,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9095265062656546e-05,
+      "loss": 0.5516,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9092361090470688e-05,
+      "loss": 0.5427,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.908945268677849e-05,
+      "loss": 0.5449,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.908653985299747e-05,
+      "loss": 0.5432,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9083622590547313e-05,
+      "loss": 0.5449,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9080700900849855e-05,
+      "loss": 0.5464,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.907777478532909e-05,
+      "loss": 0.533,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.907484424541117e-05,
+      "loss": 0.5379,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.907190928252441e-05,
+      "loss": 0.5557,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.906896989809927e-05,
+      "loss": 0.5324,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.906602609356838e-05,
+      "loss": 0.5444,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9063077870366504e-05,
+      "loss": 0.5266,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9060125229930572e-05,
+      "loss": 0.5519,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9057168173699664e-05,
+      "loss": 0.5408,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.905420670311502e-05,
+      "loss": 0.5376,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9051240819620018e-05,
+      "loss": 0.5453,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9048270524660197e-05,
+      "loss": 0.5514,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.904529581968324e-05,
+      "loss": 0.5386,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9042316706138987e-05,
+      "loss": 0.5394,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.903933318547942e-05,
+      "loss": 0.5382,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9036345259158667e-05,
+      "loss": 0.5356,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.903335292863301e-05,
+      "loss": 0.5327,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9030356195360875e-05,
+      "loss": 0.539,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.902735506080283e-05,
+      "loss": 0.5402,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9024349526421596e-05,
+      "loss": 0.5411,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.902133959368203e-05,
+      "loss": 0.5365,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.901832526405114e-05,
+      "loss": 0.5345,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.901530653899807e-05,
+      "loss": 0.5468,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9012283419994115e-05,
+      "loss": 0.5482,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9009255908512704e-05,
+      "loss": 0.5355,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9006224006029404e-05,
+      "loss": 0.5408,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9003187714021936e-05,
+      "loss": 0.5408,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9000147033970148e-05,
+      "loss": 0.5414,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.899710196735603e-05,
+      "loss": 0.5418,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.899405251566371e-05,
+      "loss": 0.5383,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8990998680379458e-05,
+      "loss": 0.5314,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8987940462991673e-05,
+      "loss": 0.5382,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8984877864990888e-05,
+      "loss": 0.5363,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8981810887869784e-05,
+      "loss": 0.5388,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.897873953312317e-05,
+      "loss": 0.5335,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8975663802247978e-05,
+      "loss": 0.5359,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8972583696743284e-05,
+      "loss": 0.5426,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8969499218110302e-05,
+      "loss": 0.5283,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.896641036785236e-05,
+      "loss": 0.5345,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.896331714747493e-05,
+      "loss": 0.5419,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.896021955848561e-05,
+      "loss": 0.5379,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.5415,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.895401128071234e-05,
+      "loss": 0.5386,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8950900594954226e-05,
+      "loss": 0.545,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8947785546635905e-05,
+      "loss": 0.544,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.89446661372756e-05,
+      "loss": 0.5272,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8941542368393683e-05,
+      "loss": 0.5367,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.893841424151264e-05,
+      "loss": 0.5488,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.893528175815708e-05,
+      "loss": 0.5434,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.893214491985374e-05,
+      "loss": 0.535,
+      "step": 668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.892900372813147e-05,
+      "loss": 0.5448,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.892585818452126e-05,
+      "loss": 0.5431,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8922708290556197e-05,
+      "loss": 0.5397,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8919554047771508e-05,
+      "loss": 0.539,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8916395457704536e-05,
+      "loss": 0.5431,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8913232521894734e-05,
+      "loss": 0.5424,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.5278,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.890689361921507e-05,
+      "loss": 0.5334,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8903717655434708e-05,
+      "loss": 0.5349,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8900537352090523e-05,
+      "loss": 0.538,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8897352710732564e-05,
+      "loss": 0.5271,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.889416373291298e-05,
+      "loss": 0.5374,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8890970420186035e-05,
+      "loss": 0.5359,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8887772774108116e-05,
+      "loss": 0.5389,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.888457079623772e-05,
+      "loss": 0.5438,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8881364488135448e-05,
+      "loss": 0.5376,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8878153851364013e-05,
+      "loss": 0.529,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.887493888748825e-05,
+      "loss": 0.5423,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8871719598075083e-05,
+      "loss": 0.5324,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.886849598469356e-05,
+      "loss": 0.5347,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8865268048914828e-05,
+      "loss": 0.5336,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8862035792312148e-05,
+      "loss": 0.5438,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8858799216460883e-05,
+      "loss": 0.5372,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8855558322938492e-05,
+      "loss": 0.5365,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8852313113324553e-05,
+      "loss": 0.5263,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8849063589200744e-05,
+      "loss": 0.5369,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.884580975215084e-05,
+      "loss": 0.5365,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8842551603760725e-05,
+      "loss": 0.5393,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8839289145618378e-05,
+      "loss": 0.5353,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8836022379313884e-05,
+      "loss": 0.5369,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.883275130643942e-05,
+      "loss": 0.5247,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 0.5373,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.882619624735982e-05,
+      "loss": 0.5323,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8822912264349535e-05,
+      "loss": 0.5431,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8819623981158996e-05,
+      "loss": 0.5425,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.881633139939087e-05,
+      "loss": 0.5314,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8813034520649923e-05,
+      "loss": 0.5357,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8809733346543013e-05,
+      "loss": 0.5245,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.880642787867909e-05,
+      "loss": 0.5325,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8803118118669203e-05,
+      "loss": 0.5335,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8799804068126487e-05,
+      "loss": 0.5383,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.879648572866617e-05,
+      "loss": 0.5393,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8793163101905562e-05,
+      "loss": 0.5338,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.878983618946409e-05,
+      "loss": 0.535,
+      "step": 712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.878650499296323e-05,
+      "loss": 0.5369,
+      "step": 713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.878316951402658e-05,
+      "loss": 0.5448,
+      "step": 714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8779829754279806e-05,
+      "loss": 0.537,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8776485715350672e-05,
+      "loss": 0.5345,
+      "step": 716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8773137398869017e-05,
+      "loss": 0.5626,
+      "step": 717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8769784806466768e-05,
+      "loss": 0.5368,
+      "step": 718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8766427939777943e-05,
+      "loss": 0.5266,
+      "step": 719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 0.5301,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8759701390087026e-05,
+      "loss": 0.5394,
+      "step": 721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8756331710363375e-05,
+      "loss": 0.5312,
+      "step": 722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8752957762910016e-05,
+      "loss": 0.5313,
+      "step": 723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.874957954937138e-05,
+      "loss": 0.5473,
+      "step": 724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.874619707139396e-05,
+      "loss": 0.541,
+      "step": 725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8742810330626338e-05,
+      "loss": 0.5345,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.873941932871917e-05,
+      "loss": 0.5329,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8736024067325188e-05,
+      "loss": 0.5472,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8732624548099204e-05,
+      "loss": 0.5255,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8729220772698096e-05,
+      "loss": 0.5385,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8725812742780832e-05,
+      "loss": 0.535,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8722400460008437e-05,
+      "loss": 0.5403,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.871898392604402e-05,
+      "loss": 0.5299,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8715563142552758e-05,
+      "loss": 0.533,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8712138111201898e-05,
+      "loss": 0.5346,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8708708833660755e-05,
+      "loss": 0.5221,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8705275311600724e-05,
+      "loss": 0.5311,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.870183754669526e-05,
+      "loss": 0.5287,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8698395540619883e-05,
+      "loss": 0.5368,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.869494929505219e-05,
+      "loss": 0.5453,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.869149881167184e-05,
+      "loss": 0.5281,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8688044092160554e-05,
+      "loss": 0.5305,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8684585138202122e-05,
+      "loss": 0.5216,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8681121951482397e-05,
+      "loss": 0.5303,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8677654533689287e-05,
+      "loss": 0.5327,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8674182886512776e-05,
+      "loss": 0.5381,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.86707070116449e-05,
+      "loss": 0.5339,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8667226910779767e-05,
+      "loss": 0.5241,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.866374258561352e-05,
+      "loss": 0.5259,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.5432,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8656761269172645e-05,
+      "loss": 0.5303,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8653264281300622e-05,
+      "loss": 0.5321,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.864976307593271e-05,
+      "loss": 0.5285,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.864625765477535e-05,
+      "loss": 0.5291,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.864274801953705e-05,
+      "loss": 0.5349,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8639234171928355e-05,
+      "loss": 0.5281,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8635716113661876e-05,
+      "loss": 0.5335,
+      "step": 757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.863219384645227e-05,
+      "loss": 0.5285,
+      "step": 758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.862866737201625e-05,
+      "loss": 0.527,
+      "step": 759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8625136692072577e-05,
+      "loss": 0.5195,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.862160180834206e-05,
+      "loss": 0.5228,
+      "step": 761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.861806272254755e-05,
+      "loss": 0.5379,
+      "step": 762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8614519436413968e-05,
+      "loss": 0.5385,
+      "step": 763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8610971951668265e-05,
+      "loss": 0.5283,
+      "step": 764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.860742027003944e-05,
+      "loss": 0.5448,
+      "step": 765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8603864393258534e-05,
+      "loss": 0.5423,
+      "step": 766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.860030432305865e-05,
+      "loss": 0.5251,
+      "step": 767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8596740061174912e-05,
+      "loss": 0.5278,
+      "step": 768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8593171609344505e-05,
+      "loss": 0.5296,
+      "step": 769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8589598969306646e-05,
+      "loss": 0.5366,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8586022142802597e-05,
+      "loss": 0.5331,
+      "step": 771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8582441131575658e-05,
+      "loss": 0.5295,
+      "step": 772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8578855937371176e-05,
+      "loss": 0.5228,
+      "step": 773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8575266561936526e-05,
+      "loss": 0.5263,
+      "step": 774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8571673007021124e-05,
+      "loss": 0.5283,
+      "step": 775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.856807527437643e-05,
+      "loss": 0.5277,
+      "step": 776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8564473365755936e-05,
+      "loss": 0.5305,
+      "step": 777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8560867282915164e-05,
+      "loss": 0.5383,
+      "step": 778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8557257027611677e-05,
+      "loss": 0.5339,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.855364260160507e-05,
+      "loss": 0.526,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8550024006656967e-05,
+      "loss": 0.5197,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.854640124453103e-05,
+      "loss": 0.5313,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8542774316992953e-05,
+      "loss": 0.5344,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8539143225810453e-05,
+      "loss": 0.5303,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8535507972753275e-05,
+      "loss": 0.5291,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8531868559593205e-05,
+      "loss": 0.5252,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8528224988104044e-05,
+      "loss": 0.5307,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8524577260061628e-05,
+      "loss": 0.5161,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8520925377243812e-05,
+      "loss": 0.527,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.851726934143048e-05,
+      "loss": 0.5312,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8513609154403535e-05,
+      "loss": 0.5324,
+      "step": 791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.850994481794692e-05,
+      "loss": 0.5335,
+      "step": 792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.850627633384658e-05,
+      "loss": 0.5238,
+      "step": 793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8502603703890488e-05,
+      "loss": 0.5238,
+      "step": 794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.849892692986864e-05,
+      "loss": 0.5303,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8495246013573057e-05,
+      "loss": 0.5218,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8491560956797766e-05,
+      "loss": 0.5304,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.848787176133882e-05,
+      "loss": 0.5367,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.848417842899429e-05,
+      "loss": 0.5263,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 0.5276,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8476779360850833e-05,
+      "loss": 0.5386,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8473073628658123e-05,
+      "loss": 0.5179,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8469363766792258e-05,
+      "loss": 0.5306,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8465649777061377e-05,
+      "loss": 0.5246,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8461931661275642e-05,
+      "loss": 0.5177,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8458209421247208e-05,
+      "loss": 0.5226,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8454483058790254e-05,
+      "loss": 0.5299,
+      "step": 807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8450752575720967e-05,
+      "loss": 0.5327,
+      "step": 808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.844701797385753e-05,
+      "loss": 0.5339,
+      "step": 809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8443279255020153e-05,
+      "loss": 0.53,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8439536421031035e-05,
+      "loss": 0.5279,
+      "step": 811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.843578947371439e-05,
+      "loss": 0.5345,
+      "step": 812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8432038414896432e-05,
+      "loss": 0.5254,
+      "step": 813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842828324640539e-05,
+      "loss": 0.5177,
+      "step": 814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842452397007148e-05,
+      "loss": 0.5259,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8420760587726925e-05,
+      "loss": 0.5321,
+      "step": 816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8416993101205957e-05,
+      "loss": 0.5319,
+      "step": 817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8413221512344805e-05,
+      "loss": 0.5289,
+      "step": 818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8409445822981694e-05,
+      "loss": 0.517,
+      "step": 819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8405666034956842e-05,
+      "loss": 0.5361,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8401882150112485e-05,
+      "loss": 0.5307,
+      "step": 821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.839809417029283e-05,
+      "loss": 0.5371,
+      "step": 822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8394302097344103e-05,
+      "loss": 0.5308,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8390505933114503e-05,
+      "loss": 0.5343,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8386705679454243e-05,
+      "loss": 0.5351,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8382901338215515e-05,
+      "loss": 0.5273,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8379092911252515e-05,
+      "loss": 0.5277,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.837528040042142e-05,
+      "loss": 0.5267,
+      "step": 828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.83714638075804e-05,
+      "loss": 0.5331,
+      "step": 829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.836764313458962e-05,
+      "loss": 0.5193,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8363818383311226e-05,
+      "loss": 0.519,
+      "step": 831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8359989555609355e-05,
+      "loss": 0.541,
+      "step": 832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8356156653350138e-05,
+      "loss": 0.532,
+      "step": 833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8352319678401677e-05,
+      "loss": 0.5277,
+      "step": 834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8348478632634067e-05,
+      "loss": 0.526,
+      "step": 835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.834463351791939e-05,
+      "loss": 0.5327,
+      "step": 836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8340784336131715e-05,
+      "loss": 0.5258,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8336931089147076e-05,
+      "loss": 0.5357,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.83330737788435e-05,
+      "loss": 0.5371,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8329212407100996e-05,
+      "loss": 0.5299,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.832534697580155e-05,
+      "loss": 0.5198,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8321477486829128e-05,
+      "loss": 0.5154,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8317603942069665e-05,
+      "loss": 0.5308,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8313726343411085e-05,
+      "loss": 0.528,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8309844692743283e-05,
+      "loss": 0.5172,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.830595899195813e-05,
+      "loss": 0.5167,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.830206924294946e-05,
+      "loss": 0.5234,
+      "step": 847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.82981754476131e-05,
+      "loss": 0.5274,
+      "step": 848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8294277607846834e-05,
+      "loss": 0.5256,
+      "step": 849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8290375725550417e-05,
+      "loss": 0.5321,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.828646980262559e-05,
+      "loss": 0.5199,
+      "step": 851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8282559840976043e-05,
+      "loss": 0.5315,
+      "step": 852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8278645842507448e-05,
+      "loss": 0.5323,
+      "step": 853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.827472780912744e-05,
+      "loss": 0.5308,
+      "step": 854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.827080574274562e-05,
+      "loss": 0.5263,
+      "step": 855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8266879645273557e-05,
+      "loss": 0.5327,
+      "step": 856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.826294951862478e-05,
+      "loss": 0.5242,
+      "step": 857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8259015364714786e-05,
+      "loss": 0.5176,
+      "step": 858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.825507718546104e-05,
+      "loss": 0.5234,
+      "step": 859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8251134982782952e-05,
+      "loss": 0.5129,
+      "step": 860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8247188758601912e-05,
+      "loss": 0.5263,
+      "step": 861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.824323851484126e-05,
+      "loss": 0.5276,
+      "step": 862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8239284253426294e-05,
+      "loss": 0.5224,
+      "step": 863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8235325976284276e-05,
+      "loss": 0.5259,
+      "step": 864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8231363685344422e-05,
+      "loss": 0.5239,
+      "step": 865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.82273973825379e-05,
+      "loss": 0.5379,
+      "step": 866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8223427069797845e-05,
+      "loss": 0.5204,
+      "step": 867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8219452749059332e-05,
+      "loss": 0.5211,
+      "step": 868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8215474422259403e-05,
+      "loss": 0.5132,
+      "step": 869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.821149209133704e-05,
+      "loss": 0.5256,
+      "step": 870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.820750575823319e-05,
+      "loss": 0.5298,
+      "step": 871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8203515424890738e-05,
+      "loss": 0.5349,
+      "step": 872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8199521093254524e-05,
+      "loss": 0.5209,
+      "step": 873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.819552276527134e-05,
+      "loss": 0.5202,
+      "step": 874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.819152044288992e-05,
+      "loss": 0.5257,
+      "step": 875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8187514128060946e-05,
+      "loss": 0.5081,
+      "step": 876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.818350382273705e-05,
+      "loss": 0.5265,
+      "step": 877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8179489528872808e-05,
+      "loss": 0.5196,
+      "step": 878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.817547124842473e-05,
+      "loss": 0.5367,
+      "step": 879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8171448983351284e-05,
+      "loss": 0.5202,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8167422735612877e-05,
+      "loss": 0.5177,
+      "step": 881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.816339250717184e-05,
+      "loss": 0.5239,
+      "step": 882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.815935829999247e-05,
+      "loss": 0.5168,
+      "step": 883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8155320116040983e-05,
+      "loss": 0.5247,
+      "step": 884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.815127795728554e-05,
+      "loss": 0.5311,
+      "step": 885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.814723182569625e-05,
+      "loss": 0.5279,
+      "step": 886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.814318172324514e-05,
+      "loss": 0.5196,
+      "step": 887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8139127651906183e-05,
+      "loss": 0.5194,
+      "step": 888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.813506961365528e-05,
+      "loss": 0.5226,
+      "step": 889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8131007610470278e-05,
+      "loss": 0.5207,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.812694164433094e-05,
+      "loss": 0.5248,
+      "step": 891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.812287171721897e-05,
+      "loss": 0.521,
+      "step": 892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.811879783111801e-05,
+      "loss": 0.5283,
+      "step": 893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8114719988013612e-05,
+      "loss": 0.5215,
+      "step": 894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8110638189893267e-05,
+      "loss": 0.5199,
+      "step": 895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.81065524387464e-05,
+      "loss": 0.5206,
+      "step": 896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8102462736564355e-05,
+      "loss": 0.5148,
+      "step": 897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.80983690853404e-05,
+      "loss": 0.5243,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8094271487069733e-05,
+      "loss": 0.5135,
+      "step": 899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.5208,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8086064457378667e-05,
+      "loss": 0.5156,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8081955029958272e-05,
+      "loss": 0.5196,
+      "step": 902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8077841663491174e-05,
+      "loss": 0.526,
+      "step": 903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8073724359982184e-05,
+      "loss": 0.5272,
+      "step": 904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.806960312143802e-05,
+      "loss": 0.5237,
+      "step": 905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8065477949867327e-05,
+      "loss": 0.52,
+      "step": 906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.806134884728066e-05,
+      "loss": 0.5257,
+      "step": 907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8057215815690494e-05,
+      "loss": 0.5205,
+      "step": 908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8053078857111218e-05,
+      "loss": 0.5222,
+      "step": 909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.804893797355914e-05,
+      "loss": 0.5176,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8044793167052476e-05,
+      "loss": 0.53,
+      "step": 911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8040644439611348e-05,
+      "loss": 0.5209,
+      "step": 912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.80364917932578e-05,
+      "loss": 0.5251,
+      "step": 913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.803233523001578e-05,
+      "loss": 0.5249,
+      "step": 914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8028174751911147e-05,
+      "loss": 0.522,
+      "step": 915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.802401036097167e-05,
+      "loss": 0.5203,
+      "step": 916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.801984205922701e-05,
+      "loss": 0.5226,
+      "step": 917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8015669848708768e-05,
+      "loss": 0.53,
+      "step": 918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8011493731450412e-05,
+      "loss": 0.5178,
+      "step": 919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8007313709487334e-05,
+      "loss": 0.512,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8003129784856832e-05,
+      "loss": 0.5099,
+      "step": 921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7998941959598097e-05,
+      "loss": 0.5218,
+      "step": 922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.799475023575222e-05,
+      "loss": 0.5171,
+      "step": 923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.79905546153622e-05,
+      "loss": 0.5248,
+      "step": 924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.798635510047293e-05,
+      "loss": 0.5122,
+      "step": 925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7982151693131206e-05,
+      "loss": 0.5197,
+      "step": 926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7977944395385713e-05,
+      "loss": 0.516,
+      "step": 927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7973733209287036e-05,
+      "loss": 0.5199,
+      "step": 928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7969518136887664e-05,
+      "loss": 0.5266,
+      "step": 929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7965299180241963e-05,
+      "loss": 0.5153,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.796107634140621e-05,
+      "loss": 0.5211,
+      "step": 931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7956849622438554e-05,
+      "loss": 0.5175,
+      "step": 932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.795261902539906e-05,
+      "loss": 0.5196,
+      "step": 933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.794838455234966e-05,
+      "loss": 0.5148,
+      "step": 934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7944146205354182e-05,
+      "loss": 0.5228,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7939903986478354e-05,
+      "loss": 0.5177,
+      "step": 936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.793565789778978e-05,
+      "loss": 0.5153,
+      "step": 937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.793140794135795e-05,
+      "loss": 0.5249,
+      "step": 938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7927154119254234e-05,
+      "loss": 0.523,
+      "step": 939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.792289643355191e-05,
+      "loss": 0.5091,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.791863488632611e-05,
+      "loss": 0.5185,
+      "step": 941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7914369479653858e-05,
+      "loss": 0.511,
+      "step": 942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.791010021561407e-05,
+      "loss": 0.518,
+      "step": 943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7905827096287532e-05,
+      "loss": 0.5293,
+      "step": 944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7901550123756906e-05,
+      "loss": 0.5188,
+      "step": 945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.789726930010674e-05,
+      "loss": 0.5221,
+      "step": 946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.789298462742345e-05,
+      "loss": 0.5325,
+      "step": 947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7888696107795343e-05,
+      "loss": 0.5253,
+      "step": 948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7884403743312583e-05,
+      "loss": 0.5192,
+      "step": 949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.788010753606722e-05,
+      "loss": 0.517,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7875807488153173e-05,
+      "loss": 0.5168,
+      "step": 951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7871503601666233e-05,
+      "loss": 0.5121,
+      "step": 952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7867195878704062e-05,
+      "loss": 0.5074,
+      "step": 953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.786288432136619e-05,
+      "loss": 0.5113,
+      "step": 954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.785856893175402e-05,
+      "loss": 0.5201,
+      "step": 955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.785424971197082e-05,
+      "loss": 0.5158,
+      "step": 956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7849926664121726e-05,
+      "loss": 0.5264,
+      "step": 957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7845599790313735e-05,
+      "loss": 0.5216,
+      "step": 958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7841269092655714e-05,
+      "loss": 0.5251,
+      "step": 959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.78369345732584e-05,
+      "loss": 0.5246,
+      "step": 960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7832596234234376e-05,
+      "loss": 0.5312,
+      "step": 961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.78282540776981e-05,
+      "loss": 0.5229,
+      "step": 962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7823908105765883e-05,
+      "loss": 0.5173,
+      "step": 963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7819558320555902e-05,
+      "loss": 0.5138,
+      "step": 964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.781520472418819e-05,
+      "loss": 0.5211,
+      "step": 965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7810847318784632e-05,
+      "loss": 0.5117,
+      "step": 966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7806486106468983e-05,
+      "loss": 0.5252,
+      "step": 967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.780212108936684e-05,
+      "loss": 0.5244,
+      "step": 968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7797752269605654e-05,
+      "loss": 0.5143,
+      "step": 969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7793379649314743e-05,
+      "loss": 0.5146,
+      "step": 970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7789003230625266e-05,
+      "loss": 0.5214,
+      "step": 971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7784623015670237e-05,
+      "loss": 0.5169,
+      "step": 972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7780239006584515e-05,
+      "loss": 0.5264,
+      "step": 973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7775851205504823e-05,
+      "loss": 0.5149,
+      "step": 974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.777145961456971e-05,
+      "loss": 0.5294,
+      "step": 975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7767064235919594e-05,
+      "loss": 0.5236,
+      "step": 976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.776266507169672e-05,
+      "loss": 0.5159,
+      "step": 977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7758262124045195e-05,
+      "loss": 0.5166,
+      "step": 978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.775385539511096e-05,
+      "loss": 0.5144,
+      "step": 979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7749444887041797e-05,
+      "loss": 0.5138,
+      "step": 980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7745030601987338e-05,
+      "loss": 0.5183,
+      "step": 981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7740612542099054e-05,
+      "loss": 0.5225,
+      "step": 982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.773619070953025e-05,
+      "loss": 0.5165,
+      "step": 983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7731765106436073e-05,
+      "loss": 0.5128,
+      "step": 984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7727335734973512e-05,
+      "loss": 0.5205,
+      "step": 985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7722902597301385e-05,
+      "loss": 0.5123,
+      "step": 986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.771846569558035e-05,
+      "loss": 0.518,
+      "step": 987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7714025031972904e-05,
+      "loss": 0.5144,
+      "step": 988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7709580608643364e-05,
+      "loss": 0.5252,
+      "step": 989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7705132427757895e-05,
+      "loss": 0.5104,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.770068049148448e-05,
+      "loss": 0.5197,
+      "step": 991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7696224801992947e-05,
+      "loss": 0.5136,
+      "step": 992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.769176536145494e-05,
+      "loss": 0.5109,
+      "step": 993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7687302172043933e-05,
+      "loss": 0.5126,
+      "step": 994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7682835235935236e-05,
+      "loss": 0.5179,
+      "step": 995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.767836455530598e-05,
+      "loss": 0.5143,
+      "step": 996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.767389013233511e-05,
+      "loss": 0.5214,
+      "step": 997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7669411969203417e-05,
+      "loss": 0.5184,
+      "step": 998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.76649300680935e-05,
+      "loss": 0.5141,
+      "step": 999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.51,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7655955060678508e-05,
+      "loss": 0.5218,
+      "step": 1001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7651461958747745e-05,
+      "loss": 0.5163,
+      "step": 1002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7646965127587373e-05,
+      "loss": 0.5257,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.764246456938909e-05,
+      "loss": 0.5183,
+      "step": 1004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7637960286346423e-05,
+      "loss": 0.5136,
+      "step": 1005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.76334522806547e-05,
+      "loss": 0.5165,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7628940554511064e-05,
+      "loss": 0.5256,
+      "step": 1007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.762442511011448e-05,
+      "loss": 0.5144,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.761990594966572e-05,
+      "loss": 0.5239,
+      "step": 1009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.761538307536737e-05,
+      "loss": 0.5242,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.761085648942382e-05,
+      "loss": 0.515,
+      "step": 1011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7606326194041274e-05,
+      "loss": 0.5151,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.760179219142774e-05,
+      "loss": 0.5087,
+      "step": 1013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.759725448379305e-05,
+      "loss": 0.5174,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.759271307334881e-05,
+      "loss": 0.502,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7588167962308458e-05,
+      "loss": 0.5138,
+      "step": 1016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7583619152887222e-05,
+      "loss": 0.5096,
+      "step": 1017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7579066647302134e-05,
+      "loss": 0.5203,
+      "step": 1018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.757451044777204e-05,
+      "loss": 0.5217,
+      "step": 1019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7569950556517566e-05,
+      "loss": 0.5156,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.756538697576115e-05,
+      "loss": 0.5155,
+      "step": 1021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7560819707727034e-05,
+      "loss": 0.5082,
+      "step": 1022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7556248754641237e-05,
+      "loss": 0.5142,
+      "step": 1023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7551674118731592e-05,
+      "loss": 0.5103,
+      "step": 1024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7547095802227723e-05,
+      "loss": 0.5196,
+      "step": 1025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.754251380736104e-05,
+      "loss": 0.5234,
+      "step": 1026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7537928136364756e-05,
+      "loss": 0.512,
+      "step": 1027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7533338791473872e-05,
+      "loss": 0.5063,
+      "step": 1028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7528745774925175e-05,
+      "loss": 0.4994,
+      "step": 1029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7524149088957244e-05,
+      "loss": 0.5105,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7519548735810456e-05,
+      "loss": 0.5159,
+      "step": 1031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7514944717726962e-05,
+      "loss": 0.5083,
+      "step": 1032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7510337036950703e-05,
+      "loss": 0.5207,
+      "step": 1033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7505725695727414e-05,
+      "loss": 0.5135,
+      "step": 1034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7501110696304598e-05,
+      "loss": 0.5237,
+      "step": 1035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.749649204093155e-05,
+      "loss": 0.5181,
+      "step": 1036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7491869731859353e-05,
+      "loss": 0.5236,
+      "step": 1037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7487243771340862e-05,
+      "loss": 0.5166,
+      "step": 1038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7482614161630714e-05,
+      "loss": 0.5119,
+      "step": 1039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.747798090498532e-05,
+      "loss": 0.5171,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7473344003662877e-05,
+      "loss": 0.5205,
+      "step": 1041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.746870345992336e-05,
+      "loss": 0.5281,
+      "step": 1042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7464059276028497e-05,
+      "loss": 0.5074,
+      "step": 1043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7459411454241822e-05,
+      "loss": 0.505,
+      "step": 1044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7454759996828622e-05,
+      "loss": 0.5172,
+      "step": 1045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7450104906055963e-05,
+      "loss": 0.5108,
+      "step": 1046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7445446184192674e-05,
+      "loss": 0.5191,
+      "step": 1047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7440783833509366e-05,
+      "loss": 0.5169,
+      "step": 1048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.743611785627841e-05,
+      "loss": 0.5131,
+      "step": 1049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 0.5163,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7426775031271876e-05,
+      "loss": 0.5163,
+      "step": 1051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7422098188049885e-05,
+      "loss": 0.5101,
+      "step": 1052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7417417727387392e-05,
+      "loss": 0.5074,
+      "step": 1053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.741273365156561e-05,
+      "loss": 0.5081,
+      "step": 1054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.74080459628675e-05,
+      "loss": 0.5114,
+      "step": 1055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7403354663577782e-05,
+      "loss": 0.5122,
+      "step": 1056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7398659755982937e-05,
+      "loss": 0.5109,
+      "step": 1057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7393961242371203e-05,
+      "loss": 0.5135,
+      "step": 1058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.738925912503259e-05,
+      "loss": 0.5221,
+      "step": 1059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7384553406258842e-05,
+      "loss": 0.5086,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.737984408834347e-05,
+      "loss": 0.5123,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.737513117358174e-05,
+      "loss": 0.5016,
+      "step": 1062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7370414664270675e-05,
+      "loss": 0.5112,
+      "step": 1063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7365694562709034e-05,
+      "loss": 0.5055,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7360970871197347e-05,
+      "loss": 0.5096,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7356243592037876e-05,
+      "loss": 0.5012,
+      "step": 1066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7351512727534645e-05,
+      "loss": 0.5113,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7346778279993417e-05,
+      "loss": 0.5149,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7342040251721702e-05,
+      "loss": 0.5173,
+      "step": 1069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7337298645028764e-05,
+      "loss": 0.5046,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7332553462225604e-05,
+      "loss": 0.4996,
+      "step": 1071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.732780470562496e-05,
+      "loss": 0.5177,
+      "step": 1072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.732305237754132e-05,
+      "loss": 0.514,
+      "step": 1073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7318296480290912e-05,
+      "loss": 0.5051,
+      "step": 1074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7313537016191706e-05,
+      "loss": 0.5155,
+      "step": 1075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7308773987563406e-05,
+      "loss": 0.5199,
+      "step": 1076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.730400739672745e-05,
+      "loss": 0.5243,
+      "step": 1077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7299237246007018e-05,
+      "loss": 0.5175,
+      "step": 1078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7294463537727026e-05,
+      "loss": 0.5138,
+      "step": 1079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 0.5094,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7284905457796678e-05,
+      "loss": 0.5052,
+      "step": 1081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7280121090804813e-05,
+      "loss": 0.5105,
+      "step": 1082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.727533317557037e-05,
+      "loss": 0.5163,
+      "step": 1083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.727054171442692e-05,
+      "loss": 0.5123,
+      "step": 1084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7265746709709762e-05,
+      "loss": 0.5138,
+      "step": 1085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7260948163755918e-05,
+      "loss": 0.517,
+      "step": 1086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7256146078904153e-05,
+      "loss": 0.5141,
+      "step": 1087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7251340457494934e-05,
+      "loss": 0.5173,
+      "step": 1088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7246531301870467e-05,
+      "loss": 0.5222,
+      "step": 1089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7241718614374678e-05,
+      "loss": 0.5097,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7236902397353204e-05,
+      "loss": 0.5127,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7232082653153422e-05,
+      "loss": 0.5088,
+      "step": 1092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7227259384124408e-05,
+      "loss": 0.5207,
+      "step": 1093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.722243259261697e-05,
+      "loss": 0.5147,
+      "step": 1094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7217602280983622e-05,
+      "loss": 0.5162,
+      "step": 1095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.721276845157861e-05,
+      "loss": 0.5118,
+      "step": 1096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7207931106757867e-05,
+      "loss": 0.5102,
+      "step": 1097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.720309024887907e-05,
+      "loss": 0.516,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.719824588030159e-05,
+      "loss": 0.5072,
+      "step": 1099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7193398003386514e-05,
+      "loss": 0.51,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7188546620496634e-05,
+      "loss": 0.5234,
+      "step": 1101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7183691733996463e-05,
+      "loss": 0.5009,
+      "step": 1102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7178833346252208e-05,
+      "loss": 0.5206,
+      "step": 1103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.717397145963179e-05,
+      "loss": 0.5139,
+      "step": 1104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.716910607650483e-05,
+      "loss": 0.5043,
+      "step": 1105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.716423719924266e-05,
+      "loss": 0.5137,
+      "step": 1106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7159364830218312e-05,
+      "loss": 0.5145,
+      "step": 1107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.715448897180652e-05,
+      "loss": 0.5161,
+      "step": 1108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7149609626383718e-05,
+      "loss": 0.5106,
+      "step": 1109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7144726796328034e-05,
+      "loss": 0.5067,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.713984048401931e-05,
+      "loss": 0.5112,
+      "step": 1111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7134950691839063e-05,
+      "loss": 0.5096,
+      "step": 1112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.713005742217053e-05,
+      "loss": 0.5078,
+      "step": 1113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7125160677398625e-05,
+      "loss": 0.5169,
+      "step": 1114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.712026045990997e-05,
+      "loss": 0.5078,
+      "step": 1115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7115356772092858e-05,
+      "loss": 0.5122,
+      "step": 1116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.711044961633729e-05,
+      "loss": 0.5175,
+      "step": 1117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.710553899503496e-05,
+      "loss": 0.506,
+      "step": 1118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.710062491057925e-05,
+      "loss": 0.5117,
+      "step": 1119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.709570736536521e-05,
+      "loss": 0.5061,
+      "step": 1120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7090786361789602e-05,
+      "loss": 0.5101,
+      "step": 1121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7085861902250864e-05,
+      "loss": 0.5098,
+      "step": 1122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7080933989149112e-05,
+      "loss": 0.5051,
+      "step": 1123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7076002624886156e-05,
+      "loss": 0.5095,
+      "step": 1124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.5096,
+      "step": 1125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.706612955249225e-05,
+      "loss": 0.506,
+      "step": 1126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7061187849173318e-05,
+      "loss": 0.5016,
+      "step": 1127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.705624270431721e-05,
+      "loss": 0.508,
+      "step": 1128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7051294120334126e-05,
+      "loss": 0.5113,
+      "step": 1129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7046342099635948e-05,
+      "loss": 0.5056,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.704138664463623e-05,
+      "loss": 0.5048,
+      "step": 1131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7036427757750205e-05,
+      "loss": 0.5043,
+      "step": 1132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7031465441394766e-05,
+      "loss": 0.5107,
+      "step": 1133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7026499697988496e-05,
+      "loss": 0.5094,
+      "step": 1134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7021530529951627e-05,
+      "loss": 0.5016,
+      "step": 1135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7016557939706075e-05,
+      "loss": 0.5065,
+      "step": 1136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7011581929675424e-05,
+      "loss": 0.5118,
+      "step": 1137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.700660250228492e-05,
+      "loss": 0.5176,
+      "step": 1138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7001619659961467e-05,
+      "loss": 0.5154,
+      "step": 1139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6996633405133656e-05,
+      "loss": 0.5078,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6991643740231714e-05,
+      "loss": 0.5046,
+      "step": 1141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6986650667687552e-05,
+      "loss": 0.509,
+      "step": 1142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.698165418993473e-05,
+      "loss": 0.5173,
+      "step": 1143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6976654309408464e-05,
+      "loss": 0.5034,
+      "step": 1144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.697165102854565e-05,
+      "loss": 0.5061,
+      "step": 1145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.696664434978481e-05,
+      "loss": 0.5114,
+      "step": 1146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6961634275566147e-05,
+      "loss": 0.503,
+      "step": 1147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.695662080833151e-05,
+      "loss": 0.5146,
+      "step": 1148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.69516039505244e-05,
+      "loss": 0.5182,
+      "step": 1149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6946583704589973e-05,
+      "loss": 0.5035,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.694156007297504e-05,
+      "loss": 0.4986,
+      "step": 1151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.693653305812805e-05,
+      "loss": 0.507,
+      "step": 1152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6931502662499116e-05,
+      "loss": 0.5112,
+      "step": 1153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6926468888539988e-05,
+      "loss": 0.5038,
+      "step": 1154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.692143173870407e-05,
+      "loss": 0.4965,
+      "step": 1155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6916391215446403e-05,
+      "loss": 0.5042,
+      "step": 1156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.691134732122368e-05,
+      "loss": 0.5074,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.690630005849423e-05,
+      "loss": 0.5066,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6901249429718033e-05,
+      "loss": 0.5076,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.68961954373567e-05,
+      "loss": 0.512,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6891138083873486e-05,
+      "loss": 0.511,
+      "step": 1161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6886077371733285e-05,
+      "loss": 0.5042,
+      "step": 1162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.688101330340263e-05,
+      "loss": 0.5078,
+      "step": 1163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6875945881349676e-05,
+      "loss": 0.5119,
+      "step": 1164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6870875108044233e-05,
+      "loss": 0.5112,
+      "step": 1165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.686580098595773e-05,
+      "loss": 0.5106,
+      "step": 1166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6860723517563232e-05,
+      "loss": 0.5201,
+      "step": 1167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6855642705335438e-05,
+      "loss": 0.5137,
+      "step": 1168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.685055855175067e-05,
+      "loss": 0.5139,
+      "step": 1169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.684547105928689e-05,
+      "loss": 0.4994,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.684038023042367e-05,
+      "loss": 0.5101,
+      "step": 1171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6835286067642228e-05,
+      "loss": 0.5111,
+      "step": 1172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.683018857342539e-05,
+      "loss": 0.5062,
+      "step": 1173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6825087750257617e-05,
+      "loss": 0.5063,
+      "step": 1174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6819983600624986e-05,
+      "loss": 0.5122,
+      "step": 1175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.68148761270152e-05,
+      "loss": 0.5079,
+      "step": 1176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6809765331917576e-05,
+      "loss": 0.5033,
+      "step": 1177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6804651217823055e-05,
+      "loss": 0.5087,
+      "step": 1178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6799533787224192e-05,
+      "loss": 0.5052,
+      "step": 1179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6794413042615168e-05,
+      "loss": 0.5127,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6789288986491764e-05,
+      "loss": 0.4995,
+      "step": 1181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6784161621351384e-05,
+      "loss": 0.5104,
+      "step": 1182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6779030949693044e-05,
+      "loss": 0.5179,
+      "step": 1183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6773896974017373e-05,
+      "loss": 0.515,
+      "step": 1184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6768759696826608e-05,
+      "loss": 0.5097,
+      "step": 1185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6763619120624595e-05,
+      "loss": 0.5048,
+      "step": 1186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6758475247916786e-05,
+      "loss": 0.5135,
+      "step": 1187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6753328081210244e-05,
+      "loss": 0.5056,
+      "step": 1188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6748177623013638e-05,
+      "loss": 0.5059,
+      "step": 1189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6743023875837233e-05,
+      "loss": 0.5034,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6737866842192908e-05,
+      "loss": 0.507,
+      "step": 1191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6732706524594138e-05,
+      "loss": 0.4981,
+      "step": 1192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6727542925556e-05,
+      "loss": 0.5081,
+      "step": 1193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6722376047595163e-05,
+      "loss": 0.5106,
+      "step": 1194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6717205893229904e-05,
+      "loss": 0.495,
+      "step": 1195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6712032464980094e-05,
+      "loss": 0.4962,
+      "step": 1196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6706855765367202e-05,
+      "loss": 0.5137,
+      "step": 1197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6701675796914284e-05,
+      "loss": 0.5004,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6696492562145996e-05,
+      "loss": 0.5014,
+      "step": 1199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.5105,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6686116303769884e-05,
+      "loss": 0.5079,
+      "step": 1201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.668092328521932e-05,
+      "loss": 0.5218,
+      "step": 1202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.667572701046791e-05,
+      "loss": 0.5159,
+      "step": 1203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6670527482048246e-05,
+      "loss": 0.5068,
+      "step": 1204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6665324702494524e-05,
+      "loss": 0.5001,
+      "step": 1205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.666011867434252e-05,
+      "loss": 0.5133,
+      "step": 1206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6654909400129575e-05,
+      "loss": 0.5073,
+      "step": 1207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6649696882394635e-05,
+      "loss": 0.5062,
+      "step": 1208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.664448112367822e-05,
+      "loss": 0.5079,
+      "step": 1209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6639262126522417e-05,
+      "loss": 0.5137,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6634039893470912e-05,
+      "loss": 0.5039,
+      "step": 1211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6628814427068954e-05,
+      "loss": 0.5086,
+      "step": 1212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.662358572986337e-05,
+      "loss": 0.5069,
+      "step": 1213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6618353804402567e-05,
+      "loss": 0.5133,
+      "step": 1214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.661311865323652e-05,
+      "loss": 0.5052,
+      "step": 1215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6607880278916778e-05,
+      "loss": 0.5007,
+      "step": 1216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6602638683996462e-05,
+      "loss": 0.499,
+      "step": 1217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6597393871030264e-05,
+      "loss": 0.5025,
+      "step": 1218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6592145842574433e-05,
+      "loss": 0.5137,
+      "step": 1219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6586894601186804e-05,
+      "loss": 0.5025,
+      "step": 1220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6581640149426766e-05,
+      "loss": 0.4956,
+      "step": 1221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6576382489855274e-05,
+      "loss": 0.4982,
+      "step": 1222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6571121625034847e-05,
+      "loss": 0.5005,
+      "step": 1223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6565857557529567e-05,
+      "loss": 0.5145,
+      "step": 1224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6560590289905074e-05,
+      "loss": 0.5038,
+      "step": 1225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6555319824728577e-05,
+      "loss": 0.5086,
+      "step": 1226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6550046164568827e-05,
+      "loss": 0.5065,
+      "step": 1227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.654476931199615e-05,
+      "loss": 0.5018,
+      "step": 1228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6539489269582414e-05,
+      "loss": 0.5071,
+      "step": 1229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6534206039901057e-05,
+      "loss": 0.494,
+      "step": 1230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.652891962552705e-05,
+      "loss": 0.4968,
+      "step": 1231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.652363002903693e-05,
+      "loss": 0.5172,
+      "step": 1232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.651833725300879e-05,
+      "loss": 0.5051,
+      "step": 1233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6513041300022253e-05,
+      "loss": 0.5027,
+      "step": 1234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.650774217265851e-05,
+      "loss": 0.5087,
+      "step": 1235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.650243987350029e-05,
+      "loss": 0.4982,
+      "step": 1236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.649713440513187e-05,
+      "loss": 0.506,
+      "step": 1237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.649182577013906e-05,
+      "loss": 0.5067,
+      "step": 1238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6486513971109245e-05,
+      "loss": 0.5086,
+      "step": 1239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6481199010631312e-05,
+      "loss": 0.5091,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6475880891295716e-05,
+      "loss": 0.5054,
+      "step": 1241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6470559615694445e-05,
+      "loss": 0.5089,
+      "step": 1242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6465235186421024e-05,
+      "loss": 0.5071,
+      "step": 1243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6459907606070513e-05,
+      "loss": 0.5188,
+      "step": 1244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.645457687723951e-05,
+      "loss": 0.5083,
+      "step": 1245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6449243002526146e-05,
+      "loss": 0.5075,
+      "step": 1246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6443905984530092e-05,
+      "loss": 0.4984,
+      "step": 1247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.643856582585254e-05,
+      "loss": 0.5029,
+      "step": 1248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.643322252909622e-05,
+      "loss": 0.5069,
+      "step": 1249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.5069,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6422526531765846e-05,
+      "loss": 0.5114,
+      "step": 1251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6417173836404888e-05,
+      "loss": 0.5039,
+      "step": 1252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6411818013391357e-05,
+      "loss": 0.4977,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6406459065335616e-05,
+      "loss": 0.4989,
+      "step": 1254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6401096994849558e-05,
+      "loss": 0.5106,
+      "step": 1255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6395731804546582e-05,
+      "loss": 0.5049,
+      "step": 1256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.639036349704162e-05,
+      "loss": 0.4937,
+      "step": 1257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6384992074951124e-05,
+      "loss": 0.5069,
+      "step": 1258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6379617540893056e-05,
+      "loss": 0.4971,
+      "step": 1259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 0.5049,
+      "step": 1260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.636885914735365e-05,
+      "loss": 0.5106,
+      "step": 1261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6363475293115824e-05,
+      "loss": 0.5038,
+      "step": 1262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6358088337397444e-05,
+      "loss": 0.5002,
+      "step": 1263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6352698282824045e-05,
+      "loss": 0.5057,
+      "step": 1264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6347305132022677e-05,
+      "loss": 0.5074,
+      "step": 1265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6341908887621897e-05,
+      "loss": 0.5016,
+      "step": 1266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6336509552251766e-05,
+      "loss": 0.5009,
+      "step": 1267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6331107128543856e-05,
+      "loss": 0.5122,
+      "step": 1268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6325701619131246e-05,
+      "loss": 0.4941,
+      "step": 1269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.632029302664851e-05,
+      "loss": 0.5074,
+      "step": 1270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6314881353731733e-05,
+      "loss": 0.5076,
+      "step": 1271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6309466603018497e-05,
+      "loss": 0.5016,
+      "step": 1272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.630404877714789e-05,
+      "loss": 0.513,
+      "step": 1273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6298627878760488e-05,
+      "loss": 0.5009,
+      "step": 1274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6293203910498375e-05,
+      "loss": 0.5117,
+      "step": 1275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.628777687500513e-05,
+      "loss": 0.4886,
+      "step": 1276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6282346774925816e-05,
+      "loss": 0.5051,
+      "step": 1277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6276913612907005e-05,
+      "loss": 0.4965,
+      "step": 1278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6271477391596754e-05,
+      "loss": 0.4899,
+      "step": 1279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6266038113644605e-05,
+      "loss": 0.5001,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6260595781701605e-05,
+      "loss": 0.5101,
+      "step": 1281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6255150398420273e-05,
+      "loss": 0.5061,
+      "step": 1282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6249701966454626e-05,
+      "loss": 0.506,
+      "step": 1283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.624425048846016e-05,
+      "loss": 0.5092,
+      "step": 1284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6238795967093865e-05,
+      "loss": 0.4978,
+      "step": 1285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6233338405014204e-05,
+      "loss": 0.5082,
+      "step": 1286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6227877804881126e-05,
+      "loss": 0.5017,
+      "step": 1287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6222414169356066e-05,
+      "loss": 0.5106,
+      "step": 1288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.621694750110193e-05,
+      "loss": 0.4982,
+      "step": 1289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6211477802783105e-05,
+      "loss": 0.4997,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6206005077065457e-05,
+      "loss": 0.5018,
+      "step": 1291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.620052932661633e-05,
+      "loss": 0.4989,
+      "step": 1292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.619505055410453e-05,
+      "loss": 0.488,
+      "step": 1293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.618956876220035e-05,
+      "loss": 0.4898,
+      "step": 1294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6184083953575543e-05,
+      "loss": 0.501,
+      "step": 1295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6178596130903345e-05,
+      "loss": 0.5045,
+      "step": 1296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.617310529685845e-05,
+      "loss": 0.4916,
+      "step": 1297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6167611454117027e-05,
+      "loss": 0.5036,
+      "step": 1298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6162114605356704e-05,
+      "loss": 0.5008,
+      "step": 1299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6156614753256583e-05,
+      "loss": 0.4959,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6151111900497225e-05,
+      "loss": 0.5123,
+      "step": 1301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6145606049760644e-05,
+      "loss": 0.4938,
+      "step": 1302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.614009720373034e-05,
+      "loss": 0.4929,
+      "step": 1303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6134585365091243e-05,
+      "loss": 0.5066,
+      "step": 1304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6129070536529767e-05,
+      "loss": 0.5007,
+      "step": 1305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6123552720733767e-05,
+      "loss": 0.4933,
+      "step": 1306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.611803192039256e-05,
+      "loss": 0.509,
+      "step": 1307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.611250813819692e-05,
+      "loss": 0.5049,
+      "step": 1308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6106981376839064e-05,
+      "loss": 0.5067,
+      "step": 1309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.610145163901268e-05,
+      "loss": 0.5053,
+      "step": 1310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6095918927412883e-05,
+      "loss": 0.4973,
+      "step": 1311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6090383244736256e-05,
+      "loss": 0.4932,
+      "step": 1312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.608484459368082e-05,
+      "loss": 0.5014,
+      "step": 1313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6079302976946055e-05,
+      "loss": 0.5049,
+      "step": 1314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.607375839723287e-05,
+      "loss": 0.4952,
+      "step": 1315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6068210857243625e-05,
+      "loss": 0.4998,
+      "step": 1316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6062660359682124e-05,
+      "loss": 0.4968,
+      "step": 1317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6057106907253617e-05,
+      "loss": 0.5085,
+      "step": 1318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.605155050266478e-05,
+      "loss": 0.5022,
+      "step": 1319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6045991148623752e-05,
+      "loss": 0.5017,
+      "step": 1320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6040428847840078e-05,
+      "loss": 0.5043,
+      "step": 1321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6034863603024768e-05,
+      "loss": 0.5006,
+      "step": 1322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.602929541689025e-05,
+      "loss": 0.4999,
+      "step": 1323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6023724292150387e-05,
+      "loss": 0.4934,
+      "step": 1324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6018150231520486e-05,
+      "loss": 0.4979,
+      "step": 1325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.601257323771727e-05,
+      "loss": 0.4979,
+      "step": 1326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6006993313458896e-05,
+      "loss": 0.503,
+      "step": 1327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6001410461464955e-05,
+      "loss": 0.511,
+      "step": 1328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5995824684456465e-05,
+      "loss": 0.5058,
+      "step": 1329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.599023598515586e-05,
+      "loss": 0.5073,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5984644366287007e-05,
+      "loss": 0.5025,
+      "step": 1331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.597904983057519e-05,
+      "loss": 0.4971,
+      "step": 1332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5973452380747125e-05,
+      "loss": 0.501,
+      "step": 1333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.596785201953093e-05,
+      "loss": 0.4954,
+      "step": 1334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5962248749656158e-05,
+      "loss": 0.4957,
+      "step": 1335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5956642573853784e-05,
+      "loss": 0.4951,
+      "step": 1336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5951033494856174e-05,
+      "loss": 0.4955,
+      "step": 1337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5945421515397135e-05,
+      "loss": 0.4977,
+      "step": 1338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.593980663821187e-05,
+      "loss": 0.5036,
+      "step": 1339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5934188866037017e-05,
+      "loss": 0.502,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5928568201610593e-05,
+      "loss": 0.502,
+      "step": 1341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.592294464767205e-05,
+      "loss": 0.4996,
+      "step": 1342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.591731820696224e-05,
+      "loss": 0.494,
+      "step": 1343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.591168888222342e-05,
+      "loss": 0.5058,
+      "step": 1344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5906056676199256e-05,
+      "loss": 0.5053,
+      "step": 1345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5900421591634813e-05,
+      "loss": 0.5016,
+      "step": 1346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.589478363127657e-05,
+      "loss": 0.5003,
+      "step": 1347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5889142797872387e-05,
+      "loss": 0.4933,
+      "step": 1348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5883499094171556e-05,
+      "loss": 0.5019,
+      "step": 1349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.5089,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5872203086883996e-05,
+      "loss": 0.4966,
+      "step": 1351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5866550788802815e-05,
+      "loss": 0.5049,
+      "step": 1352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5860895631436044e-05,
+      "loss": 0.4907,
+      "step": 1353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5855237617539943e-05,
+      "loss": 0.5018,
+      "step": 1354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.584957674987216e-05,
+      "loss": 0.5009,
+      "step": 1355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5843913031191722e-05,
+      "loss": 0.501,
+      "step": 1356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.583824646425907e-05,
+      "loss": 0.5031,
+      "step": 1357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5832577051836016e-05,
+      "loss": 0.5032,
+      "step": 1358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5826904796685763e-05,
+      "loss": 0.4873,
+      "step": 1359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5821229701572897e-05,
+      "loss": 0.4869,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5815551769263387e-05,
+      "loss": 0.4932,
+      "step": 1361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5809871002524602e-05,
+      "loss": 0.4992,
+      "step": 1362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.580418740412526e-05,
+      "loss": 0.5055,
+      "step": 1363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5798500976835493e-05,
+      "loss": 0.4928,
+      "step": 1364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5792811723426787e-05,
+      "loss": 0.4939,
+      "step": 1365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5787119646672025e-05,
+      "loss": 0.501,
+      "step": 1366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5781424749345447e-05,
+      "loss": 0.5028,
+      "step": 1367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5775727034222675e-05,
+      "loss": 0.4863,
+      "step": 1368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.577002650408072e-05,
+      "loss": 0.4938,
+      "step": 1369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5764323161697933e-05,
+      "loss": 0.5047,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5758617009854068e-05,
+      "loss": 0.4915,
+      "step": 1371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.575290805133023e-05,
+      "loss": 0.4974,
+      "step": 1372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5747196288908887e-05,
+      "loss": 0.5054,
+      "step": 1373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.57414817253739e-05,
+      "loss": 0.4968,
+      "step": 1374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.573576436351046e-05,
+      "loss": 0.5065,
+      "step": 1375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5730044206105156e-05,
+      "loss": 0.5036,
+      "step": 1376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.572432125594591e-05,
+      "loss": 0.4983,
+      "step": 1377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5718595515822027e-05,
+      "loss": 0.502,
+      "step": 1378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5712866988524157e-05,
+      "loss": 0.4935,
+      "step": 1379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.570713567684432e-05,
+      "loss": 0.4939,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5701401583575883e-05,
+      "loss": 0.4967,
+      "step": 1381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5695664711513575e-05,
+      "loss": 0.4941,
+      "step": 1382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5689925063453483e-05,
+      "loss": 0.501,
+      "step": 1383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.568418264219303e-05,
+      "loss": 0.4993,
+      "step": 1384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5678437450531014e-05,
+      "loss": 0.498,
+      "step": 1385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.567268949126757e-05,
+      "loss": 0.4987,
+      "step": 1386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5666938767204173e-05,
+      "loss": 0.5074,
+      "step": 1387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5661185281143666e-05,
+      "loss": 0.5025,
+      "step": 1388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.565542903589023e-05,
+      "loss": 0.4981,
+      "step": 1389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.564967003424938e-05,
+      "loss": 0.4885,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5643908279027994e-05,
+      "loss": 0.4969,
+      "step": 1391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5638143773034268e-05,
+      "loss": 0.4949,
+      "step": 1392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.563237651907777e-05,
+      "loss": 0.5033,
+      "step": 1393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.562660651996937e-05,
+      "loss": 0.493,
+      "step": 1394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5620833778521306e-05,
+      "loss": 0.4961,
+      "step": 1395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5615058297547144e-05,
+      "loss": 0.5056,
+      "step": 1396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.560928007986178e-05,
+      "loss": 0.5035,
+      "step": 1397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5603499128281447e-05,
+      "loss": 0.499,
+      "step": 1398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5597715445623714e-05,
+      "loss": 0.4991,
+      "step": 1399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 0.5012,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.558613989835295e-05,
+      "loss": 0.5012,
+      "step": 1401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.55803480393817e-05,
+      "loss": 0.4981,
+      "step": 1402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5574553460616608e-05,
+      "loss": 0.4874,
+      "step": 1403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.556875616488188e-05,
+      "loss": 0.4974,
+      "step": 1404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.556295615500305e-05,
+      "loss": 0.4971,
+      "step": 1405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5557153433806967e-05,
+      "loss": 0.508,
+      "step": 1406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.555134800412181e-05,
+      "loss": 0.4916,
+      "step": 1407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5545539868777075e-05,
+      "loss": 0.4938,
+      "step": 1408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5539729030603574e-05,
+      "loss": 0.5004,
+      "step": 1409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.553391549243344e-05,
+      "loss": 0.489,
+      "step": 1410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5528099257100126e-05,
+      "loss": 0.5044,
+      "step": 1411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5522280327438388e-05,
+      "loss": 0.5049,
+      "step": 1412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5516458706284306e-05,
+      "loss": 0.4917,
+      "step": 1413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5510634396475262e-05,
+      "loss": 0.5061,
+      "step": 1414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5504807400849957e-05,
+      "loss": 0.4987,
+      "step": 1415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.54989777222484e-05,
+      "loss": 0.5068,
+      "step": 1416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.54931453635119e-05,
+      "loss": 0.4983,
+      "step": 1417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5487310327483087e-05,
+      "loss": 0.5008,
+      "step": 1418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5481472617005878e-05,
+      "loss": 0.4968,
+      "step": 1419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5475632234925505e-05,
+      "loss": 0.4909,
+      "step": 1420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5469789184088498e-05,
+      "loss": 0.4989,
+      "step": 1421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5463943467342694e-05,
+      "loss": 0.4969,
+      "step": 1422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5458095087537216e-05,
+      "loss": 0.5011,
+      "step": 1423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5452244047522504e-05,
+      "loss": 0.4972,
+      "step": 1424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5446390350150272e-05,
+      "loss": 0.4946,
+      "step": 1425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.544053399827355e-05,
+      "loss": 0.5041,
+      "step": 1426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.543467499474665e-05,
+      "loss": 0.5055,
+      "step": 1427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5428813342425177e-05,
+      "loss": 0.4962,
+      "step": 1428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.542294904416603e-05,
+      "loss": 0.4896,
+      "step": 1429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.54170821028274e-05,
+      "loss": 0.4943,
+      "step": 1430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.541121252126876e-05,
+      "loss": 0.492,
+      "step": 1431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.540534030235087e-05,
+      "loss": 0.496,
+      "step": 1432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5399465448935788e-05,
+      "loss": 0.5,
+      "step": 1433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5393587963886837e-05,
+      "loss": 0.4989,
+      "step": 1434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5387707850068633e-05,
+      "loss": 0.4978,
+      "step": 1435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5381825110347072e-05,
+      "loss": 0.4915,
+      "step": 1436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5375939747589334e-05,
+      "loss": 0.4901,
+      "step": 1437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5370051764663872e-05,
+      "loss": 0.4919,
+      "step": 1438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5364161164440413e-05,
+      "loss": 0.5009,
+      "step": 1439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 0.5053,
+      "step": 1440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5352372123584816e-05,
+      "loss": 0.4883,
+      "step": 1441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5346473688698514e-05,
+      "loss": 0.5006,
+      "step": 1442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5340572648005887e-05,
+      "loss": 0.4998,
+      "step": 1443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.533466900438303e-05,
+      "loss": 0.4922,
+      "step": 1444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.53287627607073e-05,
+      "loss": 0.5043,
+      "step": 1445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.532285391985734e-05,
+      "loss": 0.4965,
+      "step": 1446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5316942484713043e-05,
+      "loss": 0.4961,
+      "step": 1447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5311028458155567e-05,
+      "loss": 0.4957,
+      "step": 1448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5305111843067343e-05,
+      "loss": 0.4985,
+      "step": 1449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.529919264233205e-05,
+      "loss": 0.4861,
+      "step": 1450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5293270858834643e-05,
+      "loss": 0.4966,
+      "step": 1451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.528734649546132e-05,
+      "loss": 0.4893,
+      "step": 1452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5281419555099547e-05,
+      "loss": 0.4937,
+      "step": 1453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5275490040638038e-05,
+      "loss": 0.4962,
+      "step": 1454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5269557954966777e-05,
+      "loss": 0.4911,
+      "step": 1455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.526362330097698e-05,
+      "loss": 0.4922,
+      "step": 1456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5257686081561134e-05,
+      "loss": 0.4966,
+      "step": 1457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5251746299612959e-05,
+      "loss": 0.5055,
+      "step": 1458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5245803958027434e-05,
+      "loss": 0.5062,
+      "step": 1459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5239859059700794e-05,
+      "loss": 0.5055,
+      "step": 1460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5233911607530499e-05,
+      "loss": 0.5006,
+      "step": 1461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5227961604415266e-05,
+      "loss": 0.4892,
+      "step": 1462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5222009053255061e-05,
+      "loss": 0.4935,
+      "step": 1463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5216053956951081e-05,
+      "loss": 0.5048,
+      "step": 1464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5210096318405768e-05,
+      "loss": 0.4897,
+      "step": 1465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5204136140522799e-05,
+      "loss": 0.4829,
+      "step": 1466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5198173426207095e-05,
+      "loss": 0.4922,
+      "step": 1467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5192208178364815e-05,
+      "loss": 0.5007,
+      "step": 1468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5186240399903343e-05,
+      "loss": 0.4971,
+      "step": 1469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5180270093731305e-05,
+      "loss": 0.4931,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5174297262758551e-05,
+      "loss": 0.505,
+      "step": 1471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5168321909896171e-05,
+      "loss": 0.4922,
+      "step": 1472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5162344038056476e-05,
+      "loss": 0.4854,
+      "step": 1473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5156363650153012e-05,
+      "loss": 0.4949,
+      "step": 1474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5150380749100545e-05,
+      "loss": 0.4995,
+      "step": 1475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5144395337815066e-05,
+      "loss": 0.4855,
+      "step": 1476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5138407419213797e-05,
+      "loss": 0.4957,
+      "step": 1477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5132416996215171e-05,
+      "loss": 0.4917,
+      "step": 1478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5126424071738853e-05,
+      "loss": 0.4966,
+      "step": 1479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5120428648705716e-05,
+      "loss": 0.498,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.511443073003786e-05,
+      "loss": 0.4871,
+      "step": 1481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.51084303186586e-05,
+      "loss": 0.4883,
+      "step": 1482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.510242741749246e-05,
+      "loss": 0.484,
+      "step": 1483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5096422029465178e-05,
+      "loss": 0.4901,
+      "step": 1484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5090414157503715e-05,
+      "loss": 0.4994,
+      "step": 1485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.508440380453623e-05,
+      "loss": 0.4912,
+      "step": 1486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5078390973492094e-05,
+      "loss": 0.4888,
+      "step": 1487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5072375667301893e-05,
+      "loss": 0.4961,
+      "step": 1488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.506635788889741e-05,
+      "loss": 0.5053,
+      "step": 1489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5060337641211637e-05,
+      "loss": 0.496,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5054314927178779e-05,
+      "loss": 0.4877,
+      "step": 1491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.504828974973422e-05,
+      "loss": 0.4854,
+      "step": 1492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5042262111814566e-05,
+      "loss": 0.493,
+      "step": 1493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.503623201635761e-05,
+      "loss": 0.4904,
+      "step": 1494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5030199466302354e-05,
+      "loss": 0.4961,
+      "step": 1495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5024164464588982e-05,
+      "loss": 0.4973,
+      "step": 1496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5018127014158886e-05,
+      "loss": 0.5015,
+      "step": 1497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5012087117954643e-05,
+      "loss": 0.4959,
+      "step": 1498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5006044778920028e-05,
+      "loss": 0.5042,
+      "step": 1499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4845,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4993952784140716e-05,
+      "loss": 0.4928,
+      "step": 1501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.498790313428951e-05,
+      "loss": 0.4843,
+      "step": 1502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.498185105339491e-05,
+      "loss": 0.4878,
+      "step": 1503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4975796544406627e-05,
+      "loss": 0.4985,
+      "step": 1504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4969739610275556e-05,
+      "loss": 0.5016,
+      "step": 1505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.496368025395377e-05,
+      "loss": 0.493,
+      "step": 1506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4957618478394529e-05,
+      "loss": 0.488,
+      "step": 1507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4951554286552266e-05,
+      "loss": 0.4788,
+      "step": 1508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4945487681382597e-05,
+      "loss": 0.4954,
+      "step": 1509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.493941866584231e-05,
+      "loss": 0.4925,
+      "step": 1510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4933347242889371e-05,
+      "loss": 0.5001,
+      "step": 1511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4927273415482916e-05,
+      "loss": 0.4949,
+      "step": 1512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4921197186583256e-05,
+      "loss": 0.4962,
+      "step": 1513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4915118559151871e-05,
+      "loss": 0.5008,
+      "step": 1514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.490903753615141e-05,
+      "loss": 0.4954,
+      "step": 1515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4902954120545687e-05,
+      "loss": 0.4885,
+      "step": 1516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4896868315299692e-05,
+      "loss": 0.491,
+      "step": 1517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4890780123379565e-05,
+      "loss": 0.4863,
+      "step": 1518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.488468954775262e-05,
+      "loss": 0.4963,
+      "step": 1519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4878596591387329e-05,
+      "loss": 0.4878,
+      "step": 1520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4872501257253325e-05,
+      "loss": 0.4908,
+      "step": 1521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4866403548321402e-05,
+      "loss": 0.4892,
+      "step": 1522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4860303467563504e-05,
+      "loss": 0.4909,
+      "step": 1523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.485420101795274e-05,
+      "loss": 0.4919,
+      "step": 1524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4848096202463373e-05,
+      "loss": 0.4856,
+      "step": 1525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4841989024070809e-05,
+      "loss": 0.4985,
+      "step": 1526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4835879485751617e-05,
+      "loss": 0.5053,
+      "step": 1527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4829767590483508e-05,
+      "loss": 0.49,
+      "step": 1528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4823653341245353e-05,
+      "loss": 0.4906,
+      "step": 1529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4817536741017153e-05,
+      "loss": 0.4888,
+      "step": 1530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4811417792780074e-05,
+      "loss": 0.4854,
+      "step": 1531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4805296499516408e-05,
+      "loss": 0.4914,
+      "step": 1532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4799172864209607e-05,
+      "loss": 0.4911,
+      "step": 1533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4793046889844252e-05,
+      "loss": 0.4896,
+      "step": 1534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.478691857940607e-05,
+      "loss": 0.4994,
+      "step": 1535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4780787935881925e-05,
+      "loss": 0.4906,
+      "step": 1536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4774654962259813e-05,
+      "loss": 0.4892,
+      "step": 1537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4768519661528879e-05,
+      "loss": 0.4929,
+      "step": 1538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4762382036679393e-05,
+      "loss": 0.489,
+      "step": 1539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4756242090702756e-05,
+      "loss": 0.4932,
+      "step": 1540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.47500998265915e-05,
+      "loss": 0.4811,
+      "step": 1541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4743955247339292e-05,
+      "loss": 0.4906,
+      "step": 1542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4737808355940932e-05,
+      "loss": 0.4919,
+      "step": 1543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4731659155392332e-05,
+      "loss": 0.4855,
+      "step": 1544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4725507648690542e-05,
+      "loss": 0.4913,
+      "step": 1545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4719353838833729e-05,
+      "loss": 0.4946,
+      "step": 1546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4713197728821185e-05,
+      "loss": 0.4832,
+      "step": 1547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.470703932165333e-05,
+      "loss": 0.4893,
+      "step": 1548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4700878620331684e-05,
+      "loss": 0.4909,
+      "step": 1549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.469471562785891e-05,
+      "loss": 0.4905,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.468855034723877e-05,
+      "loss": 0.4934,
+      "step": 1551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4682382781476146e-05,
+      "loss": 0.4847,
+      "step": 1552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.467621293357704e-05,
+      "loss": 0.4903,
+      "step": 1553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4670040806548555e-05,
+      "loss": 0.4915,
+      "step": 1554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4663866403398915e-05,
+      "loss": 0.4889,
+      "step": 1555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4657689727137443e-05,
+      "loss": 0.4841,
+      "step": 1556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4651510780774585e-05,
+      "loss": 0.4938,
+      "step": 1557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.464532956732188e-05,
+      "loss": 0.4887,
+      "step": 1558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4639146089791972e-05,
+      "loss": 0.4959,
+      "step": 1559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.463296035119862e-05,
+      "loss": 0.4819,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.462677235455667e-05,
+      "loss": 0.4892,
+      "step": 1561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4620582102882088e-05,
+      "loss": 0.489,
+      "step": 1562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4614389599191917e-05,
+      "loss": 0.492,
+      "step": 1563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4608194846504311e-05,
+      "loss": 0.4925,
+      "step": 1564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4601997847838518e-05,
+      "loss": 0.4937,
+      "step": 1565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4595798606214882e-05,
+      "loss": 0.4916,
+      "step": 1566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4589597124654834e-05,
+      "loss": 0.4752,
+      "step": 1567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4583393406180898e-05,
+      "loss": 0.497,
+      "step": 1568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4577187453816702e-05,
+      "loss": 0.4932,
+      "step": 1569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4570979270586944e-05,
+      "loss": 0.4926,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4564768859517417e-05,
+      "loss": 0.4889,
+      "step": 1571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4558556223635004e-05,
+      "loss": 0.4926,
+      "step": 1572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.455234136596766e-05,
+      "loss": 0.5053,
+      "step": 1573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.454612428954444e-05,
+      "loss": 0.4949,
+      "step": 1574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 0.4838,
+      "step": 1575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4533683492551954e-05,
+      "loss": 0.4918,
+      "step": 1576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.452745977804618e-05,
+      "loss": 0.4913,
+      "step": 1577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4521233856911507e-05,
+      "loss": 0.4869,
+      "step": 1578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4515005732182384e-05,
+      "loss": 0.4923,
+      "step": 1579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4508775406894308e-05,
+      "loss": 0.4885,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4502542884083876e-05,
+      "loss": 0.4913,
+      "step": 1581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.449630816678874e-05,
+      "loss": 0.4821,
+      "step": 1582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4490071258047625e-05,
+      "loss": 0.4946,
+      "step": 1583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4483832160900326e-05,
+      "loss": 0.4927,
+      "step": 1584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4477590878387697e-05,
+      "loss": 0.5026,
+      "step": 1585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4471347413551673e-05,
+      "loss": 0.4832,
+      "step": 1586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4465101769435235e-05,
+      "loss": 0.4876,
+      "step": 1587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4458853949082443e-05,
+      "loss": 0.4877,
+      "step": 1588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4452603955538397e-05,
+      "loss": 0.4924,
+      "step": 1589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4446351791849276e-05,
+      "loss": 0.4887,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4440097461062308e-05,
+      "loss": 0.4925,
+      "step": 1591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4433840966225772e-05,
+      "loss": 0.4944,
+      "step": 1592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.442758231038902e-05,
+      "loss": 0.4772,
+      "step": 1593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4421321496602428e-05,
+      "loss": 0.4865,
+      "step": 1594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4415058527917454e-05,
+      "loss": 0.4919,
+      "step": 1595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4408793407386587e-05,
+      "loss": 0.4843,
+      "step": 1596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4402526138063373e-05,
+      "loss": 0.4945,
+      "step": 1597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.43962567230024e-05,
+      "loss": 0.4858,
+      "step": 1598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4389985165259308e-05,
+      "loss": 0.4968,
+      "step": 1599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4383711467890776e-05,
+      "loss": 0.4895,
+      "step": 1600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4377435633954528e-05,
+      "loss": 0.4954,
+      "step": 1601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.437115766650933e-05,
+      "loss": 0.4912,
+      "step": 1602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.436487756861499e-05,
+      "loss": 0.492,
+      "step": 1603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4358595343332342e-05,
+      "loss": 0.4914,
+      "step": 1604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4352310993723277e-05,
+      "loss": 0.4898,
+      "step": 1605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4346024522850704e-05,
+      "loss": 0.4952,
+      "step": 1606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4339735933778576e-05,
+      "loss": 0.494,
+      "step": 1607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4333445229571874e-05,
+      "loss": 0.476,
+      "step": 1608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4327152413296607e-05,
+      "loss": 0.4902,
+      "step": 1609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4320857488019826e-05,
+      "loss": 0.4949,
+      "step": 1610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4314560456809592e-05,
+      "loss": 0.4918,
+      "step": 1611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4308261322735006e-05,
+      "loss": 0.4925,
+      "step": 1612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4301960088866187e-05,
+      "loss": 0.4867,
+      "step": 1613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4295656758274283e-05,
+      "loss": 0.4806,
+      "step": 1614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4289351334031461e-05,
+      "loss": 0.5051,
+      "step": 1615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4283043819210905e-05,
+      "loss": 0.4881,
+      "step": 1616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4276734216886823e-05,
+      "loss": 0.4909,
+      "step": 1617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4270422530134433e-05,
+      "loss": 0.5004,
+      "step": 1618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4264108762029989e-05,
+      "loss": 0.4904,
+      "step": 1619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 0.4915,
+      "step": 1620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4251474994074927e-05,
+      "loss": 0.4865,
+      "step": 1621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.424515500038186e-05,
+      "loss": 0.4933,
+      "step": 1622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4238832937651816e-05,
+      "loss": 0.4837,
+      "step": 1623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4232508808966097e-05,
+      "loss": 0.4879,
+      "step": 1624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4226182617406996e-05,
+      "loss": 0.4771,
+      "step": 1625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4219854366057831e-05,
+      "loss": 0.4862,
+      "step": 1626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.421352405800291e-05,
+      "loss": 0.4872,
+      "step": 1627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.420719169632755e-05,
+      "loss": 0.4835,
+      "step": 1628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4200857284118067e-05,
+      "loss": 0.4849,
+      "step": 1629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4194520824461773e-05,
+      "loss": 0.4851,
+      "step": 1630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4188182320446985e-05,
+      "loss": 0.4858,
+      "step": 1631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4181841775163014e-05,
+      "loss": 0.4941,
+      "step": 1632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4175499191700169e-05,
+      "loss": 0.495,
+      "step": 1633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4169154573149737e-05,
+      "loss": 0.4976,
+      "step": 1634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4162807922604014e-05,
+      "loss": 0.4871,
+      "step": 1635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.415645924315628e-05,
+      "loss": 0.4873,
+      "step": 1636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4150108537900805e-05,
+      "loss": 0.4851,
+      "step": 1637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4143755809932843e-05,
+      "loss": 0.4827,
+      "step": 1638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4137401062348639e-05,
+      "loss": 0.4857,
+      "step": 1639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.413104429824542e-05,
+      "loss": 0.4978,
+      "step": 1640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4124685520721393e-05,
+      "loss": 0.4925,
+      "step": 1641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.411832473287575e-05,
+      "loss": 0.4881,
+      "step": 1642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4111961937808665e-05,
+      "loss": 0.4943,
+      "step": 1643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4105597138621281e-05,
+      "loss": 0.4873,
+      "step": 1644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4099230338415728e-05,
+      "loss": 0.4834,
+      "step": 1645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4092861540295109e-05,
+      "loss": 0.4869,
+      "step": 1646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4086490747363492e-05,
+      "loss": 0.4843,
+      "step": 1647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4080117962725929e-05,
+      "loss": 0.4884,
+      "step": 1648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4073743189488436e-05,
+      "loss": 0.4846,
+      "step": 1649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.492,
+      "step": 1650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4060987689642581e-05,
+      "loss": 0.4892,
+      "step": 1651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4054606969251095e-05,
+      "loss": 0.4824,
+      "step": 1652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4048224272693426e-05,
+      "loss": 0.4926,
+      "step": 1653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4041839603080423e-05,
+      "loss": 0.4922,
+      "step": 1654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4035452963523903e-05,
+      "loss": 0.4907,
+      "step": 1655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4029064357136628e-05,
+      "loss": 0.4811,
+      "step": 1656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4022673787032333e-05,
+      "loss": 0.4937,
+      "step": 1657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4016281256325702e-05,
+      "loss": 0.4825,
+      "step": 1658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4009886768132375e-05,
+      "loss": 0.4856,
+      "step": 1659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4003490325568953e-05,
+      "loss": 0.4781,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3997091931752978e-05,
+      "loss": 0.4937,
+      "step": 1661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3990691589802955e-05,
+      "loss": 0.4881,
+      "step": 1662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3984289302838327e-05,
+      "loss": 0.488,
+      "step": 1663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.39778850739795e-05,
+      "loss": 0.4828,
+      "step": 1664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3971478906347806e-05,
+      "loss": 0.4803,
+      "step": 1665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3965070803065543e-05,
+      "loss": 0.4791,
+      "step": 1666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3958660767255938e-05,
+      "loss": 0.4851,
+      "step": 1667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3952248802043166e-05,
+      "loss": 0.4818,
+      "step": 1668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.394583491055234e-05,
+      "loss": 0.4837,
+      "step": 1669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3939419095909513e-05,
+      "loss": 0.4929,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3933001361241674e-05,
+      "loss": 0.4792,
+      "step": 1671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3926581709676752e-05,
+      "loss": 0.4874,
+      "step": 1672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3920160144343604e-05,
+      "loss": 0.4811,
+      "step": 1673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3913736668372027e-05,
+      "loss": 0.4856,
+      "step": 1674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3907311284892737e-05,
+      "loss": 0.4883,
+      "step": 1675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3900883997037398e-05,
+      "loss": 0.4857,
+      "step": 1676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3894454807938587e-05,
+      "loss": 0.4831,
+      "step": 1677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.388802372072981e-05,
+      "loss": 0.4891,
+      "step": 1678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3881590738545508e-05,
+      "loss": 0.4898,
+      "step": 1679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3875155864521031e-05,
+      "loss": 0.4834,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3868719101792664e-05,
+      "loss": 0.4816,
+      "step": 1681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3862280453497601e-05,
+      "loss": 0.4882,
+      "step": 1682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3855839922773968e-05,
+      "loss": 0.4853,
+      "step": 1683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3849397512760797e-05,
+      "loss": 0.4861,
+      "step": 1684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3842953226598036e-05,
+      "loss": 0.4884,
+      "step": 1685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3836507067426565e-05,
+      "loss": 0.4824,
+      "step": 1686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3830059038388153e-05,
+      "loss": 0.4899,
+      "step": 1687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3823609142625492e-05,
+      "loss": 0.4937,
+      "step": 1688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3817157383282184e-05,
+      "loss": 0.4848,
+      "step": 1689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3810703763502744e-05,
+      "loss": 0.4869,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3804248286432577e-05,
+      "loss": 0.4836,
+      "step": 1691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3797790955218014e-05,
+      "loss": 0.4875,
+      "step": 1692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3791331773006272e-05,
+      "loss": 0.4882,
+      "step": 1693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3784870742945482e-05,
+      "loss": 0.4761,
+      "step": 1694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3778407868184674e-05,
+      "loss": 0.4899,
+      "step": 1695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.3771943151873768e-05,
+      "loss": 0.4828,
+      "step": 1696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3765476597163595e-05,
+      "loss": 0.4908,
+      "step": 1697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3759008207205869e-05,
+      "loss": 0.4811,
+      "step": 1698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.375253798515321e-05,
+      "loss": 0.4885,
+      "step": 1699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3746065934159123e-05,
+      "loss": 0.4859,
+      "step": 1700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3739592057378005e-05,
+      "loss": 0.4836,
+      "step": 1701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.373311635796515e-05,
+      "loss": 0.4855,
+      "step": 1702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3726638839076732e-05,
+      "loss": 0.4853,
+      "step": 1703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3720159503869816e-05,
+      "loss": 0.4841,
+      "step": 1704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.371367835550235e-05,
+      "loss": 0.4884,
+      "step": 1705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3707195397133165e-05,
+      "loss": 0.4891,
+      "step": 1706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3700710631921984e-05,
+      "loss": 0.5018,
+      "step": 1707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3694224063029396e-05,
+      "loss": 0.4771,
+      "step": 1708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3687735693616876e-05,
+      "loss": 0.4885,
+      "step": 1709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3681245526846782e-05,
+      "loss": 0.4839,
+      "step": 1710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3674753565882336e-05,
+      "loss": 0.4818,
+      "step": 1711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3668259813887644e-05,
+      "loss": 0.4851,
+      "step": 1712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3661764274027678e-05,
+      "loss": 0.4894,
+      "step": 1713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.365526694946829e-05,
+      "loss": 0.4878,
+      "step": 1714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3648767843376196e-05,
+      "loss": 0.4806,
+      "step": 1715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3642266958918985e-05,
+      "loss": 0.4869,
+      "step": 1716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.36357642992651e-05,
+      "loss": 0.4899,
+      "step": 1717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3629259867583864e-05,
+      "loss": 0.4641,
+      "step": 1718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3622753667045459e-05,
+      "loss": 0.4773,
+      "step": 1719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3616245700820922e-05,
+      "loss": 0.488,
+      "step": 1720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3609735972082168e-05,
+      "loss": 0.4801,
+      "step": 1721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3603224484001949e-05,
+      "loss": 0.4792,
+      "step": 1722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3596711239753889e-05,
+      "loss": 0.4845,
+      "step": 1723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3590196242512463e-05,
+      "loss": 0.4805,
+      "step": 1724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3583679495453e-05,
+      "loss": 0.4782,
+      "step": 1725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3577161001751696e-05,
+      "loss": 0.4805,
+      "step": 1726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3570640764585567e-05,
+      "loss": 0.4776,
+      "step": 1727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3564118787132507e-05,
+      "loss": 0.4874,
+      "step": 1728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.355759507257125e-05,
+      "loss": 0.4845,
+      "step": 1729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3551069624081372e-05,
+      "loss": 0.475,
+      "step": 1730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3544542444843298e-05,
+      "loss": 0.4838,
+      "step": 1731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3538013538038295e-05,
+      "loss": 0.4887,
+      "step": 1732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3531482906848474e-05,
+      "loss": 0.4802,
+      "step": 1733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3524950554456786e-05,
+      "loss": 0.48,
+      "step": 1734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3518416484047018e-05,
+      "loss": 0.4869,
+      "step": 1735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3511880698803801e-05,
+      "loss": 0.4832,
+      "step": 1736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.350534320191259e-05,
+      "loss": 0.4851,
+      "step": 1737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.349880399655969e-05,
+      "loss": 0.4908,
+      "step": 1738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3492263085932224e-05,
+      "loss": 0.4807,
+      "step": 1739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3485720473218153e-05,
+      "loss": 0.4741,
+      "step": 1740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3479176161606269e-05,
+      "loss": 0.4821,
+      "step": 1741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.347263015428619e-05,
+      "loss": 0.4855,
+      "step": 1742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3466082454448364e-05,
+      "loss": 0.4815,
+      "step": 1743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3459533065284049e-05,
+      "loss": 0.485,
+      "step": 1744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3452981989985347e-05,
+      "loss": 0.4832,
+      "step": 1745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.344642923174517e-05,
+      "loss": 0.4788,
+      "step": 1746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3439874793757255e-05,
+      "loss": 0.4825,
+      "step": 1747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3433318679216154e-05,
+      "loss": 0.4791,
+      "step": 1748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3426760891317236e-05,
+      "loss": 0.487,
+      "step": 1749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.4822,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3413640308231511e-05,
+      "loss": 0.481,
+      "step": 1751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.340707751943952e-05,
+      "loss": 0.4869,
+      "step": 1752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.340051307007933e-05,
+      "loss": 0.4805,
+      "step": 1753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3393946963350381e-05,
+      "loss": 0.4769,
+      "step": 1754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3387379202452917e-05,
+      "loss": 0.4812,
+      "step": 1755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3380809790587975e-05,
+      "loss": 0.4981,
+      "step": 1756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3374238730957414e-05,
+      "loss": 0.4927,
+      "step": 1757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3367666026763884e-05,
+      "loss": 0.4751,
+      "step": 1758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3361091681210846e-05,
+      "loss": 0.4902,
+      "step": 1759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3354515697502552e-05,
+      "loss": 0.4819,
+      "step": 1760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3347938078844058e-05,
+      "loss": 0.4828,
+      "step": 1761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3341358828441217e-05,
+      "loss": 0.4786,
+      "step": 1762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3334777949500673e-05,
+      "loss": 0.4736,
+      "step": 1763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3328195445229869e-05,
+      "loss": 0.4898,
+      "step": 1764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3321611318837033e-05,
+      "loss": 0.4811,
+      "step": 1765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3315025573531198e-05,
+      "loss": 0.4747,
+      "step": 1766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3308438212522164e-05,
+      "loss": 0.4829,
+      "step": 1767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3301849239020537e-05,
+      "loss": 0.4746,
+      "step": 1768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3295258656237703e-05,
+      "loss": 0.4806,
+      "step": 1769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3288666467385834e-05,
+      "loss": 0.4796,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.328207267567788e-05,
+      "loss": 0.4905,
+      "step": 1771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.327547728432757e-05,
+      "loss": 0.483,
+      "step": 1772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3268880296549424e-05,
+      "loss": 0.4775,
+      "step": 1773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3262281715558736e-05,
+      "loss": 0.4796,
+      "step": 1774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3255681544571568e-05,
+      "loss": 0.4838,
+      "step": 1775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3249079786804765e-05,
+      "loss": 0.4686,
+      "step": 1776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3242476445475945e-05,
+      "loss": 0.4845,
+      "step": 1777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3235871523803496e-05,
+      "loss": 0.4701,
+      "step": 1778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3229265025006577e-05,
+      "loss": 0.4818,
+      "step": 1779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3222656952305113e-05,
+      "loss": 0.4829,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.32160473089198e-05,
+      "loss": 0.4708,
+      "step": 1781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3209436098072095e-05,
+      "loss": 0.4868,
+      "step": 1782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3202823322984228e-05,
+      "loss": 0.4794,
+      "step": 1783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.319620898687918e-05,
+      "loss": 0.4845,
+      "step": 1784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3189593092980701e-05,
+      "loss": 0.4759,
+      "step": 1785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3182975644513296e-05,
+      "loss": 0.4768,
+      "step": 1786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3176356644702225e-05,
+      "loss": 0.4845,
+      "step": 1787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.316973609677352e-05,
+      "loss": 0.4712,
+      "step": 1788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.316311400395394e-05,
+      "loss": 0.4782,
+      "step": 1789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3156490369471026e-05,
+      "loss": 0.4897,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3149865196553049e-05,
+      "loss": 0.4865,
+      "step": 1791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3143238488429042e-05,
+      "loss": 0.4857,
+      "step": 1792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3136610248328779e-05,
+      "loss": 0.4719,
+      "step": 1793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3129980479482783e-05,
+      "loss": 0.485,
+      "step": 1794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3123349185122328e-05,
+      "loss": 0.4777,
+      "step": 1795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3116716368479418e-05,
+      "loss": 0.485,
+      "step": 1796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.311008203278682e-05,
+      "loss": 0.472,
+      "step": 1797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3103446181278015e-05,
+      "loss": 0.4786,
+      "step": 1798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3096808817187243e-05,
+      "loss": 0.4881,
+      "step": 1799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.4784,
+      "step": 1800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3083529564200417e-05,
+      "loss": 0.4701,
+      "step": 1801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3076887681776509e-05,
+      "loss": 0.4837,
+      "step": 1802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.307024429971492e-05,
+      "loss": 0.4714,
+      "step": 1803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.306359942125356e-05,
+      "loss": 0.4725,
+      "step": 1804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3056953049631059e-05,
+      "loss": 0.4781,
+      "step": 1805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3050305188086778e-05,
+      "loss": 0.4788,
+      "step": 1806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3043655839860803e-05,
+      "loss": 0.4746,
+      "step": 1807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3037005008193944e-05,
+      "loss": 0.4831,
+      "step": 1808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3030352696327741e-05,
+      "loss": 0.4815,
+      "step": 1809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3023698907504447e-05,
+      "loss": 0.4818,
+      "step": 1810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3017043644967036e-05,
+      "loss": 0.4841,
+      "step": 1811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3010386911959207e-05,
+      "loss": 0.4833,
+      "step": 1812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3003728711725364e-05,
+      "loss": 0.4823,
+      "step": 1813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.299706904751064e-05,
+      "loss": 0.4766,
+      "step": 1814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2990407922560869e-05,
+      "loss": 0.486,
+      "step": 1815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2983745340122604e-05,
+      "loss": 0.4841,
+      "step": 1816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2977081303443107e-05,
+      "loss": 0.4669,
+      "step": 1817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.297041581577035e-05,
+      "loss": 0.4775,
+      "step": 1818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2963748880353011e-05,
+      "loss": 0.484,
+      "step": 1819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2957080500440469e-05,
+      "loss": 0.4809,
+      "step": 1820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2950410679282815e-05,
+      "loss": 0.4872,
+      "step": 1821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2943739420130837e-05,
+      "loss": 0.4783,
+      "step": 1822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2937066726236029e-05,
+      "loss": 0.4814,
+      "step": 1823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2930392600850574e-05,
+      "loss": 0.4916,
+      "step": 1824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2923717047227368e-05,
+      "loss": 0.4782,
+      "step": 1825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.291704006861999e-05,
+      "loss": 0.4834,
+      "step": 1826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2910361668282718e-05,
+      "loss": 0.4777,
+      "step": 1827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2903681849470528e-05,
+      "loss": 0.4875,
+      "step": 1828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2897000615439075e-05,
+      "loss": 0.4744,
+      "step": 1829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2890317969444716e-05,
+      "loss": 0.4738,
+      "step": 1830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2883633914744493e-05,
+      "loss": 0.4836,
+      "step": 1831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.287694845459613e-05,
+      "loss": 0.4791,
+      "step": 1832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2870261592258038e-05,
+      "loss": 0.4882,
+      "step": 1833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2863573330989315e-05,
+      "loss": 0.4767,
+      "step": 1834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2856883674049736e-05,
+      "loss": 0.4822,
+      "step": 1835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2850192624699762e-05,
+      "loss": 0.4875,
+      "step": 1836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2843500186200529e-05,
+      "loss": 0.4773,
+      "step": 1837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2836806361813846e-05,
+      "loss": 0.4728,
+      "step": 1838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2830111154802203e-05,
+      "loss": 0.4741,
+      "step": 1839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2823414568428767e-05,
+      "loss": 0.482,
+      "step": 1840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2816716605957366e-05,
+      "loss": 0.4805,
+      "step": 1841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2810017270652513e-05,
+      "loss": 0.4887,
+      "step": 1842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2803316565779378e-05,
+      "loss": 0.4764,
+      "step": 1843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.27966144946038e-05,
+      "loss": 0.4806,
+      "step": 1844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2789911060392295e-05,
+      "loss": 0.4749,
+      "step": 1845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.278320626641203e-05,
+      "loss": 0.4743,
+      "step": 1846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2776500115930842e-05,
+      "loss": 0.478,
+      "step": 1847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2769792612217224e-05,
+      "loss": 0.4699,
+      "step": 1848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2763083758540337e-05,
+      "loss": 0.48,
+      "step": 1849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2756373558169992e-05,
+      "loss": 0.4735,
+      "step": 1850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2749662014376662e-05,
+      "loss": 0.4668,
+      "step": 1851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2742949130431468e-05,
+      "loss": 0.4773,
+      "step": 1852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2736234909606186e-05,
+      "loss": 0.4752,
+      "step": 1853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2729519355173254e-05,
+      "loss": 0.4708,
+      "step": 1854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2722802470405744e-05,
+      "loss": 0.4723,
+      "step": 1855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2716084258577388e-05,
+      "loss": 0.4803,
+      "step": 1856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.270936472296256e-05,
+      "loss": 0.4727,
+      "step": 1857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.270264386683628e-05,
+      "loss": 0.4859,
+      "step": 1858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2695921693474211e-05,
+      "loss": 0.4818,
+      "step": 1859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2689198206152657e-05,
+      "loss": 0.4803,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.268247340814857e-05,
+      "loss": 0.4806,
+      "step": 1861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2675747302739528e-05,
+      "loss": 0.4801,
+      "step": 1862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2669019893203758e-05,
+      "loss": 0.4766,
+      "step": 1863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2662291182820115e-05,
+      "loss": 0.4783,
+      "step": 1864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.265556117486809e-05,
+      "loss": 0.4782,
+      "step": 1865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2648829872627809e-05,
+      "loss": 0.4812,
+      "step": 1866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2642097279380025e-05,
+      "loss": 0.4618,
+      "step": 1867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.263536339840613e-05,
+      "loss": 0.4758,
+      "step": 1868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2628628232988123e-05,
+      "loss": 0.4825,
+      "step": 1869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2621891786408648e-05,
+      "loss": 0.4724,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.261515406195097e-05,
+      "loss": 0.4747,
+      "step": 1871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2608415062898971e-05,
+      "loss": 0.4753,
+      "step": 1872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2601674792537157e-05,
+      "loss": 0.4835,
+      "step": 1873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2594933254150654e-05,
+      "loss": 0.4819,
+      "step": 1874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.4721,
+      "step": 1875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2581446386447178e-05,
+      "loss": 0.4681,
+      "step": 1876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.257470106370354e-05,
+      "loss": 0.4716,
+      "step": 1877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.256795448608188e-05,
+      "loss": 0.4839,
+      "step": 1878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2561206656870397e-05,
+      "loss": 0.4876,
+      "step": 1879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2554457579357906e-05,
+      "loss": 0.4765,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2547707256833823e-05,
+      "loss": 0.4745,
+      "step": 1881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2540955692588173e-05,
+      "loss": 0.4808,
+      "step": 1882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2534202889911584e-05,
+      "loss": 0.4814,
+      "step": 1883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2527448852095295e-05,
+      "loss": 0.4743,
+      "step": 1884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.252069358243114e-05,
+      "loss": 0.4813,
+      "step": 1885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.251393708421155e-05,
+      "loss": 0.4794,
+      "step": 1886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2507179360729569e-05,
+      "loss": 0.4665,
+      "step": 1887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2500420415278822e-05,
+      "loss": 0.4852,
+      "step": 1888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.249366025115354e-05,
+      "loss": 0.4783,
+      "step": 1889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2486898871648552e-05,
+      "loss": 0.4801,
+      "step": 1890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2480136280059256e-05,
+      "loss": 0.4749,
+      "step": 1891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2473372479681671e-05,
+      "loss": 0.47,
+      "step": 1892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2466607473812386e-05,
+      "loss": 0.4692,
+      "step": 1893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2459841265748582e-05,
+      "loss": 0.475,
+      "step": 1894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2453073858788027e-05,
+      "loss": 0.4826,
+      "step": 1895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2446305256229074e-05,
+      "loss": 0.4779,
+      "step": 1896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2439535461370658e-05,
+      "loss": 0.4789,
+      "step": 1897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2432764477512294e-05,
+      "loss": 0.4867,
+      "step": 1898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2425992307954075e-05,
+      "loss": 0.4743,
+      "step": 1899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2419218955996677e-05,
+      "loss": 0.4785,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.241244442494135e-05,
+      "loss": 0.477,
+      "step": 1901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2405668718089918e-05,
+      "loss": 0.4712,
+      "step": 1902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2398891838744777e-05,
+      "loss": 0.4872,
+      "step": 1903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2392113790208895e-05,
+      "loss": 0.4697,
+      "step": 1904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.238533457578581e-05,
+      "loss": 0.4797,
+      "step": 1905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2378554198779632e-05,
+      "loss": 0.4758,
+      "step": 1906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2371772662495031e-05,
+      "loss": 0.4765,
+      "step": 1907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.236498997023725e-05,
+      "loss": 0.4841,
+      "step": 1908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2358206125312085e-05,
+      "loss": 0.4687,
+      "step": 1909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.23514211310259e-05,
+      "loss": 0.4781,
+      "step": 1910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2344634990685624e-05,
+      "loss": 0.4881,
+      "step": 1911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2337847707598738e-05,
+      "loss": 0.4639,
+      "step": 1912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.233105928507328e-05,
+      "loss": 0.4829,
+      "step": 1913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2324269726417841e-05,
+      "loss": 0.4658,
+      "step": 1914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2317479034941572e-05,
+      "loss": 0.4748,
+      "step": 1915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2310687213954182e-05,
+      "loss": 0.4763,
+      "step": 1916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2303894266765908e-05,
+      "loss": 0.4808,
+      "step": 1917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2297100196687557e-05,
+      "loss": 0.4816,
+      "step": 1918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2290305007030479e-05,
+      "loss": 0.4887,
+      "step": 1919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2283508701106559e-05,
+      "loss": 0.4741,
+      "step": 1920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2276711282228241e-05,
+      "loss": 0.4783,
+      "step": 1921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2269912753708502e-05,
+      "loss": 0.4791,
+      "step": 1922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.226311311886086e-05,
+      "loss": 0.4749,
+      "step": 1923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2256312380999376e-05,
+      "loss": 0.4709,
+      "step": 1924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2249510543438652e-05,
+      "loss": 0.4712,
+      "step": 1925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2242707609493814e-05,
+      "loss": 0.4765,
+      "step": 1926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.223590358248053e-05,
+      "loss": 0.4806,
+      "step": 1927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2229098465715005e-05,
+      "loss": 0.4773,
+      "step": 1928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2222292262513967e-05,
+      "loss": 0.4774,
+      "step": 1929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2215484976194675e-05,
+      "loss": 0.4785,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.220867661007492e-05,
+      "loss": 0.4762,
+      "step": 1931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2201867167473015e-05,
+      "loss": 0.4713,
+      "step": 1932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2195056651707806e-05,
+      "loss": 0.4761,
+      "step": 1933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2188245066098647e-05,
+      "loss": 0.4807,
+      "step": 1934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2181432413965428e-05,
+      "loss": 0.4789,
+      "step": 1935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.217461869862855e-05,
+      "loss": 0.4752,
+      "step": 1936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2167803923408935e-05,
+      "loss": 0.4709,
+      "step": 1937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2160988091628023e-05,
+      "loss": 0.4668,
+      "step": 1938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2154171206607765e-05,
+      "loss": 0.4733,
+      "step": 1939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2147353271670634e-05,
+      "loss": 0.4752,
+      "step": 1940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2140534290139601e-05,
+      "loss": 0.4733,
+      "step": 1941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2133714265338162e-05,
+      "loss": 0.476,
+      "step": 1942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2126893200590309e-05,
+      "loss": 0.4827,
+      "step": 1943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.212007109922055e-05,
+      "loss": 0.4862,
+      "step": 1944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.211324796455389e-05,
+      "loss": 0.4674,
+      "step": 1945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2106423799915841e-05,
+      "loss": 0.4838,
+      "step": 1946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2099598608632427e-05,
+      "loss": 0.4739,
+      "step": 1947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2092772394030153e-05,
+      "loss": 0.4774,
+      "step": 1948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.208594515943604e-05,
+      "loss": 0.4804,
+      "step": 1949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 0.4695,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2072287643582825e-05,
+      "loss": 0.4743,
+      "step": 1951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2065457368980236e-05,
+      "loss": 0.4793,
+      "step": 1952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2058626087698814e-05,
+      "loss": 0.483,
+      "step": 1953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2051793803068046e-05,
+      "loss": 0.4766,
+      "step": 1954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2044960518417902e-05,
+      "loss": 0.4768,
+      "step": 1955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.203812623707885e-05,
+      "loss": 0.4811,
+      "step": 1956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2031290962381823e-05,
+      "loss": 0.474,
+      "step": 1957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.202445469765826e-05,
+      "loss": 0.471,
+      "step": 1958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.201761744624007e-05,
+      "loss": 0.4756,
+      "step": 1959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2010779211459649e-05,
+      "loss": 0.4736,
+      "step": 1960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2003939996649864e-05,
+      "loss": 0.4737,
+      "step": 1961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1997099805144071e-05,
+      "loss": 0.4707,
+      "step": 1962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1990258640276094e-05,
+      "loss": 0.4734,
+      "step": 1963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1983416505380234e-05,
+      "loss": 0.4808,
+      "step": 1964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1976573403791263e-05,
+      "loss": 0.4804,
+      "step": 1965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1969729338844429e-05,
+      "loss": 0.4694,
+      "step": 1966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.196288431387544e-05,
+      "loss": 0.4693,
+      "step": 1967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1956038332220484e-05,
+      "loss": 0.4748,
+      "step": 1968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1949191397216207e-05,
+      "loss": 0.4689,
+      "step": 1969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.194234351219972e-05,
+      "loss": 0.4721,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1935494680508606e-05,
+      "loss": 0.4793,
+      "step": 1971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.192864490548089e-05,
+      "loss": 0.4707,
+      "step": 1972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1921794190455082e-05,
+      "loss": 0.4747,
+      "step": 1973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.191494253877013e-05,
+      "loss": 0.474,
+      "step": 1974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.190808995376545e-05,
+      "loss": 0.4692,
+      "step": 1975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1901236438780902e-05,
+      "loss": 0.4738,
+      "step": 1976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1894381997156814e-05,
+      "loss": 0.4714,
+      "step": 1977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1887526632233954e-05,
+      "loss": 0.478,
+      "step": 1978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.188067034735354e-05,
+      "loss": 0.4695,
+      "step": 1979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 0.4791,
+      "step": 1980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.186695503108719e-05,
+      "loss": 0.483,
+      "step": 1981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.186009600638593e-05,
+      "loss": 0.4759,
+      "step": 1982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1853236075096474e-05,
+      "loss": 0.4682,
+      "step": 1983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.184637524056227e-05,
+      "loss": 0.4815,
+      "step": 1984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1839513506127202e-05,
+      "loss": 0.4674,
+      "step": 1985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1832650875135599e-05,
+      "loss": 0.4645,
+      "step": 1986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1825787350932224e-05,
+      "loss": 0.4708,
+      "step": 1987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.181892293686227e-05,
+      "loss": 0.4741,
+      "step": 1988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1812057636271374e-05,
+      "loss": 0.4607,
+      "step": 1989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1805191452505602e-05,
+      "loss": 0.4731,
+      "step": 1990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1798324388911445e-05,
+      "loss": 0.4795,
+      "step": 1991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1791456448835825e-05,
+      "loss": 0.4643,
+      "step": 1992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1784587635626095e-05,
+      "loss": 0.4771,
+      "step": 1993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1777717952630033e-05,
+      "loss": 0.4841,
+      "step": 1994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1770847403195836e-05,
+      "loss": 0.466,
+      "step": 1995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1763975990672125e-05,
+      "loss": 0.4709,
+      "step": 1996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1757103718407948e-05,
+      "loss": 0.4772,
+      "step": 1997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1750230589752763e-05,
+      "loss": 0.4804,
+      "step": 1998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1743356608056448e-05,
+      "loss": 0.4762,
+      "step": 1999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.476,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1729606098942039e-05,
+      "loss": 0.4759,
+      "step": 2001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1722729578225769e-05,
+      "loss": 0.4765,
+      "step": 2002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.171585221787203e-05,
+      "loss": 0.4699,
+      "step": 2003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1708974021232768e-05,
+      "loss": 0.4659,
+      "step": 2004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1702094991660326e-05,
+      "loss": 0.4842,
+      "step": 2005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1695215132507465e-05,
+      "loss": 0.473,
+      "step": 2006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1688334447127338e-05,
+      "loss": 0.4703,
+      "step": 2007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1681452938873516e-05,
+      "loss": 0.4783,
+      "step": 2008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1674570611099956e-05,
+      "loss": 0.4772,
+      "step": 2009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1667687467161025e-05,
+      "loss": 0.4731,
+      "step": 2010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.166080351041148e-05,
+      "loss": 0.4771,
+      "step": 2011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1653918744206478e-05,
+      "loss": 0.4724,
+      "step": 2012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1647033171901573e-05,
+      "loss": 0.4868,
+      "step": 2013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1640146796852711e-05,
+      "loss": 0.4699,
+      "step": 2014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1633259622416224e-05,
+      "loss": 0.4711,
+      "step": 2015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1626371651948839e-05,
+      "loss": 0.4665,
+      "step": 2016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1619482888807662e-05,
+      "loss": 0.4733,
+      "step": 2017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1612593336350209e-05,
+      "loss": 0.4679,
+      "step": 2018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1605702997934345e-05,
+      "loss": 0.4603,
+      "step": 2019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.159881187691835e-05,
+      "loss": 0.4778,
+      "step": 2020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1591919976660867e-05,
+      "loss": 0.4777,
+      "step": 2021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.158502730052093e-05,
+      "loss": 0.4676,
+      "step": 2022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.157813385185794e-05,
+      "loss": 0.4659,
+      "step": 2023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.157123963403168e-05,
+      "loss": 0.4706,
+      "step": 2024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 0.4736,
+      "step": 2025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1557448904330362e-05,
+      "loss": 0.4755,
+      "step": 2026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.155055239917674e-05,
+      "loss": 0.4708,
+      "step": 2027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1543655138302714e-05,
+      "loss": 0.4771,
+      "step": 2028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1536757125069924e-05,
+      "loss": 0.4717,
+      "step": 2029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1529858362840383e-05,
+      "loss": 0.4812,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1522958854976458e-05,
+      "loss": 0.4784,
+      "step": 2031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1516058604840891e-05,
+      "loss": 0.4694,
+      "step": 2032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1509157615796775e-05,
+      "loss": 0.475,
+      "step": 2033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1502255891207572e-05,
+      "loss": 0.4731,
+      "step": 2034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1495353434437098e-05,
+      "loss": 0.4662,
+      "step": 2035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1488450248849523e-05,
+      "loss": 0.4591,
+      "step": 2036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1481546337809381e-05,
+      "loss": 0.4716,
+      "step": 2037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1474641704681551e-05,
+      "loss": 0.4778,
+      "step": 2038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1467736352831266e-05,
+      "loss": 0.4727,
+      "step": 2039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1460830285624119e-05,
+      "loss": 0.4702,
+      "step": 2040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1453923506426032e-05,
+      "loss": 0.4599,
+      "step": 2041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1447016018603293e-05,
+      "loss": 0.476,
+      "step": 2042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1440107825522522e-05,
+      "loss": 0.4718,
+      "step": 2043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1433198930550694e-05,
+      "loss": 0.4636,
+      "step": 2044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1426289337055119e-05,
+      "loss": 0.4703,
+      "step": 2045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1419379048403446e-05,
+      "loss": 0.4712,
+      "step": 2046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.141246806796367e-05,
+      "loss": 0.4761,
+      "step": 2047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.140555639910411e-05,
+      "loss": 0.4703,
+      "step": 2048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1398644045193443e-05,
+      "loss": 0.4645,
+      "step": 2049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1391731009600655e-05,
+      "loss": 0.468,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1384817295695083e-05,
+      "loss": 0.4657,
+      "step": 2051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.137790290684638e-05,
+      "loss": 0.4746,
+      "step": 2052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1370987846424547e-05,
+      "loss": 0.4678,
+      "step": 2053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1364072117799884e-05,
+      "loss": 0.4577,
+      "step": 2054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1357155724343046e-05,
+      "loss": 0.4775,
+      "step": 2055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1350238669424993e-05,
+      "loss": 0.4801,
+      "step": 2056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1343320956417015e-05,
+      "loss": 0.4634,
+      "step": 2057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1336402588690727e-05,
+      "loss": 0.4721,
+      "step": 2058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1329483569618045e-05,
+      "loss": 0.4628,
+      "step": 2059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1322563902571227e-05,
+      "loss": 0.4625,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1315643590922827e-05,
+      "loss": 0.4745,
+      "step": 2061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1308722638045724e-05,
+      "loss": 0.4741,
+      "step": 2062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1301801047313106e-05,
+      "loss": 0.4674,
+      "step": 2063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.129487882209847e-05,
+      "loss": 0.4642,
+      "step": 2064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.128795596577563e-05,
+      "loss": 0.468,
+      "step": 2065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1281032481718696e-05,
+      "loss": 0.4693,
+      "step": 2066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1274108373302095e-05,
+      "loss": 0.4711,
+      "step": 2067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1267183643900548e-05,
+      "loss": 0.4669,
+      "step": 2068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1260258296889086e-05,
+      "loss": 0.4638,
+      "step": 2069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1253332335643043e-05,
+      "loss": 0.4742,
+      "step": 2070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1246405763538047e-05,
+      "loss": 0.4747,
+      "step": 2071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1239478583950019e-05,
+      "loss": 0.4727,
+      "step": 2072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1232550800255188e-05,
+      "loss": 0.4698,
+      "step": 2073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1225622415830068e-05,
+      "loss": 0.4603,
+      "step": 2074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1218693434051475e-05,
+      "loss": 0.4627,
+      "step": 2075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1211763858296507e-05,
+      "loss": 0.4669,
+      "step": 2076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1204833691942553e-05,
+      "loss": 0.475,
+      "step": 2077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1197902938367297e-05,
+      "loss": 0.4691,
+      "step": 2078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.11909716009487e-05,
+      "loss": 0.4592,
+      "step": 2079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1184039683065014e-05,
+      "loss": 0.4664,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1177107188094765e-05,
+      "loss": 0.4738,
+      "step": 2081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1170174119416778e-05,
+      "loss": 0.4612,
+      "step": 2082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1163240480410136e-05,
+      "loss": 0.4678,
+      "step": 2083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.1156306274454218e-05,
+      "loss": 0.4634,
+      "step": 2084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1149371504928667e-05,
+      "loss": 0.4721,
+      "step": 2085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1142436175213409e-05,
+      "loss": 0.4852,
+      "step": 2086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1135500288688636e-05,
+      "loss": 0.4566,
+      "step": 2087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1128563848734817e-05,
+      "loss": 0.4722,
+      "step": 2088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.112162685873269e-05,
+      "loss": 0.4713,
+      "step": 2089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1114689322063255e-05,
+      "loss": 0.4608,
+      "step": 2090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1107751242107786e-05,
+      "loss": 0.4681,
+      "step": 2091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1100812622247823e-05,
+      "loss": 0.466,
+      "step": 2092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1093873465865156e-05,
+      "loss": 0.4667,
+      "step": 2093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1086933776341853e-05,
+      "loss": 0.4681,
+      "step": 2094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1079993557060228e-05,
+      "loss": 0.4779,
+      "step": 2095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1073052811402867e-05,
+      "loss": 0.4748,
+      "step": 2096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.10661115427526e-05,
+      "loss": 0.4768,
+      "step": 2097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.105916975449252e-05,
+      "loss": 0.4772,
+      "step": 2098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1052227450005968e-05,
+      "loss": 0.4694,
+      "step": 2099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.467,
+      "step": 2100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1038341305888074e-05,
+      "loss": 0.4666,
+      "step": 2101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1031397473024674e-05,
+      "loss": 0.468,
+      "step": 2102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1024453137470677e-05,
+      "loss": 0.4688,
+      "step": 2103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1017508302610665e-05,
+      "loss": 0.4689,
+      "step": 2104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1010562971829464e-05,
+      "loss": 0.4643,
+      "step": 2105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.1003617148512149e-05,
+      "loss": 0.4625,
+      "step": 2106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.099667083604403e-05,
+      "loss": 0.4604,
+      "step": 2107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0989724037810651e-05,
+      "loss": 0.4704,
+      "step": 2108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0982776757197799e-05,
+      "loss": 0.4729,
+      "step": 2109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0975828997591496e-05,
+      "loss": 0.4744,
+      "step": 2110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0968880762377994e-05,
+      "loss": 0.4613,
+      "step": 2111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0961932054943778e-05,
+      "loss": 0.4665,
+      "step": 2112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0954982878675564e-05,
+      "loss": 0.4616,
+      "step": 2113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0948033236960294e-05,
+      "loss": 0.4712,
+      "step": 2114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0941083133185146e-05,
+      "loss": 0.4633,
+      "step": 2115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0934132570737508e-05,
+      "loss": 0.4687,
+      "step": 2116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0927181553005001e-05,
+      "loss": 0.4598,
+      "step": 2117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0920230083375474e-05,
+      "loss": 0.4614,
+      "step": 2118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0913278165236977e-05,
+      "loss": 0.4793,
+      "step": 2119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0906325801977804e-05,
+      "loss": 0.4781,
+      "step": 2120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0899372996986439e-05,
+      "loss": 0.4601,
+      "step": 2121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0892419753651606e-05,
+      "loss": 0.4638,
+      "step": 2122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0885466075362224e-05,
+      "loss": 0.4677,
+      "step": 2123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0878511965507435e-05,
+      "loss": 0.4656,
+      "step": 2124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0871557427476585e-05,
+      "loss": 0.466,
+      "step": 2125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.086460246465923e-05,
+      "loss": 0.473,
+      "step": 2126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.085764708044514e-05,
+      "loss": 0.4692,
+      "step": 2127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0850691278224282e-05,
+      "loss": 0.4639,
+      "step": 2128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0843735061386829e-05,
+      "loss": 0.4723,
+      "step": 2129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.083677843332316e-05,
+      "loss": 0.4694,
+      "step": 2130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.082982139742384e-05,
+      "loss": 0.4671,
+      "step": 2131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0822863957079657e-05,
+      "loss": 0.4665,
+      "step": 2132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0815906115681579e-05,
+      "loss": 0.4688,
+      "step": 2133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0808947876620768e-05,
+      "loss": 0.4603,
+      "step": 2134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0801989243288588e-05,
+      "loss": 0.4733,
+      "step": 2135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.07950302190766e-05,
+      "loss": 0.4655,
+      "step": 2136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0788070807376536e-05,
+      "loss": 0.4606,
+      "step": 2137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0781111011580336e-05,
+      "loss": 0.4766,
+      "step": 2138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0774150835080119e-05,
+      "loss": 0.4805,
+      "step": 2139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0767190281268187e-05,
+      "loss": 0.4589,
+      "step": 2140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0760229353537032e-05,
+      "loss": 0.4715,
+      "step": 2141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0753268055279328e-05,
+      "loss": 0.4639,
+      "step": 2142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0746306389887924e-05,
+      "loss": 0.4645,
+      "step": 2143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0739344360755853e-05,
+      "loss": 0.4694,
+      "step": 2144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0732381971276318e-05,
+      "loss": 0.4702,
+      "step": 2145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.072541922484271e-05,
+      "loss": 0.4647,
+      "step": 2146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0718456124848584e-05,
+      "loss": 0.4603,
+      "step": 2147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.071149267468767e-05,
+      "loss": 0.4707,
+      "step": 2148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.070452887775387e-05,
+      "loss": 0.47,
+      "step": 2149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0697564737441254e-05,
+      "loss": 0.4658,
+      "step": 2150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0690600257144062e-05,
+      "loss": 0.4697,
+      "step": 2151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0683635440256689e-05,
+      "loss": 0.4657,
+      "step": 2152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.067667029017371e-05,
+      "loss": 0.4656,
+      "step": 2153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0669704810289852e-05,
+      "loss": 0.466,
+      "step": 2154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0662739004000005e-05,
+      "loss": 0.4711,
+      "step": 2155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0655772874699217e-05,
+      "loss": 0.4656,
+      "step": 2156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0648806425782697e-05,
+      "loss": 0.4666,
+      "step": 2157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0641839660645806e-05,
+      "loss": 0.469,
+      "step": 2158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0634872582684062e-05,
+      "loss": 0.4718,
+      "step": 2159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 0.4707,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0620937501868842e-05,
+      "loss": 0.4547,
+      "step": 2161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0613969505807157e-05,
+      "loss": 0.4679,
+      "step": 2162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.060700121050419e-05,
+      "loss": 0.4602,
+      "step": 2163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0600032619356208e-05,
+      "loss": 0.4744,
+      "step": 2164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0593063735759619e-05,
+      "loss": 0.4639,
+      "step": 2165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0586094563110965e-05,
+      "loss": 0.4596,
+      "step": 2166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0579125104806944e-05,
+      "loss": 0.4618,
+      "step": 2167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0572155364244383e-05,
+      "loss": 0.4587,
+      "step": 2168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0565185344820248e-05,
+      "loss": 0.4673,
+      "step": 2169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.055821504993164e-05,
+      "loss": 0.4606,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0551244482975798e-05,
+      "loss": 0.4779,
+      "step": 2171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0544273647350091e-05,
+      "loss": 0.4623,
+      "step": 2172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0537302546452022e-05,
+      "loss": 0.4689,
+      "step": 2173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.053033118367922e-05,
+      "loss": 0.4624,
+      "step": 2174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0523359562429441e-05,
+      "loss": 0.4642,
+      "step": 2175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0516387686100566e-05,
+      "loss": 0.463,
+      "step": 2176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.050941555809061e-05,
+      "loss": 0.4604,
+      "step": 2177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0502443181797696e-05,
+      "loss": 0.4688,
+      "step": 2178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0495470560620082e-05,
+      "loss": 0.4647,
+      "step": 2179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.0488497697956134e-05,
+      "loss": 0.4763,
+      "step": 2180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0481524597204342e-05,
+      "loss": 0.4635,
+      "step": 2181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0474551261763315e-05,
+      "loss": 0.4613,
+      "step": 2182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0467577695031763e-05,
+      "loss": 0.4682,
+      "step": 2183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0460603900408523e-05,
+      "loss": 0.4678,
+      "step": 2184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0453629881292537e-05,
+      "loss": 0.4691,
+      "step": 2185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0446655641082864e-05,
+      "loss": 0.4783,
+      "step": 2186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.043968118317865e-05,
+      "loss": 0.4673,
+      "step": 2187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0432706510979172e-05,
+      "loss": 0.4585,
+      "step": 2188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0425731627883798e-05,
+      "loss": 0.4575,
+      "step": 2189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0418756537291996e-05,
+      "loss": 0.4738,
+      "step": 2190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0411781242603352e-05,
+      "loss": 0.4704,
+      "step": 2191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0404805747217525e-05,
+      "loss": 0.4691,
+      "step": 2192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.03978300545343e-05,
+      "loss": 0.4628,
+      "step": 2193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0390854167953537e-05,
+      "loss": 0.4609,
+      "step": 2194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.03838780908752e-05,
+      "loss": 0.4651,
+      "step": 2195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0376901826699349e-05,
+      "loss": 0.4718,
+      "step": 2196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.036992537882612e-05,
+      "loss": 0.4669,
+      "step": 2197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.036294875065576e-05,
+      "loss": 0.4729,
+      "step": 2198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0355971945588586e-05,
+      "loss": 0.4644,
+      "step": 2199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0348994967025012e-05,
+      "loss": 0.4638,
+      "step": 2200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.034201781836553e-05,
+      "loss": 0.4526,
+      "step": 2201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0335040503010715e-05,
+      "loss": 0.4649,
+      "step": 2202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0328063024361232e-05,
+      "loss": 0.4608,
+      "step": 2203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0321085385817818e-05,
+      "loss": 0.4653,
+      "step": 2204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0314107590781284e-05,
+      "loss": 0.462,
+      "step": 2205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.030712964265253e-05,
+      "loss": 0.4701,
+      "step": 2206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0300151544832513e-05,
+      "loss": 0.4638,
+      "step": 2207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0293173300722286e-05,
+      "loss": 0.4727,
+      "step": 2208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0286194913722948e-05,
+      "loss": 0.4809,
+      "step": 2209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0279216387235691e-05,
+      "loss": 0.4665,
+      "step": 2210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0272237724661753e-05,
+      "loss": 0.467,
+      "step": 2211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.026525892940246e-05,
+      "loss": 0.468,
+      "step": 2212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0258280004859189e-05,
+      "loss": 0.4491,
+      "step": 2213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0251300954433377e-05,
+      "loss": 0.4583,
+      "step": 2214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0244321781526533e-05,
+      "loss": 0.4674,
+      "step": 2215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0237342489540221e-05,
+      "loss": 0.47,
+      "step": 2216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0230363081876065e-05,
+      "loss": 0.4648,
+      "step": 2217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0223383561935738e-05,
+      "loss": 0.4691,
+      "step": 2218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0216403933120979e-05,
+      "loss": 0.4533,
+      "step": 2219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 0.4608,
+      "step": 2220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0202444362475352e-05,
+      "loss": 0.4704,
+      "step": 2221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0195464427448213e-05,
+      "loss": 0.4649,
+      "step": 2222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0188484397154083e-05,
+      "loss": 0.4686,
+      "step": 2223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0181504274994949e-05,
+      "loss": 0.4585,
+      "step": 2224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0174524064372837e-05,
+      "loss": 0.4613,
+      "step": 2225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0167543768689816e-05,
+      "loss": 0.459,
+      "step": 2226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0160563391347998e-05,
+      "loss": 0.4555,
+      "step": 2227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0153582935749531e-05,
+      "loss": 0.4635,
+      "step": 2228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0146602405296608e-05,
+      "loss": 0.4645,
+      "step": 2229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0139621803391454e-05,
+      "loss": 0.4631,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.013264113343633e-05,
+      "loss": 0.4601,
+      "step": 2231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0125660398833528e-05,
+      "loss": 0.4719,
+      "step": 2232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0118679602985373e-05,
+      "loss": 0.4667,
+      "step": 2233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0111698749294223e-05,
+      "loss": 0.4663,
+      "step": 2234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.010471784116246e-05,
+      "loss": 0.4684,
+      "step": 2235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0097736881992492e-05,
+      "loss": 0.4735,
+      "step": 2236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0090755875186752e-05,
+      "loss": 0.4611,
+      "step": 2237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0083774824147707e-05,
+      "loss": 0.4632,
+      "step": 2238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.007679373227783e-05,
+      "loss": 0.4583,
+      "step": 2239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0069812602979617e-05,
+      "loss": 0.4565,
+      "step": 2240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0062831439655591e-05,
+      "loss": 0.4704,
+      "step": 2241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0055850245708283e-05,
+      "loss": 0.4664,
+      "step": 2242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0048869024540247e-05,
+      "loss": 0.4566,
+      "step": 2243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0041887779554041e-05,
+      "loss": 0.4623,
+      "step": 2244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0034906514152239e-05,
+      "loss": 0.4685,
+      "step": 2245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0027925231737428e-05,
+      "loss": 0.4665,
+      "step": 2246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0020943935712193e-05,
+      "loss": 0.4668,
+      "step": 2247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0013962629479145e-05,
+      "loss": 0.4697,
+      "step": 2248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0006981316440876e-05,
+      "loss": 0.4573,
+      "step": 2249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.4586,
+      "step": 2250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.993018683559126e-06,
+      "loss": 0.4767,
+      "step": 2251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.986037370520856e-06,
+      "loss": 0.4658,
+      "step": 2252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.979056064287807e-06,
+      "loss": 0.4629,
+      "step": 2253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.972074768262576e-06,
+      "loss": 0.4597,
+      "step": 2254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.965093485847766e-06,
+      "loss": 0.4561,
+      "step": 2255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.958112220445964e-06,
+      "loss": 0.466,
+      "step": 2256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.951130975459758e-06,
+      "loss": 0.4542,
+      "step": 2257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.944149754291719e-06,
+      "loss": 0.4564,
+      "step": 2258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.937168560344412e-06,
+      "loss": 0.4664,
+      "step": 2259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.930187397020385e-06,
+      "loss": 0.4703,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.923206267722173e-06,
+      "loss": 0.4678,
+      "step": 2261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.916225175852295e-06,
+      "loss": 0.4586,
+      "step": 2262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.909244124813246e-06,
+      "loss": 0.4636,
+      "step": 2263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.902263118007513e-06,
+      "loss": 0.468,
+      "step": 2264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.895282158837545e-06,
+      "loss": 0.4523,
+      "step": 2265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.88830125070578e-06,
+      "loss": 0.4557,
+      "step": 2266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.88132039701463e-06,
+      "loss": 0.4574,
+      "step": 2267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.874339601166474e-06,
+      "loss": 0.4695,
+      "step": 2268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.867358866563674e-06,
+      "loss": 0.4457,
+      "step": 2269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.860378196608549e-06,
+      "loss": 0.462,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.853397594703394e-06,
+      "loss": 0.4615,
+      "step": 2271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.84641706425047e-06,
+      "loss": 0.4658,
+      "step": 2272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.839436608652007e-06,
+      "loss": 0.4583,
+      "step": 2273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.832456231310189e-06,
+      "loss": 0.4689,
+      "step": 2274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.825475935627165e-06,
+      "loss": 0.466,
+      "step": 2275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.818495725005053e-06,
+      "loss": 0.467,
+      "step": 2276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.81151560284592e-06,
+      "loss": 0.4616,
+      "step": 2277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.80453557255179e-06,
+      "loss": 0.4571,
+      "step": 2278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.79755563752465e-06,
+      "loss": 0.4628,
+      "step": 2279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.790575801166432e-06,
+      "loss": 0.4701,
+      "step": 2280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.783596066879023e-06,
+      "loss": 0.4652,
+      "step": 2281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.776616438064265e-06,
+      "loss": 0.4608,
+      "step": 2282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.76963691812394e-06,
+      "loss": 0.4609,
+      "step": 2283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.762657510459784e-06,
+      "loss": 0.4707,
+      "step": 2284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.75567821847347e-06,
+      "loss": 0.4578,
+      "step": 2285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.748699045566626e-06,
+      "loss": 0.4635,
+      "step": 2286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.741719995140814e-06,
+      "loss": 0.4588,
+      "step": 2287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.73474107059754e-06,
+      "loss": 0.4632,
+      "step": 2288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.727762275338246e-06,
+      "loss": 0.4595,
+      "step": 2289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.720783612764314e-06,
+      "loss": 0.4612,
+      "step": 2290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.713805086277055e-06,
+      "loss": 0.4614,
+      "step": 2291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.706826699277719e-06,
+      "loss": 0.4593,
+      "step": 2292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.699848455167489e-06,
+      "loss": 0.4544,
+      "step": 2293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.692870357347474e-06,
+      "loss": 0.4618,
+      "step": 2294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.685892409218718e-06,
+      "loss": 0.4534,
+      "step": 2295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.678914614182185e-06,
+      "loss": 0.4558,
+      "step": 2296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.671936975638768e-06,
+      "loss": 0.4636,
+      "step": 2297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.664959496989286e-06,
+      "loss": 0.4524,
+      "step": 2298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.657982181634476e-06,
+      "loss": 0.4571,
+      "step": 2299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.651005032974994e-06,
+      "loss": 0.4679,
+      "step": 2300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.644028054411416e-06,
+      "loss": 0.4593,
+      "step": 2301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.637051249344244e-06,
+      "loss": 0.4691,
+      "step": 2302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.630074621173882e-06,
+      "loss": 0.4654,
+      "step": 2303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.623098173300655e-06,
+      "loss": 0.459,
+      "step": 2304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.616121909124801e-06,
+      "loss": 0.458,
+      "step": 2305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.609145832046465e-06,
+      "loss": 0.4555,
+      "step": 2306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.602169945465702e-06,
+      "loss": 0.4602,
+      "step": 2307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.595194252782476e-06,
+      "loss": 0.4547,
+      "step": 2308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.588218757396655e-06,
+      "loss": 0.4644,
+      "step": 2309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.581243462708007e-06,
+      "loss": 0.4589,
+      "step": 2310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.574268372116205e-06,
+      "loss": 0.4665,
+      "step": 2311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.567293489020831e-06,
+      "loss": 0.4643,
+      "step": 2312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.560318816821354e-06,
+      "loss": 0.459,
+      "step": 2313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.553344358917141e-06,
+      "loss": 0.4535,
+      "step": 2314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.546370118707463e-06,
+      "loss": 0.4566,
+      "step": 2315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.539396099591477e-06,
+      "loss": 0.4668,
+      "step": 2316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.532422304968243e-06,
+      "loss": 0.4591,
+      "step": 2317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.525448738236691e-06,
+      "loss": 0.4769,
+      "step": 2318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.518475402795661e-06,
+      "loss": 0.4583,
+      "step": 2319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.511502302043867e-06,
+      "loss": 0.4623,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.504529439379921e-06,
+      "loss": 0.4532,
+      "step": 2321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.497556818202306e-06,
+      "loss": 0.4604,
+      "step": 2322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.490584441909392e-06,
+      "loss": 0.4578,
+      "step": 2323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.483612313899436e-06,
+      "loss": 0.4473,
+      "step": 2324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.476640437570562e-06,
+      "loss": 0.4643,
+      "step": 2325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.469668816320785e-06,
+      "loss": 0.4636,
+      "step": 2326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.46269745354798e-06,
+      "loss": 0.4575,
+      "step": 2327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.45572635264991e-06,
+      "loss": 0.457,
+      "step": 2328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.448755517024207e-06,
+      "loss": 0.4656,
+      "step": 2329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.441784950068362e-06,
+      "loss": 0.4623,
+      "step": 2330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.434814655179756e-06,
+      "loss": 0.4522,
+      "step": 2331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.42784463575562e-06,
+      "loss": 0.4615,
+      "step": 2332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.420874895193056e-06,
+      "loss": 0.4659,
+      "step": 2333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.413905436889035e-06,
+      "loss": 0.4617,
+      "step": 2334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.406936264240386e-06,
+      "loss": 0.4669,
+      "step": 2335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.399967380643795e-06,
+      "loss": 0.4622,
+      "step": 2336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.392998789495813e-06,
+      "loss": 0.4648,
+      "step": 2337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.386030494192847e-06,
+      "loss": 0.4544,
+      "step": 2338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.379062498131161e-06,
+      "loss": 0.4688,
+      "step": 2339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.4526,
+      "step": 2340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.36512741731594e-06,
+      "loss": 0.4716,
+      "step": 2341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.358160339354194e-06,
+      "loss": 0.4616,
+      "step": 2342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.351193574217305e-06,
+      "loss": 0.4524,
+      "step": 2343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.344227125300788e-06,
+      "loss": 0.456,
+      "step": 2344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.337260996000002e-06,
+      "loss": 0.4663,
+      "step": 2345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.330295189710153e-06,
+      "loss": 0.4567,
+      "step": 2346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.323329709826294e-06,
+      "loss": 0.458,
+      "step": 2347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.316364559743315e-06,
+      "loss": 0.4561,
+      "step": 2348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.309399742855943e-06,
+      "loss": 0.4655,
+      "step": 2349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.302435262558748e-06,
+      "loss": 0.4586,
+      "step": 2350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.295471122246131e-06,
+      "loss": 0.4583,
+      "step": 2351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.288507325312334e-06,
+      "loss": 0.4557,
+      "step": 2352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.281543875151419e-06,
+      "loss": 0.4569,
+      "step": 2353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.274580775157294e-06,
+      "loss": 0.4673,
+      "step": 2354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.267618028723687e-06,
+      "loss": 0.4583,
+      "step": 2355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.260655639244152e-06,
+      "loss": 0.462,
+      "step": 2356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.253693610112079e-06,
+      "loss": 0.4657,
+      "step": 2357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.246731944720675e-06,
+      "loss": 0.4614,
+      "step": 2358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.239770646462968e-06,
+      "loss": 0.4608,
+      "step": 2359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.232809718731815e-06,
+      "loss": 0.4555,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.225849164919886e-06,
+      "loss": 0.4535,
+      "step": 2361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.218888988419668e-06,
+      "loss": 0.4591,
+      "step": 2362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.211929192623466e-06,
+      "loss": 0.4566,
+      "step": 2363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.204969780923404e-06,
+      "loss": 0.4611,
+      "step": 2364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.198010756711413e-06,
+      "loss": 0.4613,
+      "step": 2365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.191052123379234e-06,
+      "loss": 0.4555,
+      "step": 2366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.184093884318426e-06,
+      "loss": 0.4655,
+      "step": 2367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.177136042920344e-06,
+      "loss": 0.4561,
+      "step": 2368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.170178602576161e-06,
+      "loss": 0.4702,
+      "step": 2369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.163221566676847e-06,
+      "loss": 0.4658,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.156264938613176e-06,
+      "loss": 0.4634,
+      "step": 2371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.14930872177572e-06,
+      "loss": 0.4644,
+      "step": 2372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.142352919554862e-06,
+      "loss": 0.4612,
+      "step": 2373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.135397535340773e-06,
+      "loss": 0.4609,
+      "step": 2374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.128442572523418e-06,
+      "loss": 0.4561,
+      "step": 2375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.121488034492569e-06,
+      "loss": 0.4664,
+      "step": 2376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.114533924637778e-06,
+      "loss": 0.4584,
+      "step": 2377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.107580246348395e-06,
+      "loss": 0.4585,
+      "step": 2378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.100627003013563e-06,
+      "loss": 0.4559,
+      "step": 2379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.093674198022201e-06,
+      "loss": 0.4618,
+      "step": 2380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.086721834763024e-06,
+      "loss": 0.4612,
+      "step": 2381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.07976991662453e-06,
+      "loss": 0.4607,
+      "step": 2382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.072818446995e-06,
+      "loss": 0.4562,
+      "step": 2383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.065867429262497e-06,
+      "loss": 0.4512,
+      "step": 2384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.058916866814857e-06,
+      "loss": 0.4637,
+      "step": 2385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.051966763039706e-06,
+      "loss": 0.4631,
+      "step": 2386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.045017121324438e-06,
+      "loss": 0.452,
+      "step": 2387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.038067945056229e-06,
+      "loss": 0.4621,
+      "step": 2388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.031119237622011e-06,
+      "loss": 0.4653,
+      "step": 2389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.024171002408507e-06,
+      "loss": 0.4646,
+      "step": 2390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.017223242802205e-06,
+      "loss": 0.4581,
+      "step": 2391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01027596218935e-06,
+      "loss": 0.465,
+      "step": 2392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.003329163955973e-06,
+      "loss": 0.4575,
+      "step": 2393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.996382851487851e-06,
+      "loss": 0.4681,
+      "step": 2394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.989437028170537e-06,
+      "loss": 0.4591,
+      "step": 2395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.982491697389339e-06,
+      "loss": 0.461,
+      "step": 2396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.975546862529328e-06,
+      "loss": 0.4661,
+      "step": 2397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.968602526975329e-06,
+      "loss": 0.4466,
+      "step": 2398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.961658694111929e-06,
+      "loss": 0.4519,
+      "step": 2399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.4575,
+      "step": 2400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.947772549994037e-06,
+      "loss": 0.4556,
+      "step": 2401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.940830245507483e-06,
+      "loss": 0.4597,
+      "step": 2402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.933888457247402e-06,
+      "loss": 0.464,
+      "step": 2403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.926947188597133e-06,
+      "loss": 0.4542,
+      "step": 2404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.920006442939772e-06,
+      "loss": 0.4553,
+      "step": 2405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.913066223658152e-06,
+      "loss": 0.4571,
+      "step": 2406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.906126534134849e-06,
+      "loss": 0.456,
+      "step": 2407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.89918737775218e-06,
+      "loss": 0.4597,
+      "step": 2408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.892248757892215e-06,
+      "loss": 0.4612,
+      "step": 2409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.885310677936746e-06,
+      "loss": 0.4599,
+      "step": 2410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.878373141267312e-06,
+      "loss": 0.4668,
+      "step": 2411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.871436151265183e-06,
+      "loss": 0.4638,
+      "step": 2412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.864499711311362e-06,
+      "loss": 0.4604,
+      "step": 2413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.857563824786598e-06,
+      "loss": 0.4515,
+      "step": 2414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.850628495071336e-06,
+      "loss": 0.4651,
+      "step": 2415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.843693725545787e-06,
+      "loss": 0.4507,
+      "step": 2416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.836759519589869e-06,
+      "loss": 0.4545,
+      "step": 2417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829825880583228e-06,
+      "loss": 0.4618,
+      "step": 2418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822892811905237e-06,
+      "loss": 0.4648,
+      "step": 2419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815960316934991e-06,
+      "loss": 0.4573,
+      "step": 2420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.809028399051302e-06,
+      "loss": 0.455,
+      "step": 2421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802097061632706e-06,
+      "loss": 0.4571,
+      "step": 2422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79516630805745e-06,
+      "loss": 0.4569,
+      "step": 2423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788236141703498e-06,
+      "loss": 0.4542,
+      "step": 2424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781306565948528e-06,
+      "loss": 0.463,
+      "step": 2425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774377584169934e-06,
+      "loss": 0.4582,
+      "step": 2426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.767449199744813e-06,
+      "loss": 0.4617,
+      "step": 2427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.760521416049983e-06,
+      "loss": 0.4591,
+      "step": 2428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753594236461957e-06,
+      "loss": 0.456,
+      "step": 2429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746667664356957e-06,
+      "loss": 0.4547,
+      "step": 2430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739741703110914e-06,
+      "loss": 0.4546,
+      "step": 2431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.732816356099455e-06,
+      "loss": 0.4525,
+      "step": 2432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.725891626697912e-06,
+      "loss": 0.4593,
+      "step": 2433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.718967518281307e-06,
+      "loss": 0.4604,
+      "step": 2434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.712044034224374e-06,
+      "loss": 0.4661,
+      "step": 2435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.705121177901532e-06,
+      "loss": 0.4632,
+      "step": 2436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.698198952686896e-06,
+      "loss": 0.4568,
+      "step": 2437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.69127736195428e-06,
+      "loss": 0.4639,
+      "step": 2438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.684356409077177e-06,
+      "loss": 0.4669,
+      "step": 2439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.677436097428775e-06,
+      "loss": 0.4605,
+      "step": 2440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.670516430381958e-06,
+      "loss": 0.4631,
+      "step": 2441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.663597411309278e-06,
+      "loss": 0.4536,
+      "step": 2442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.656679043582986e-06,
+      "loss": 0.4542,
+      "step": 2443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.649761330575009e-06,
+      "loss": 0.4586,
+      "step": 2444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.642844275656957e-06,
+      "loss": 0.4537,
+      "step": 2445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.635927882200117e-06,
+      "loss": 0.4628,
+      "step": 2446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.629012153575458e-06,
+      "loss": 0.4569,
+      "step": 2447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.62209709315362e-06,
+      "loss": 0.4583,
+      "step": 2448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.615182704304918e-06,
+      "loss": 0.4575,
+      "step": 2449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.60826899039935e-06,
+      "loss": 0.4593,
+      "step": 2450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.601355954806562e-06,
+      "loss": 0.4487,
+      "step": 2451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.594443600895892e-06,
+      "loss": 0.4534,
+      "step": 2452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.587531932036334e-06,
+      "loss": 0.4601,
+      "step": 2453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.580620951596556e-06,
+      "loss": 0.4594,
+      "step": 2454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.573710662944884e-06,
+      "loss": 0.4501,
+      "step": 2455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.566801069449307e-06,
+      "loss": 0.455,
+      "step": 2456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.559892174477478e-06,
+      "loss": 0.4644,
+      "step": 2457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.552983981396709e-06,
+      "loss": 0.46,
+      "step": 2458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.546076493573973e-06,
+      "loss": 0.4568,
+      "step": 2459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.539169714375885e-06,
+      "loss": 0.4527,
+      "step": 2460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.532263647168735e-06,
+      "loss": 0.4542,
+      "step": 2461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.525358295318454e-06,
+      "loss": 0.4487,
+      "step": 2462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.518453662190622e-06,
+      "loss": 0.4608,
+      "step": 2463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.511549751150478e-06,
+      "loss": 0.4588,
+      "step": 2464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.504646565562907e-06,
+      "loss": 0.4515,
+      "step": 2465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.49774410879243e-06,
+      "loss": 0.4553,
+      "step": 2466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.490842384203227e-06,
+      "loss": 0.4568,
+      "step": 2467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.483941395159114e-06,
+      "loss": 0.4428,
+      "step": 2468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.477041145023546e-06,
+      "loss": 0.4505,
+      "step": 2469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.47014163715962e-06,
+      "loss": 0.4612,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.46324287493008e-06,
+      "loss": 0.451,
+      "step": 2471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.45634486169729e-06,
+      "loss": 0.4635,
+      "step": 2472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.449447600823262e-06,
+      "loss": 0.4593,
+      "step": 2473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.44255109566964e-06,
+      "loss": 0.4506,
+      "step": 2474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 0.454,
+      "step": 2475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.428760365968327e-06,
+      "loss": 0.4548,
+      "step": 2476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.421866148142066e-06,
+      "loss": 0.4627,
+      "step": 2477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.414972699479076e-06,
+      "loss": 0.4545,
+      "step": 2478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.408080023339134e-06,
+      "loss": 0.4507,
+      "step": 2479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.401188123081653e-06,
+      "loss": 0.46,
+      "step": 2480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.394297002065658e-06,
+      "loss": 0.4563,
+      "step": 2481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.387406663649796e-06,
+      "loss": 0.4545,
+      "step": 2482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.380517111192336e-06,
+      "loss": 0.4542,
+      "step": 2483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.373628348051165e-06,
+      "loss": 0.4587,
+      "step": 2484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.366740377583781e-06,
+      "loss": 0.4638,
+      "step": 2485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.35985320314729e-06,
+      "loss": 0.4476,
+      "step": 2486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.352966828098428e-06,
+      "loss": 0.4534,
+      "step": 2487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.346081255793524e-06,
+      "loss": 0.4485,
+      "step": 2488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.339196489588522e-06,
+      "loss": 0.4539,
+      "step": 2489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.332312532838978e-06,
+      "loss": 0.4444,
+      "step": 2490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.325429388900046e-06,
+      "loss": 0.4567,
+      "step": 2491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.318547061126485e-06,
+      "loss": 0.4557,
+      "step": 2492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.311665552872662e-06,
+      "loss": 0.4568,
+      "step": 2493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.30478486749254e-06,
+      "loss": 0.4511,
+      "step": 2494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.297905008339677e-06,
+      "loss": 0.4558,
+      "step": 2495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.291025978767236e-06,
+      "loss": 0.4525,
+      "step": 2496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.284147782127971e-06,
+      "loss": 0.4542,
+      "step": 2497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.277270421774234e-06,
+      "loss": 0.4573,
+      "step": 2498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.270393901057964e-06,
+      "loss": 0.4607,
+      "step": 2499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.4633,
+      "step": 2500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.25664339194355e-06,
+      "loss": 0.4601,
+      "step": 2501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.249769410247239e-06,
+      "loss": 0.4668,
+      "step": 2502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.242896281592057e-06,
+      "loss": 0.4496,
+      "step": 2503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.236024009327879e-06,
+      "loss": 0.4521,
+      "step": 2504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.22915259680417e-06,
+      "loss": 0.452,
+      "step": 2505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.222282047369972e-06,
+      "loss": 0.4591,
+      "step": 2506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.215412364373908e-06,
+      "loss": 0.4515,
+      "step": 2507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.208543551164178e-06,
+      "loss": 0.4539,
+      "step": 2508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.201675611088558e-06,
+      "loss": 0.4596,
+      "step": 2509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.194808547494401e-06,
+      "loss": 0.4548,
+      "step": 2510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.187942363728626e-06,
+      "loss": 0.4677,
+      "step": 2511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.181077063137733e-06,
+      "loss": 0.4641,
+      "step": 2512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.174212649067781e-06,
+      "loss": 0.4536,
+      "step": 2513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.167349124864406e-06,
+      "loss": 0.4646,
+      "step": 2514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.1604864938728e-06,
+      "loss": 0.4539,
+      "step": 2515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.153624759437733e-06,
+      "loss": 0.4471,
+      "step": 2516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.146763924903527e-06,
+      "loss": 0.4497,
+      "step": 2517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.139903993614069e-06,
+      "loss": 0.4539,
+      "step": 2518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.133044968912811e-06,
+      "loss": 0.4539,
+      "step": 2519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 0.4575,
+      "step": 2520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.119329652646463e-06,
+      "loss": 0.4494,
+      "step": 2521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.112473367766051e-06,
+      "loss": 0.4491,
+      "step": 2522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.10561800284319e-06,
+      "loss": 0.4629,
+      "step": 2523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.098763561219101e-06,
+      "loss": 0.4576,
+      "step": 2524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.091910046234552e-06,
+      "loss": 0.4521,
+      "step": 2525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.08505746122987e-06,
+      "loss": 0.4536,
+      "step": 2526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.078205809544918e-06,
+      "loss": 0.4461,
+      "step": 2527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.07135509451911e-06,
+      "loss": 0.453,
+      "step": 2528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.064505319491398e-06,
+      "loss": 0.4443,
+      "step": 2529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.057656487800283e-06,
+      "loss": 0.4626,
+      "step": 2530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.050808602783797e-06,
+      "loss": 0.4447,
+      "step": 2531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.04396166777952e-06,
+      "loss": 0.4543,
+      "step": 2532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.037115686124564e-06,
+      "loss": 0.4566,
+      "step": 2533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.030270661155575e-06,
+      "loss": 0.4513,
+      "step": 2534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.023426596208739e-06,
+      "loss": 0.454,
+      "step": 2535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.016583494619769e-06,
+      "loss": 0.4534,
+      "step": 2536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.009741359723906e-06,
+      "loss": 0.4568,
+      "step": 2537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.00290019485593e-06,
+      "loss": 0.4561,
+      "step": 2538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.996060003350139e-06,
+      "loss": 0.4541,
+      "step": 2539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.989220788540356e-06,
+      "loss": 0.452,
+      "step": 2540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.982382553759931e-06,
+      "loss": 0.4478,
+      "step": 2541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.975545302341743e-06,
+      "loss": 0.4513,
+      "step": 2542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.96870903761818e-06,
+      "loss": 0.4508,
+      "step": 2543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.961873762921153e-06,
+      "loss": 0.4506,
+      "step": 2544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.955039481582098e-06,
+      "loss": 0.4528,
+      "step": 2545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.948206196931953e-06,
+      "loss": 0.4644,
+      "step": 2546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.94137391230119e-06,
+      "loss": 0.4477,
+      "step": 2547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.934542631019767e-06,
+      "loss": 0.4528,
+      "step": 2548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.927712356417176e-06,
+      "loss": 0.4474,
+      "step": 2549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 0.4466,
+      "step": 2550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.914054840563962e-06,
+      "loss": 0.4461,
+      "step": 2551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.907227605969849e-06,
+      "loss": 0.457,
+      "step": 2552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.900401391367576e-06,
+      "loss": 0.4506,
+      "step": 2553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.89357620008416e-06,
+      "loss": 0.4384,
+      "step": 2554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.886752035446116e-06,
+      "loss": 0.4511,
+      "step": 2555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.879928900779457e-06,
+      "loss": 0.4602,
+      "step": 2556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.873106799409696e-06,
+      "loss": 0.4478,
+      "step": 2557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.866285734661842e-06,
+      "loss": 0.4516,
+      "step": 2558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.8594657098604e-06,
+      "loss": 0.4474,
+      "step": 2559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.852646728329368e-06,
+      "loss": 0.4569,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.845828793392236e-06,
+      "loss": 0.4526,
+      "step": 2561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.83901190837198e-06,
+      "loss": 0.4552,
+      "step": 2562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.832196076591067e-06,
+      "loss": 0.4464,
+      "step": 2563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.825381301371452e-06,
+      "loss": 0.4517,
+      "step": 2564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.818567586034578e-06,
+      "loss": 0.446,
+      "step": 2565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.811754933901358e-06,
+      "loss": 0.4526,
+      "step": 2566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.804943348292197e-06,
+      "loss": 0.453,
+      "step": 2567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.798132832526986e-06,
+      "loss": 0.4496,
+      "step": 2568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.791323389925084e-06,
+      "loss": 0.448,
+      "step": 2569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.784515023805328e-06,
+      "loss": 0.4567,
+      "step": 2570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.777707737486036e-06,
+      "loss": 0.4535,
+      "step": 2571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.770901534284996e-06,
+      "loss": 0.4545,
+      "step": 2572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.76409641751947e-06,
+      "loss": 0.456,
+      "step": 2573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.757292390506191e-06,
+      "loss": 0.4456,
+      "step": 2574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.750489456561351e-06,
+      "loss": 0.4404,
+      "step": 2575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.743687619000625e-06,
+      "loss": 0.4534,
+      "step": 2576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.736886881139143e-06,
+      "loss": 0.4438,
+      "step": 2577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.730087246291503e-06,
+      "loss": 0.453,
+      "step": 2578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.72328871777176e-06,
+      "loss": 0.445,
+      "step": 2579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.716491298893443e-06,
+      "loss": 0.4545,
+      "step": 2580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.709694992969525e-06,
+      "loss": 0.4456,
+      "step": 2581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.702899803312443e-06,
+      "loss": 0.4545,
+      "step": 2582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.696105733234099e-06,
+      "loss": 0.4533,
+      "step": 2583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.689312786045823e-06,
+      "loss": 0.4458,
+      "step": 2584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.68252096505843e-06,
+      "loss": 0.4587,
+      "step": 2585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.67573027358216e-06,
+      "loss": 0.4555,
+      "step": 2586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.668940714926724e-06,
+      "loss": 0.4528,
+      "step": 2587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.662152292401265e-06,
+      "loss": 0.4563,
+      "step": 2588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.655365009314375e-06,
+      "loss": 0.4542,
+      "step": 2589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.6485788689741e-06,
+      "loss": 0.4564,
+      "step": 2590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.641793874687918e-06,
+      "loss": 0.45,
+      "step": 2591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.635010029762755e-06,
+      "loss": 0.4601,
+      "step": 2592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.628227337504972e-06,
+      "loss": 0.4634,
+      "step": 2593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.621445801220372e-06,
+      "loss": 0.4405,
+      "step": 2594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.6146654242141935e-06,
+      "loss": 0.4585,
+      "step": 2595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.6078862097911075e-06,
+      "loss": 0.4538,
+      "step": 2596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.6011081612552265e-06,
+      "loss": 0.4506,
+      "step": 2597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.594331281910082e-06,
+      "loss": 0.4534,
+      "step": 2598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.58755557505865e-06,
+      "loss": 0.4585,
+      "step": 2599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.580781044003324e-06,
+      "loss": 0.4476,
+      "step": 2600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.574007692045928e-06,
+      "loss": 0.4523,
+      "step": 2601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.5672355224877115e-06,
+      "loss": 0.4483,
+      "step": 2602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.560464538629345e-06,
+      "loss": 0.4492,
+      "step": 2603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.553694743770928e-06,
+      "loss": 0.4596,
+      "step": 2604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.546926141211975e-06,
+      "loss": 0.4528,
+      "step": 2605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.54015873425142e-06,
+      "loss": 0.4465,
+      "step": 2606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.533392526187617e-06,
+      "loss": 0.4466,
+      "step": 2607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.526627520318329e-06,
+      "loss": 0.4522,
+      "step": 2608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.519863719940748e-06,
+      "loss": 0.4572,
+      "step": 2609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.513101128351454e-06,
+      "loss": 0.4627,
+      "step": 2610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.506339748846461e-06,
+      "loss": 0.4621,
+      "step": 2611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.49957958472118e-06,
+      "loss": 0.4542,
+      "step": 2612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.492820639270435e-06,
+      "loss": 0.4537,
+      "step": 2613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.486062915788453e-06,
+      "loss": 0.4495,
+      "step": 2614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.4793064175688635e-06,
+      "loss": 0.454,
+      "step": 2615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.472551147904708e-06,
+      "loss": 0.454,
+      "step": 2616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.465797110088417e-06,
+      "loss": 0.4457,
+      "step": 2617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.4590443074118325e-06,
+      "loss": 0.4454,
+      "step": 2618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.4522927431661805e-06,
+      "loss": 0.4429,
+      "step": 2619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.445542420642097e-06,
+      "loss": 0.4499,
+      "step": 2620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.438793343129605e-06,
+      "loss": 0.4542,
+      "step": 2621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.432045513918122e-06,
+      "loss": 0.4562,
+      "step": 2622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.4252989362964635e-06,
+      "loss": 0.4518,
+      "step": 2623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.418553613552824e-06,
+      "loss": 0.4488,
+      "step": 2624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.4427,
+      "step": 2625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.405066745849347e-06,
+      "loss": 0.4601,
+      "step": 2626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.398325207462846e-06,
+      "loss": 0.4591,
+      "step": 2627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.391584937101034e-06,
+      "loss": 0.4486,
+      "step": 2628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.384845938049033e-06,
+      "loss": 0.4444,
+      "step": 2629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.378108213591355e-06,
+      "loss": 0.4532,
+      "step": 2630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.37137176701188e-06,
+      "loss": 0.4482,
+      "step": 2631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.364636601593875e-06,
+      "loss": 0.4592,
+      "step": 2632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.357902720619976e-06,
+      "loss": 0.4441,
+      "step": 2633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.351170127372191e-06,
+      "loss": 0.4489,
+      "step": 2634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.344438825131912e-06,
+      "loss": 0.4438,
+      "step": 2635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.33770881717989e-06,
+      "loss": 0.453,
+      "step": 2636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.330980106796247e-06,
+      "loss": 0.458,
+      "step": 2637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.324252697260475e-06,
+      "loss": 0.4456,
+      "step": 2638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.3175265918514335e-06,
+      "loss": 0.4549,
+      "step": 2639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.310801793847344e-06,
+      "loss": 0.449,
+      "step": 2640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.3040783065257906e-06,
+      "loss": 0.4561,
+      "step": 2641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.297356133163722e-06,
+      "loss": 0.4561,
+      "step": 2642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.290635277037442e-06,
+      "loss": 0.454,
+      "step": 2643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.283915741422611e-06,
+      "loss": 0.4575,
+      "step": 2644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.277197529594257e-06,
+      "loss": 0.4482,
+      "step": 2645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.27048064482675e-06,
+      "loss": 0.4536,
+      "step": 2646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.263765090393817e-06,
+      "loss": 0.4533,
+      "step": 2647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.257050869568536e-06,
+      "loss": 0.4497,
+      "step": 2648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.250337985623342e-06,
+      "loss": 0.4405,
+      "step": 2649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.243626441830009e-06,
+      "loss": 0.4477,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.236916241459664e-06,
+      "loss": 0.4498,
+      "step": 2651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2302073877827775e-06,
+      "loss": 0.4547,
+      "step": 2652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.22349988406916e-06,
+      "loss": 0.4412,
+      "step": 2653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.216793733587976e-06,
+      "loss": 0.4481,
+      "step": 2654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.210088939607709e-06,
+      "loss": 0.4494,
+      "step": 2655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.203385505396203e-06,
+      "loss": 0.4574,
+      "step": 2656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.196683434220626e-06,
+      "loss": 0.4452,
+      "step": 2657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.189982729347491e-06,
+      "loss": 0.4578,
+      "step": 2658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.1832833940426346e-06,
+      "loss": 0.4555,
+      "step": 2659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.176585431571235e-06,
+      "loss": 0.4492,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.169888845197798e-06,
+      "loss": 0.4445,
+      "step": 2661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.163193638186159e-06,
+      "loss": 0.4576,
+      "step": 2662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.156499813799477e-06,
+      "loss": 0.4427,
+      "step": 2663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.149807375300239e-06,
+      "loss": 0.4522,
+      "step": 2664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.143116325950266e-06,
+      "loss": 0.4487,
+      "step": 2665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.13642666901069e-06,
+      "loss": 0.4525,
+      "step": 2666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.129738407741964e-06,
+      "loss": 0.4481,
+      "step": 2667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.123051545403874e-06,
+      "loss": 0.4465,
+      "step": 2668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.116366085255511e-06,
+      "loss": 0.4586,
+      "step": 2669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.109682030555283e-06,
+      "loss": 0.4467,
+      "step": 2670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.102999384560927e-06,
+      "loss": 0.4489,
+      "step": 2671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.096318150529476e-06,
+      "loss": 0.4579,
+      "step": 2672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.0896383317172845e-06,
+      "loss": 0.4555,
+      "step": 2673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.082959931380011e-06,
+      "loss": 0.4469,
+      "step": 2674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.076282952772634e-06,
+      "loss": 0.4523,
+      "step": 2675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.069607399149427e-06,
+      "loss": 0.4591,
+      "step": 2676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.062933273763974e-06,
+      "loss": 0.4478,
+      "step": 2677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.056260579869165e-06,
+      "loss": 0.4485,
+      "step": 2678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.049589320717186e-06,
+      "loss": 0.4556,
+      "step": 2679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.042919499559538e-06,
+      "loss": 0.4493,
+      "step": 2680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.036251119646993e-06,
+      "loss": 0.4505,
+      "step": 2681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.029584184229653e-06,
+      "loss": 0.454,
+      "step": 2682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.022918696556896e-06,
+      "loss": 0.4537,
+      "step": 2683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.016254659877398e-06,
+      "loss": 0.4496,
+      "step": 2684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.009592077439135e-06,
+      "loss": 0.4531,
+      "step": 2685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.002930952489362e-06,
+      "loss": 0.458,
+      "step": 2686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.996271288274636e-06,
+      "loss": 0.4542,
+      "step": 2687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.9896130880407965e-06,
+      "loss": 0.4504,
+      "step": 2688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.982956355032968e-06,
+      "loss": 0.4461,
+      "step": 2689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.976301092495556e-06,
+      "loss": 0.444,
+      "step": 2690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.969647303672262e-06,
+      "loss": 0.4505,
+      "step": 2691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.962994991806059e-06,
+      "loss": 0.4429,
+      "step": 2692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.956344160139201e-06,
+      "loss": 0.4546,
+      "step": 2693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.949694811913226e-06,
+      "loss": 0.4452,
+      "step": 2694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.943046950368944e-06,
+      "loss": 0.4471,
+      "step": 2695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.9364005787464406e-06,
+      "loss": 0.4507,
+      "step": 2696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.929755700285082e-06,
+      "loss": 0.4541,
+      "step": 2697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.923112318223497e-06,
+      "loss": 0.446,
+      "step": 2698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.9164704357995874e-06,
+      "loss": 0.4465,
+      "step": 2699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.4464,
+      "step": 2700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.903191182812759e-06,
+      "loss": 0.4562,
+      "step": 2701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.896553818721989e-06,
+      "loss": 0.4456,
+      "step": 2702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.889917967213184e-06,
+      "loss": 0.4518,
+      "step": 2703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.883283631520582e-06,
+      "loss": 0.4549,
+      "step": 2704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.876650814877675e-06,
+      "loss": 0.4378,
+      "step": 2705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.870019520517217e-06,
+      "loss": 0.4566,
+      "step": 2706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.863389751671225e-06,
+      "loss": 0.4478,
+      "step": 2707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.856761511570963e-06,
+      "loss": 0.4464,
+      "step": 2708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.850134803446955e-06,
+      "loss": 0.4416,
+      "step": 2709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.843509630528977e-06,
+      "loss": 0.4539,
+      "step": 2710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.836885996046061e-06,
+      "loss": 0.4456,
+      "step": 2711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.830263903226483e-06,
+      "loss": 0.448,
+      "step": 2712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.823643355297774e-06,
+      "loss": 0.447,
+      "step": 2713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.8170243554867065e-06,
+      "loss": 0.4394,
+      "step": 2714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.8104069070193e-06,
+      "loss": 0.4483,
+      "step": 2715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.803791013120822e-06,
+      "loss": 0.4412,
+      "step": 2716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.797176677015775e-06,
+      "loss": 0.4556,
+      "step": 2717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.790563901927907e-06,
+      "loss": 0.4429,
+      "step": 2718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.783952691080203e-06,
+      "loss": 0.4475,
+      "step": 2719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.777343047694891e-06,
+      "loss": 0.4441,
+      "step": 2720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.770734974993427e-06,
+      "loss": 0.4535,
+      "step": 2721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.764128476196505e-06,
+      "loss": 0.4556,
+      "step": 2722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.757523554524056e-06,
+      "loss": 0.4521,
+      "step": 2723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.750920213195238e-06,
+      "loss": 0.4497,
+      "step": 2724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.744318455428436e-06,
+      "loss": 0.4519,
+      "step": 2725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.737718284441267e-06,
+      "loss": 0.4453,
+      "step": 2726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.731119703450577e-06,
+      "loss": 0.4335,
+      "step": 2727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.7245227156724324e-06,
+      "loss": 0.4427,
+      "step": 2728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.717927324322124e-06,
+      "loss": 0.4546,
+      "step": 2729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.711333532614168e-06,
+      "loss": 0.4536,
+      "step": 2730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.704741343762296e-06,
+      "loss": 0.4459,
+      "step": 2731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.698150760979463e-06,
+      "loss": 0.4568,
+      "step": 2732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.69156178747784e-06,
+      "loss": 0.4501,
+      "step": 2733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.684974426468809e-06,
+      "loss": 0.4514,
+      "step": 2734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.67838868116297e-06,
+      "loss": 0.4456,
+      "step": 2735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.671804554770135e-06,
+      "loss": 0.4514,
+      "step": 2736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.6652220504993305e-06,
+      "loss": 0.447,
+      "step": 2737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.658641171558785e-06,
+      "loss": 0.4479,
+      "step": 2738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.6520619211559435e-06,
+      "loss": 0.4512,
+      "step": 2739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.645484302497452e-06,
+      "loss": 0.4575,
+      "step": 2740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.638908318789156e-06,
+      "loss": 0.4495,
+      "step": 2741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.63233397323612e-06,
+      "loss": 0.455,
+      "step": 2742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.62576126904259e-06,
+      "loss": 0.4519,
+      "step": 2743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.6191902094120295e-06,
+      "loss": 0.4501,
+      "step": 2744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.612620797547087e-06,
+      "loss": 0.4509,
+      "step": 2745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.60605303664962e-06,
+      "loss": 0.4528,
+      "step": 2746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.5994869299206736e-06,
+      "loss": 0.4479,
+      "step": 2747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.5929224805604845e-06,
+      "loss": 0.4498,
+      "step": 2748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.58635969176849e-06,
+      "loss": 0.4476,
+      "step": 2749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.4441,
+      "step": 2750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.573239108682769e-06,
+      "loss": 0.4529,
+      "step": 2751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.566681320783849e-06,
+      "loss": 0.4455,
+      "step": 2752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.560125206242746e-06,
+      "loss": 0.4428,
+      "step": 2753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.553570768254831e-06,
+      "loss": 0.453,
+      "step": 2754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.547018010014654e-06,
+      "loss": 0.4374,
+      "step": 2755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.540466934715953e-06,
+      "loss": 0.4519,
+      "step": 2756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.53391754555164e-06,
+      "loss": 0.4529,
+      "step": 2757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.52736984571381e-06,
+      "loss": 0.4465,
+      "step": 2758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.520823838393732e-06,
+      "loss": 0.4498,
+      "step": 2759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.5142795267818505e-06,
+      "loss": 0.4452,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.5077369140677815e-06,
+      "loss": 0.4403,
+      "step": 2761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.501196003440313e-06,
+      "loss": 0.4516,
+      "step": 2762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.494656798087412e-06,
+      "loss": 0.4436,
+      "step": 2763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.488119301196201e-06,
+      "loss": 0.4433,
+      "step": 2764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.481583515952983e-06,
+      "loss": 0.4458,
+      "step": 2765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.475049445543215e-06,
+      "loss": 0.4431,
+      "step": 2766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.468517093151525e-06,
+      "loss": 0.4453,
+      "step": 2767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.461986461961706e-06,
+      "loss": 0.4377,
+      "step": 2768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.455457555156706e-06,
+      "loss": 0.4438,
+      "step": 2769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.448930375918632e-06,
+      "loss": 0.449,
+      "step": 2770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.442404927428751e-06,
+      "loss": 0.4443,
+      "step": 2771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.435881212867494e-06,
+      "loss": 0.4402,
+      "step": 2772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.4293592354144365e-06,
+      "loss": 0.4448,
+      "step": 2773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.422838998248308e-06,
+      "loss": 0.4417,
+      "step": 2774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.4163205045469975e-06,
+      "loss": 0.4474,
+      "step": 2775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.409803757487539e-06,
+      "loss": 0.4427,
+      "step": 2776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.403288760246112e-06,
+      "loss": 0.4505,
+      "step": 2777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.396775515998055e-06,
+      "loss": 0.4418,
+      "step": 2778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.390264027917836e-06,
+      "loss": 0.4448,
+      "step": 2779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.383754299179079e-06,
+      "loss": 0.4538,
+      "step": 2780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.377246332954544e-06,
+      "loss": 0.4514,
+      "step": 2781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.370740132416138e-06,
+      "loss": 0.4383,
+      "step": 2782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.364235700734903e-06,
+      "loss": 0.4518,
+      "step": 2783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.357733041081018e-06,
+      "loss": 0.4416,
+      "step": 2784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.351232156623803e-06,
+      "loss": 0.4354,
+      "step": 2785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.344733050531713e-06,
+      "loss": 0.4505,
+      "step": 2786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.338235725972326e-06,
+      "loss": 0.4558,
+      "step": 2787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.33174018611236e-06,
+      "loss": 0.4465,
+      "step": 2788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.325246434117669e-06,
+      "loss": 0.447,
+      "step": 2789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.318754473153221e-06,
+      "loss": 0.4541,
+      "step": 2790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.3122643063831245e-06,
+      "loss": 0.4458,
+      "step": 2791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.305775936970606e-06,
+      "loss": 0.4543,
+      "step": 2792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.299289368078016e-06,
+      "loss": 0.4452,
+      "step": 2793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.292804602866833e-06,
+      "loss": 0.4353,
+      "step": 2794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.286321644497655e-06,
+      "loss": 0.446,
+      "step": 2795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.27984049613019e-06,
+      "loss": 0.444,
+      "step": 2796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.273361160923271e-06,
+      "loss": 0.4618,
+      "step": 2797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.2668836420348535e-06,
+      "loss": 0.4466,
+      "step": 2798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.260407942621998e-06,
+      "loss": 0.4515,
+      "step": 2799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.25393406584088e-06,
+      "loss": 0.457,
+      "step": 2800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.247462014846793e-06,
+      "loss": 0.4542,
+      "step": 2801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.240991792794133e-06,
+      "loss": 0.4445,
+      "step": 2802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.234523402836408e-06,
+      "loss": 0.4506,
+      "step": 2803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.228056848126236e-06,
+      "loss": 0.4434,
+      "step": 2804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.22159213181533e-06,
+      "loss": 0.4416,
+      "step": 2805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.2151292570545215e-06,
+      "loss": 0.445,
+      "step": 2806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.208668226993731e-06,
+      "loss": 0.447,
+      "step": 2807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.202209044781991e-06,
+      "loss": 0.4479,
+      "step": 2808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.195751713567426e-06,
+      "loss": 0.4533,
+      "step": 2809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.18929623649726e-06,
+      "loss": 0.4403,
+      "step": 2810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.182842616717817e-06,
+      "loss": 0.4362,
+      "step": 2811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.176390857374508e-06,
+      "loss": 0.4429,
+      "step": 2812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.169940961611853e-06,
+      "loss": 0.4403,
+      "step": 2813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.1634929325734385e-06,
+      "loss": 0.451,
+      "step": 2814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.157046773401964e-06,
+      "loss": 0.445,
+      "step": 2815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.150602487239207e-06,
+      "loss": 0.4368,
+      "step": 2816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.144160077226035e-06,
+      "loss": 0.4467,
+      "step": 2817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.137719546502401e-06,
+      "loss": 0.4408,
+      "step": 2818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.131280898207339e-06,
+      "loss": 0.4468,
+      "step": 2819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.124844135478971e-06,
+      "loss": 0.4418,
+      "step": 2820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.118409261454494e-06,
+      "loss": 0.4431,
+      "step": 2821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.1119762792701935e-06,
+      "loss": 0.4372,
+      "step": 2822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.1055451920614165e-06,
+      "loss": 0.4428,
+      "step": 2823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.099116002962604e-06,
+      "loss": 0.4528,
+      "step": 2824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.092688715107265e-06,
+      "loss": 0.4404,
+      "step": 2825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.086263331627976e-06,
+      "loss": 0.4427,
+      "step": 2826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.079839855656397e-06,
+      "loss": 0.4484,
+      "step": 2827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.073418290323251e-06,
+      "loss": 0.4465,
+      "step": 2828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.066998638758326e-06,
+      "loss": 0.437,
+      "step": 2829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.06058090409049e-06,
+      "loss": 0.4508,
+      "step": 2830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.054165089447663e-06,
+      "loss": 0.4545,
+      "step": 2831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.047751197956838e-06,
+      "loss": 0.4406,
+      "step": 2832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0413392327440635e-06,
+      "loss": 0.4543,
+      "step": 2833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0349291969344595e-06,
+      "loss": 0.4367,
+      "step": 2834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.028521093652195e-06,
+      "loss": 0.4509,
+      "step": 2835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.022114926020504e-06,
+      "loss": 0.4483,
+      "step": 2836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.015710697161674e-06,
+      "loss": 0.4471,
+      "step": 2837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.009308410197048e-06,
+      "loss": 0.4434,
+      "step": 2838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.002908068247024e-06,
+      "loss": 0.4467,
+      "step": 2839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.996509674431053e-06,
+      "loss": 0.4472,
+      "step": 2840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.990113231867629e-06,
+      "loss": 0.4474,
+      "step": 2841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.983718743674302e-06,
+      "loss": 0.4364,
+      "step": 2842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.977326212967671e-06,
+      "loss": 0.4406,
+      "step": 2843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.970935642863375e-06,
+      "loss": 0.452,
+      "step": 2844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9645470364761e-06,
+      "loss": 0.4289,
+      "step": 2845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.958160396919577e-06,
+      "loss": 0.448,
+      "step": 2846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.951775727306577e-06,
+      "loss": 0.4428,
+      "step": 2847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.94539303074891e-06,
+      "loss": 0.4455,
+      "step": 2848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.939012310357422e-06,
+      "loss": 0.4531,
+      "step": 2849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 0.4492,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926256810511566e-06,
+      "loss": 0.4431,
+      "step": 2851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9198820372740726e-06,
+      "loss": 0.4424,
+      "step": 2852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.913509252636511e-06,
+      "loss": 0.4369,
+      "step": 2853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.907138459704895e-06,
+      "loss": 0.4378,
+      "step": 2854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.900769661584273e-06,
+      "loss": 0.4447,
+      "step": 2855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.894402861378721e-06,
+      "loss": 0.4464,
+      "step": 2856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.88803806219134e-06,
+      "loss": 0.4492,
+      "step": 2857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.881675267124254e-06,
+      "loss": 0.4409,
+      "step": 2858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.8753144792786095e-06,
+      "loss": 0.4445,
+      "step": 2859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.868955701754584e-06,
+      "loss": 0.4403,
+      "step": 2860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.862598937651364e-06,
+      "loss": 0.4392,
+      "step": 2861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.85624419006716e-06,
+      "loss": 0.4534,
+      "step": 2862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.849891462099199e-06,
+      "loss": 0.4453,
+      "step": 2863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.843540756843722e-06,
+      "loss": 0.4495,
+      "step": 2864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.83719207739599e-06,
+      "loss": 0.4483,
+      "step": 2865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.830845426850268e-06,
+      "loss": 0.4387,
+      "step": 2866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.824500808299836e-06,
+      "loss": 0.4405,
+      "step": 2867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.818158224836987e-06,
+      "loss": 0.4393,
+      "step": 2868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.811817679553018e-06,
+      "loss": 0.4434,
+      "step": 2869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8054791755382286e-06,
+      "loss": 0.4566,
+      "step": 2870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.799142715881938e-06,
+      "loss": 0.4603,
+      "step": 2871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.792808303672454e-06,
+      "loss": 0.4412,
+      "step": 2872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.786475941997094e-06,
+      "loss": 0.4433,
+      "step": 2873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.780145633942173e-06,
+      "loss": 0.4537,
+      "step": 2874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.773817382593008e-06,
+      "loss": 0.4483,
+      "step": 2875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.7674911910339094e-06,
+      "loss": 0.4438,
+      "step": 2876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.761167062348187e-06,
+      "loss": 0.4421,
+      "step": 2877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.754844999618144e-06,
+      "loss": 0.4414,
+      "step": 2878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.748525005925074e-06,
+      "loss": 0.4498,
+      "step": 2879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 0.4373,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.735891237970015e-06,
+      "loss": 0.4365,
+      "step": 2881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.729577469865566e-06,
+      "loss": 0.4344,
+      "step": 2882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.723265783113181e-06,
+      "loss": 0.4446,
+      "step": 2883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.716956180789098e-06,
+      "loss": 0.4425,
+      "step": 2884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.710648665968543e-06,
+      "loss": 0.4429,
+      "step": 2885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.704343241725719e-06,
+      "loss": 0.4394,
+      "step": 2886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.698039911133816e-06,
+      "loss": 0.4377,
+      "step": 2887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.691738677265e-06,
+      "loss": 0.4424,
+      "step": 2888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.685439543190409e-06,
+      "loss": 0.4441,
+      "step": 2889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.679142511980176e-06,
+      "loss": 0.4324,
+      "step": 2890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.672847586703393e-06,
+      "loss": 0.4485,
+      "step": 2891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.666554770428129e-06,
+      "loss": 0.4372,
+      "step": 2892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.660264066221426e-06,
+      "loss": 0.4439,
+      "step": 2893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.653975477149298e-06,
+      "loss": 0.4424,
+      "step": 2894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.647689006276727e-06,
+      "loss": 0.4558,
+      "step": 2895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.641404656667661e-06,
+      "loss": 0.4401,
+      "step": 2896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.6351224313850165e-06,
+      "loss": 0.439,
+      "step": 2897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.628842333490674e-06,
+      "loss": 0.4411,
+      "step": 2898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.622564366045472e-06,
+      "loss": 0.4448,
+      "step": 2899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.616288532109225e-06,
+      "loss": 0.4432,
+      "step": 2900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.610014834740694e-06,
+      "loss": 0.4427,
+      "step": 2901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.603743276997607e-06,
+      "loss": 0.4481,
+      "step": 2902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.59747386193663e-06,
+      "loss": 0.4397,
+      "step": 2903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.591206592613416e-06,
+      "loss": 0.4417,
+      "step": 2904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.584941472082549e-06,
+      "loss": 0.4497,
+      "step": 2905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5786785033975745e-06,
+      "loss": 0.4396,
+      "step": 2906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.572417689610987e-06,
+      "loss": 0.4413,
+      "step": 2907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5661590337742255e-06,
+      "loss": 0.4415,
+      "step": 2908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.559902538937694e-06,
+      "loss": 0.4416,
+      "step": 2909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.553648208150728e-06,
+      "loss": 0.4329,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5473960444616085e-06,
+      "loss": 0.4424,
+      "step": 2911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5411460509175605e-06,
+      "loss": 0.4398,
+      "step": 2912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.534898230564765e-06,
+      "loss": 0.4553,
+      "step": 2913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.5286525864483285e-06,
+      "loss": 0.4363,
+      "step": 2914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.522409121612304e-06,
+      "loss": 0.4411,
+      "step": 2915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.516167839099679e-06,
+      "loss": 0.4459,
+      "step": 2916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.50992874195238e-06,
+      "loss": 0.4414,
+      "step": 2917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.50369183321126e-06,
+      "loss": 0.4446,
+      "step": 2918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.497457115916127e-06,
+      "loss": 0.4416,
+      "step": 2919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.491224593105695e-06,
+      "loss": 0.4371,
+      "step": 2920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.484994267817624e-06,
+      "loss": 0.4315,
+      "step": 2921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.478766143088492e-06,
+      "loss": 0.4326,
+      "step": 2922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.472540221953824e-06,
+      "loss": 0.4354,
+      "step": 2923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.466316507448049e-06,
+      "loss": 0.4444,
+      "step": 2924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.4431,
+      "step": 2925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.453875710455562e-06,
+      "loss": 0.4453,
+      "step": 2926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.447658634032338e-06,
+      "loss": 0.4404,
+      "step": 2927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.441443776365003e-06,
+      "loss": 0.4436,
+      "step": 2928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.435231140482588e-06,
+      "loss": 0.4401,
+      "step": 2929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.429020729413062e-06,
+      "loss": 0.4381,
+      "step": 2930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.4228125461833026e-06,
+      "loss": 0.4374,
+      "step": 2931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.416606593819102e-06,
+      "loss": 0.4348,
+      "step": 2932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.41040287534517e-06,
+      "loss": 0.4417,
+      "step": 2933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.404201393785123e-06,
+      "loss": 0.4448,
+      "step": 2934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.398002152161484e-06,
+      "loss": 0.444,
+      "step": 2935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.391805153495693e-06,
+      "loss": 0.4473,
+      "step": 2936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.385610400808088e-06,
+      "loss": 0.4498,
+      "step": 2937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.379417897117917e-06,
+      "loss": 0.4416,
+      "step": 2938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.373227645443332e-06,
+      "loss": 0.4434,
+      "step": 2939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.367039648801386e-06,
+      "loss": 0.4418,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.360853910208028e-06,
+      "loss": 0.4459,
+      "step": 2941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.354670432678124e-06,
+      "loss": 0.4515,
+      "step": 2942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.348489219225417e-06,
+      "loss": 0.4477,
+      "step": 2943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.342310272862558e-06,
+      "loss": 0.4398,
+      "step": 2944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.336133596601089e-06,
+      "loss": 0.4459,
+      "step": 2945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3299591934514485e-06,
+      "loss": 0.4316,
+      "step": 2946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.323787066422964e-06,
+      "loss": 0.4384,
+      "step": 2947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.317617218523856e-06,
+      "loss": 0.4521,
+      "step": 2948
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.311449652761235e-06,
+      "loss": 0.4441,
+      "step": 2949
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.305284372141095e-06,
+      "loss": 0.4424,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.299121379668316e-06,
+      "loss": 0.4388,
+      "step": 2951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.292960678346674e-06,
+      "loss": 0.4418,
+      "step": 2952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.286802271178815e-06,
+      "loss": 0.4452,
+      "step": 2953
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.280646161166274e-06,
+      "loss": 0.4286,
+      "step": 2954
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.274492351309462e-06,
+      "loss": 0.437,
+      "step": 2955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.26834084460767e-06,
+      "loss": 0.4459,
+      "step": 2956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.262191644059071e-06,
+      "loss": 0.4396,
+      "step": 2957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.256044752660709e-06,
+      "loss": 0.4466,
+      "step": 2958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2499001734085045e-06,
+      "loss": 0.4405,
+      "step": 2959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.243757909297247e-06,
+      "loss": 0.4385,
+      "step": 2960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.237617963320608e-06,
+      "loss": 0.4401,
+      "step": 2961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.23148033847112e-06,
+      "loss": 0.4386,
+      "step": 2962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.225345037740186e-06,
+      "loss": 0.4374,
+      "step": 2963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.219212064118079e-06,
+      "loss": 0.4439,
+      "step": 2964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.213081420593933e-06,
+      "loss": 0.4449,
+      "step": 2965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.2069531101557505e-06,
+      "loss": 0.4392,
+      "step": 2966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.200827135790396e-06,
+      "loss": 0.4505,
+      "step": 2967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.194703500483593e-06,
+      "loss": 0.4416,
+      "step": 2968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.188582207219931e-06,
+      "loss": 0.4376,
+      "step": 2969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1824632589828465e-06,
+      "loss": 0.4587,
+      "step": 2970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.176346658754648e-06,
+      "loss": 0.4481,
+      "step": 2971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.1702324095164955e-06,
+      "loss": 0.4379,
+      "step": 2972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.16412051424839e-06,
+      "loss": 0.4408,
+      "step": 2973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.158010975929193e-06,
+      "loss": 0.43,
+      "step": 2974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.151903797536631e-06,
+      "loss": 0.4387,
+      "step": 2975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.145798982047261e-06,
+      "loss": 0.4324,
+      "step": 2976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.139696532436499e-06,
+      "loss": 0.4246,
+      "step": 2977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.133596451678603e-06,
+      "loss": 0.4395,
+      "step": 2978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.127498742746675e-06,
+      "loss": 0.4375,
+      "step": 2979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.121403408612672e-06,
+      "loss": 0.4444,
+      "step": 2980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.115310452247386e-06,
+      "loss": 0.4386,
+      "step": 2981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.109219876620441e-06,
+      "loss": 0.4432,
+      "step": 2982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.103131684700315e-06,
+      "loss": 0.4372,
+      "step": 2983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0970458794543135e-06,
+      "loss": 0.4364,
+      "step": 2984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.090962463848592e-06,
+      "loss": 0.445,
+      "step": 2985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0848814408481305e-06,
+      "loss": 0.4381,
+      "step": 2986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.078802813416746e-06,
+      "loss": 0.4367,
+      "step": 2987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.072726584517086e-06,
+      "loss": 0.4477,
+      "step": 2988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.066652757110628e-06,
+      "loss": 0.4385,
+      "step": 2989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.060581334157693e-06,
+      "loss": 0.4336,
+      "step": 2990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.054512318617406e-06,
+      "loss": 0.4361,
+      "step": 2991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.048445713447738e-06,
+      "loss": 0.4432,
+      "step": 2992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.042381521605473e-06,
+      "loss": 0.4428,
+      "step": 2993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.036319746046232e-06,
+      "loss": 0.4331,
+      "step": 2994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.030260389724447e-06,
+      "loss": 0.4375,
+      "step": 2995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.024203455593375e-06,
+      "loss": 0.4477,
+      "step": 2996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.018148946605092e-06,
+      "loss": 0.4439,
+      "step": 2997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.012096865710494e-06,
+      "loss": 0.4401,
+      "step": 2998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.0060472158592885e-06,
+      "loss": 0.4443,
+      "step": 2999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.4447,
+      "step": 3000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.993955221079976e-06,
+      "loss": 0.4364,
+      "step": 3001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98791288204536e-06,
+      "loss": 0.439,
+      "step": 3002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.981872985841115e-06,
+      "loss": 0.442,
+      "step": 3003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.97583553541102e-06,
+      "loss": 0.4396,
+      "step": 3004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.96980053369765e-06,
+      "loss": 0.4339,
+      "step": 3005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9637679836423926e-06,
+      "loss": 0.4406,
+      "step": 3006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.957737888185439e-06,
+      "loss": 0.4424,
+      "step": 3007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.951710250265785e-06,
+      "loss": 0.4337,
+      "step": 3008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.945685072821227e-06,
+      "loss": 0.4451,
+      "step": 3009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.939662358788364e-06,
+      "loss": 0.4499,
+      "step": 3010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.933642111102595e-06,
+      "loss": 0.4394,
+      "step": 3011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.927624332698109e-06,
+      "loss": 0.4529,
+      "step": 3012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.921609026507907e-06,
+      "loss": 0.4447,
+      "step": 3013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.915596195463773e-06,
+      "loss": 0.4433,
+      "step": 3014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.909585842496287e-06,
+      "loss": 0.4376,
+      "step": 3015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.903577970534823e-06,
+      "loss": 0.4417,
+      "step": 3016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.897572582507544e-06,
+      "loss": 0.4391,
+      "step": 3017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.891569681341403e-06,
+      "loss": 0.4426,
+      "step": 3018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.885569269962142e-06,
+      "loss": 0.4372,
+      "step": 3019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.879571351294287e-06,
+      "loss": 0.4362,
+      "step": 3020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.873575928261151e-06,
+      "loss": 0.4355,
+      "step": 3021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8675830037848295e-06,
+      "loss": 0.4387,
+      "step": 3022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.861592580786205e-06,
+      "loss": 0.4352,
+      "step": 3023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.855604662184935e-06,
+      "loss": 0.4421,
+      "step": 3024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849619250899458e-06,
+      "loss": 0.4442,
+      "step": 3025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.843636349846991e-06,
+      "loss": 0.4467,
+      "step": 3026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.837655961943526e-06,
+      "loss": 0.4376,
+      "step": 3027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.831678090103832e-06,
+      "loss": 0.4358,
+      "step": 3028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.825702737241452e-06,
+      "loss": 0.4366,
+      "step": 3029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8197299062687e-06,
+      "loss": 0.4402,
+      "step": 3030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.813759600096661e-06,
+      "loss": 0.4437,
+      "step": 3031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.807791821635186e-06,
+      "loss": 0.4399,
+      "step": 3032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.801826573792905e-06,
+      "loss": 0.4412,
+      "step": 3033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.795863859477207e-06,
+      "loss": 0.4315,
+      "step": 3034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.78990368159424e-06,
+      "loss": 0.4437,
+      "step": 3035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.783946043048922e-06,
+      "loss": 0.4382,
+      "step": 3036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7779909467449416e-06,
+      "loss": 0.4394,
+      "step": 3037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.772038395584735e-06,
+      "loss": 0.4446,
+      "step": 3038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7660883924695055e-06,
+      "loss": 0.4442,
+      "step": 3039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.76014094029921e-06,
+      "loss": 0.4298,
+      "step": 3040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.754196041972563e-06,
+      "loss": 0.439,
+      "step": 3041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7482537003870425e-06,
+      "loss": 0.4416,
+      "step": 3042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7423139184388725e-06,
+      "loss": 0.438,
+      "step": 3043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.736376699023023e-06,
+      "loss": 0.443,
+      "step": 3044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7304420450332244e-06,
+      "loss": 0.4385,
+      "step": 3045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.724509959361961e-06,
+      "loss": 0.4395,
+      "step": 3046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.718580444900457e-06,
+      "loss": 0.4471,
+      "step": 3047
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.712653504538684e-06,
+      "loss": 0.4531,
+      "step": 3048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.706729141165362e-06,
+      "loss": 0.4439,
+      "step": 3049
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.700807357667953e-06,
+      "loss": 0.449,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.694888156932657e-06,
+      "loss": 0.4333,
+      "step": 3051
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.688971541844436e-06,
+      "loss": 0.4347,
+      "step": 3052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6830575152869615e-06,
+      "loss": 0.437,
+      "step": 3053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.677146080142664e-06,
+      "loss": 0.4389,
+      "step": 3054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.671237239292699e-06,
+      "loss": 0.4443,
+      "step": 3055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6653309956169745e-06,
+      "loss": 0.4361,
+      "step": 3056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.659427351994116e-06,
+      "loss": 0.4363,
+      "step": 3057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6535263113014885e-06,
+      "loss": 0.4504,
+      "step": 3058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.647627876415186e-06,
+      "loss": 0.4435,
+      "step": 3059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 0.4402,
+      "step": 3060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.635838835559591e-06,
+      "loss": 0.4386,
+      "step": 3061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.629948235336133e-06,
+      "loss": 0.4356,
+      "step": 3062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.62406025241067e-06,
+      "loss": 0.4465,
+      "step": 3063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.618174889652928e-06,
+      "loss": 0.4318,
+      "step": 3064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.612292149931369e-06,
+      "loss": 0.4351,
+      "step": 3065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.606412036113166e-06,
+      "loss": 0.438,
+      "step": 3066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.600534551064215e-06,
+      "loss": 0.4374,
+      "step": 3067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.59465969764913e-06,
+      "loss": 0.4375,
+      "step": 3068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.588787478731242e-06,
+      "loss": 0.4458,
+      "step": 3069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.582917897172603e-06,
+      "loss": 0.4359,
+      "step": 3070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.577050955833972e-06,
+      "loss": 0.4388,
+      "step": 3071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.571186657574828e-06,
+      "loss": 0.4379,
+      "step": 3072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.565325005253356e-06,
+      "loss": 0.4445,
+      "step": 3073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.559466001726451e-06,
+      "loss": 0.4344,
+      "step": 3074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5536096498497295e-06,
+      "loss": 0.4365,
+      "step": 3075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5477559524775e-06,
+      "loss": 0.4464,
+      "step": 3076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.541904912462785e-06,
+      "loss": 0.4455,
+      "step": 3077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.53605653265731e-06,
+      "loss": 0.4339,
+      "step": 3078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.530210815911504e-06,
+      "loss": 0.4445,
+      "step": 3079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.524367765074499e-06,
+      "loss": 0.4446,
+      "step": 3080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.518527382994127e-06,
+      "loss": 0.442,
+      "step": 3081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.512689672516918e-06,
+      "loss": 0.4416,
+      "step": 3082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.506854636488103e-06,
+      "loss": 0.4388,
+      "step": 3083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.501022277751602e-06,
+      "loss": 0.4369,
+      "step": 3084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.495192599150045e-06,
+      "loss": 0.4422,
+      "step": 3085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.48936560352474e-06,
+      "loss": 0.4424,
+      "step": 3086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.483541293715699e-06,
+      "loss": 0.4348,
+      "step": 3087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477719672561615e-06,
+      "loss": 0.434,
+      "step": 3088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471900742899876e-06,
+      "loss": 0.4454,
+      "step": 3089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46608450756656e-06,
+      "loss": 0.4404,
+      "step": 3090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4602709693964296e-06,
+      "loss": 0.4301,
+      "step": 3091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4544601312229295e-06,
+      "loss": 0.4393,
+      "step": 3092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.44865199587819e-06,
+      "loss": 0.4384,
+      "step": 3093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.442846566193034e-06,
+      "loss": 0.4436,
+      "step": 3094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.437043844996952e-06,
+      "loss": 0.4338,
+      "step": 3095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4312438351181246e-06,
+      "loss": 0.4376,
+      "step": 3096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.425446539383394e-06,
+      "loss": 0.4347,
+      "step": 3097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.419651960618302e-06,
+      "loss": 0.4323,
+      "step": 3098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.413860101647055e-06,
+      "loss": 0.4311,
+      "step": 3099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.408070965292534e-06,
+      "loss": 0.4335,
+      "step": 3100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.402284554376292e-06,
+      "loss": 0.4411,
+      "step": 3101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3965008717185555e-06,
+      "loss": 0.4488,
+      "step": 3102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.39071992013822e-06,
+      "loss": 0.4345,
+      "step": 3103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.384941702452856e-06,
+      "loss": 0.4384,
+      "step": 3104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.379166221478697e-06,
+      "loss": 0.4429,
+      "step": 3105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.373393480030637e-06,
+      "loss": 0.432,
+      "step": 3106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.367623480922236e-06,
+      "loss": 0.444,
+      "step": 3107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.361856226965733e-06,
+      "loss": 0.4421,
+      "step": 3108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.356091720972011e-06,
+      "loss": 0.4289,
+      "step": 3109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.350329965750622e-06,
+      "loss": 0.4346,
+      "step": 3110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.344570964109775e-06,
+      "loss": 0.4368,
+      "step": 3111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.338814718856333e-06,
+      "loss": 0.4272,
+      "step": 3112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3330612327958265e-06,
+      "loss": 0.4418,
+      "step": 3113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3273105087324375e-06,
+      "loss": 0.4352,
+      "step": 3114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.321562549468991e-06,
+      "loss": 0.4428,
+      "step": 3115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.315817357806974e-06,
+      "loss": 0.4368,
+      "step": 3116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.310074936546521e-06,
+      "loss": 0.4419,
+      "step": 3117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.304335288486426e-06,
+      "loss": 0.4447,
+      "step": 3118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.29859841642412e-06,
+      "loss": 0.4425,
+      "step": 3119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 0.4363,
+      "step": 3120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.287133011475847e-06,
+      "loss": 0.4341,
+      "step": 3121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.281404484177974e-06,
+      "loss": 0.4414,
+      "step": 3122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.275678744054094e-06,
+      "loss": 0.4347,
+      "step": 3123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.26995579389485e-06,
+      "loss": 0.4339,
+      "step": 3124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.264235636489542e-06,
+      "loss": 0.44,
+      "step": 3125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.258518274626103e-06,
+      "loss": 0.4251,
+      "step": 3126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2528037110911126e-06,
+      "loss": 0.4405,
+      "step": 3127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.247091948669775e-06,
+      "loss": 0.4293,
+      "step": 3128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.2413829901459345e-06,
+      "loss": 0.438,
+      "step": 3129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.235676838302069e-06,
+      "loss": 0.4345,
+      "step": 3130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.229973495919286e-06,
+      "loss": 0.4372,
+      "step": 3131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.224272965777326e-06,
+      "loss": 0.4341,
+      "step": 3132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.218575250654559e-06,
+      "loss": 0.4463,
+      "step": 3133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.21288035332798e-06,
+      "loss": 0.4273,
+      "step": 3134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.207188276573214e-06,
+      "loss": 0.4395,
+      "step": 3135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.201499023164508e-06,
+      "loss": 0.4294,
+      "step": 3136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.19581259587474e-06,
+      "loss": 0.4409,
+      "step": 3137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.190128997475402e-06,
+      "loss": 0.4307,
+      "step": 3138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.184448230736613e-06,
+      "loss": 0.4307,
+      "step": 3139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.178770298427107e-06,
+      "loss": 0.433,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.173095203314241e-06,
+      "loss": 0.4345,
+      "step": 3141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.167422948163986e-06,
+      "loss": 0.4366,
+      "step": 3142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.161753535740932e-06,
+      "loss": 0.4427,
+      "step": 3143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.15608696880828e-06,
+      "loss": 0.4412,
+      "step": 3144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.150423250127846e-06,
+      "loss": 0.4427,
+      "step": 3145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.144762382460059e-06,
+      "loss": 0.436,
+      "step": 3146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.1391043685639576e-06,
+      "loss": 0.4351,
+      "step": 3147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.133449211197188e-06,
+      "loss": 0.4379,
+      "step": 3148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.127796913116004e-06,
+      "loss": 0.4363,
+      "step": 3149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.4343,
+      "step": 3150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1165009058284496e-06,
+      "loss": 0.4362,
+      "step": 3151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.110857202127615e-06,
+      "loss": 0.435,
+      "step": 3152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.105216368723437e-06,
+      "loss": 0.4397,
+      "step": 3153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.099578408365192e-06,
+      "loss": 0.4294,
+      "step": 3154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.093943323800746e-06,
+      "loss": 0.4408,
+      "step": 3155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.08831111777658e-06,
+      "loss": 0.4312,
+      "step": 3156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.08268179303776e-06,
+      "loss": 0.4329,
+      "step": 3157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0770553523279535e-06,
+      "loss": 0.4411,
+      "step": 3158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.071431798389408e-06,
+      "loss": 0.4279,
+      "step": 3159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.065811133962987e-06,
+      "loss": 0.432,
+      "step": 3160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.06019336178813e-06,
+      "loss": 0.4498,
+      "step": 3161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.05457848460287e-06,
+      "loss": 0.4516,
+      "step": 3162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.048966505143831e-06,
+      "loss": 0.4285,
+      "step": 3163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.04335742614622e-06,
+      "loss": 0.429,
+      "step": 3164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.037751250343841e-06,
+      "loss": 0.4332,
+      "step": 3165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.032147980469072e-06,
+      "loss": 0.4274,
+      "step": 3166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.026547619252883e-06,
+      "loss": 0.4323,
+      "step": 3167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.020950169424815e-06,
+      "loss": 0.4345,
+      "step": 3168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.015355633712996e-06,
+      "loss": 0.4354,
+      "step": 3169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.009764014844143e-06,
+      "loss": 0.4359,
+      "step": 3170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.004175315543538e-06,
+      "loss": 0.4403,
+      "step": 3171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998589538535046e-06,
+      "loss": 0.4313,
+      "step": 3172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.993006686541108e-06,
+      "loss": 0.4257,
+      "step": 3173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.987426762282733e-06,
+      "loss": 0.4428,
+      "step": 3174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.981849768479516e-06,
+      "loss": 0.4365,
+      "step": 3175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.976275707849616e-06,
+      "loss": 0.4424,
+      "step": 3176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.970704583109755e-06,
+      "loss": 0.4397,
+      "step": 3177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.965136396975235e-06,
+      "loss": 0.4459,
+      "step": 3178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.959571152159922e-06,
+      "loss": 0.4361,
+      "step": 3179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.954008851376252e-06,
+      "loss": 0.4352,
+      "step": 3180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.94844949733522e-06,
+      "loss": 0.4334,
+      "step": 3181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.942893092746387e-06,
+      "loss": 0.4423,
+      "step": 3182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.937339640317879e-06,
+      "loss": 0.4264,
+      "step": 3183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.931789142756377e-06,
+      "loss": 0.4367,
+      "step": 3184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9262416027671354e-06,
+      "loss": 0.4388,
+      "step": 3185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.920697023053949e-06,
+      "loss": 0.4385,
+      "step": 3186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.915155406319181e-06,
+      "loss": 0.4342,
+      "step": 3187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9096167552637454e-06,
+      "loss": 0.4386,
+      "step": 3188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.90408107258712e-06,
+      "loss": 0.4367,
+      "step": 3189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.898548360987325e-06,
+      "loss": 0.4303,
+      "step": 3190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.893018623160938e-06,
+      "loss": 0.4321,
+      "step": 3191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.887491861803085e-06,
+      "loss": 0.4344,
+      "step": 3192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88196807960744e-06,
+      "loss": 0.4381,
+      "step": 3193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.876447279266238e-06,
+      "loss": 0.4264,
+      "step": 3194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8709294634702374e-06,
+      "loss": 0.4434,
+      "step": 3195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.86541463490876e-06,
+      "loss": 0.438,
+      "step": 3196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.859902796269664e-06,
+      "loss": 0.435,
+      "step": 3197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.854393950239356e-06,
+      "loss": 0.429,
+      "step": 3198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.848888099502779e-06,
+      "loss": 0.4291,
+      "step": 3199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 0.4331,
+      "step": 3200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8378853946432956e-06,
+      "loss": 0.4337,
+      "step": 3201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.832388545882975e-06,
+      "loss": 0.4347,
+      "step": 3202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.826894703141552e-06,
+      "loss": 0.4271,
+      "step": 3203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.821403869096658e-06,
+      "loss": 0.4362,
+      "step": 3204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.81591604642446e-06,
+      "loss": 0.4382,
+      "step": 3205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8104312377996576e-06,
+      "loss": 0.4275,
+      "step": 3206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.804949445895473e-06,
+      "loss": 0.4307,
+      "step": 3207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7994706733836738e-06,
+      "loss": 0.433,
+      "step": 3208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.793994922934544e-06,
+      "loss": 0.4289,
+      "step": 3209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7885221972168974e-06,
+      "loss": 0.4282,
+      "step": 3210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.783052498898073e-06,
+      "loss": 0.4343,
+      "step": 3211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7775858306439374e-06,
+      "loss": 0.4325,
+      "step": 3212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.772122195118877e-06,
+      "loss": 0.4316,
+      "step": 3213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.766661594985801e-06,
+      "loss": 0.4331,
+      "step": 3214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7612040329061405e-06,
+      "loss": 0.4373,
+      "step": 3215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7557495115398446e-06,
+      "loss": 0.4339,
+      "step": 3216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7502980335453777e-06,
+      "loss": 0.4345,
+      "step": 3217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7448496015797296e-06,
+      "loss": 0.4264,
+      "step": 3218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7394042182983983e-06,
+      "loss": 0.4308,
+      "step": 3219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7339618863553983e-06,
+      "loss": 0.4311,
+      "step": 3220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.728522608403249e-06,
+      "loss": 0.4263,
+      "step": 3221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.723086387092997e-06,
+      "loss": 0.4349,
+      "step": 3222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7176532250741857e-06,
+      "loss": 0.4328,
+      "step": 3223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7122231249948747e-06,
+      "loss": 0.442,
+      "step": 3224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7067960895016277e-06,
+      "loss": 0.4261,
+      "step": 3225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7013721212395128e-06,
+      "loss": 0.4354,
+      "step": 3226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6959512228521123e-06,
+      "loss": 0.4362,
+      "step": 3227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6905333969815038e-06,
+      "loss": 0.4305,
+      "step": 3228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.685118646268272e-06,
+      "loss": 0.4347,
+      "step": 3229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.679706973351491e-06,
+      "loss": 0.4315,
+      "step": 3230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.674298380868756e-06,
+      "loss": 0.4405,
+      "step": 3231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6688928714561444e-06,
+      "loss": 0.4268,
+      "step": 3232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.663490447748236e-06,
+      "loss": 0.4356,
+      "step": 3233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.658091112378106e-06,
+      "loss": 0.4287,
+      "step": 3234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6526948679773256e-06,
+      "loss": 0.4278,
+      "step": 3235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6473017171759563e-06,
+      "loss": 0.4269,
+      "step": 3236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6419116626025585e-06,
+      "loss": 0.4309,
+      "step": 3237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.636524706884181e-06,
+      "loss": 0.4355,
+      "step": 3238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6311408526463554e-06,
+      "loss": 0.4303,
+      "step": 3239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 0.4269,
+      "step": 3240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.620382459106946e-06,
+      "loss": 0.4358,
+      "step": 3241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.615007925048878e-06,
+      "loss": 0.432,
+      "step": 3242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6096365029583803e-06,
+      "loss": 0.4414,
+      "step": 3243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.604268195453421e-06,
+      "loss": 0.4464,
+      "step": 3244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.598903005150444e-06,
+      "loss": 0.4461,
+      "step": 3245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.5935409346643835e-06,
+      "loss": 0.4422,
+      "step": 3246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.5881819866086485e-06,
+      "loss": 0.4352,
+      "step": 3247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.582826163595119e-06,
+      "loss": 0.4352,
+      "step": 3248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5774734682341563e-06,
+      "loss": 0.4344,
+      "step": 3249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.4283,
+      "step": 3250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5667774709037804e-06,
+      "loss": 0.4344,
+      "step": 3251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5614341741474633e-06,
+      "loss": 0.4417,
+      "step": 3252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5560940154699133e-06,
+      "loss": 0.4295,
+      "step": 3253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5507569974738575e-06,
+      "loss": 0.4316,
+      "step": 3254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.545423122760493e-06,
+      "loss": 0.4302,
+      "step": 3255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.540092393929494e-06,
+      "loss": 0.4331,
+      "step": 3256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5347648135789823e-06,
+      "loss": 0.4378,
+      "step": 3257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5294403843055604e-06,
+      "loss": 0.4371,
+      "step": 3258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.524119108704286e-06,
+      "loss": 0.4239,
+      "step": 3259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5188009893686916e-06,
+      "loss": 0.4351,
+      "step": 3260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.5134860288907602e-06,
+      "loss": 0.4341,
+      "step": 3261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.50817422986094e-06,
+      "loss": 0.4318,
+      "step": 3262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.502865594868136e-06,
+      "loss": 0.4342,
+      "step": 3263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4975601264997094e-06,
+      "loss": 0.4394,
+      "step": 3264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.492257827341492e-06,
+      "loss": 0.4365,
+      "step": 3265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4869586999777492e-06,
+      "loss": 0.4298,
+      "step": 3266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4816627469912147e-06,
+      "loss": 0.4278,
+      "step": 3267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.476369970963072e-06,
+      "loss": 0.4376,
+      "step": 3268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4710803744729517e-06,
+      "loss": 0.4243,
+      "step": 3269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4657939600989453e-06,
+      "loss": 0.4317,
+      "step": 3270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4605107304175855e-06,
+      "loss": 0.4349,
+      "step": 3271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.455230688003852e-06,
+      "loss": 0.4361,
+      "step": 3272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4499538354311757e-06,
+      "loss": 0.432,
+      "step": 3273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4446801752714287e-06,
+      "loss": 0.4402,
+      "step": 3274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4394097100949286e-06,
+      "loss": 0.4329,
+      "step": 3275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4341424424704373e-06,
+      "loss": 0.4313,
+      "step": 3276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4288783749651568e-06,
+      "loss": 0.4426,
+      "step": 3277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4236175101447265e-06,
+      "loss": 0.4374,
+      "step": 3278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.418359850573234e-06,
+      "loss": 0.4287,
+      "step": 3279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4131053988131947e-06,
+      "loss": 0.4367,
+      "step": 3280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4078541574255664e-06,
+      "loss": 0.4264,
+      "step": 3281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.4026061289697397e-06,
+      "loss": 0.4296,
+      "step": 3282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.397361316003539e-06,
+      "loss": 0.4332,
+      "step": 3283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3921197210832235e-06,
+      "loss": 0.4286,
+      "step": 3284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3868813467634833e-06,
+      "loss": 0.4269,
+      "step": 3285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.381646195597437e-06,
+      "loss": 0.428,
+      "step": 3286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.376414270136633e-06,
+      "loss": 0.436,
+      "step": 3287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3711855729310482e-06,
+      "loss": 0.4358,
+      "step": 3288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3659601065290893e-06,
+      "loss": 0.4342,
+      "step": 3289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.360737873477584e-06,
+      "loss": 0.4358,
+      "step": 3290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.355518876321787e-06,
+      "loss": 0.427,
+      "step": 3291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3503031176053657e-06,
+      "loss": 0.4289,
+      "step": 3292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3450905998704274e-06,
+      "loss": 0.4315,
+      "step": 3293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3398813256574847e-06,
+      "loss": 0.4341,
+      "step": 3294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3346752975054763e-06,
+      "loss": 0.4387,
+      "step": 3295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3294725179517573e-06,
+      "loss": 0.4364,
+      "step": 3296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3242729895320945e-06,
+      "loss": 0.4371,
+      "step": 3297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3190767147806825e-06,
+      "loss": 0.4217,
+      "step": 3298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3138836962301192e-06,
+      "loss": 0.4371,
+      "step": 3299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.4323,
+      "step": 3300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3035074378540087e-06,
+      "loss": 0.4257,
+      "step": 3301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2983242030857177e-06,
+      "loss": 0.4289,
+      "step": 3302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2931442346328e-06,
+      "loss": 0.445,
+      "step": 3303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.287967535019908e-06,
+      "loss": 0.4271,
+      "step": 3304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2827941067700996e-06,
+      "loss": 0.4242,
+      "step": 3305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2776239524048426e-06,
+      "loss": 0.4382,
+      "step": 3306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.272457074444003e-06,
+      "loss": 0.4282,
+      "step": 3307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2672934754058615e-06,
+      "loss": 0.4349,
+      "step": 3308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2621331578070936e-06,
+      "loss": 0.4349,
+      "step": 3309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2569761241627694e-06,
+      "loss": 0.4283,
+      "step": 3310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2518223769863633e-06,
+      "loss": 0.4274,
+      "step": 3311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2466719187897555e-06,
+      "loss": 0.4361,
+      "step": 3312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.241524752083215e-06,
+      "loss": 0.4311,
+      "step": 3313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2363808793754082e-06,
+      "loss": 0.4336,
+      "step": 3314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2312403031733943e-06,
+      "loss": 0.4306,
+      "step": 3315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2261030259826287e-06,
+      "loss": 0.4376,
+      "step": 3316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2209690503069545e-06,
+      "loss": 0.4273,
+      "step": 3317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2158383786486204e-06,
+      "loss": 0.4329,
+      "step": 3318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.210711013508242e-06,
+      "loss": 0.4416,
+      "step": 3319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2055869573848374e-06,
+      "loss": 0.4332,
+      "step": 3320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.200466212775808e-06,
+      "loss": 0.4414,
+      "step": 3321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.195348782176948e-06,
+      "loss": 0.4398,
+      "step": 3322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.190234668082427e-06,
+      "loss": 0.4349,
+      "step": 3323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1851238729848033e-06,
+      "loss": 0.434,
+      "step": 3324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1800163993750166e-06,
+      "loss": 0.429,
+      "step": 3325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.174912249742382e-06,
+      "loss": 0.4326,
+      "step": 3326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1698114265746126e-06,
+      "loss": 0.4228,
+      "step": 3327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.164713932357776e-06,
+      "loss": 0.4389,
+      "step": 3328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.159619769576333e-06,
+      "loss": 0.429,
+      "step": 3329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1545289407131128e-06,
+      "loss": 0.4254,
+      "step": 3330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.149441448249331e-06,
+      "loss": 0.438,
+      "step": 3331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.144357294664565e-06,
+      "loss": 0.4319,
+      "step": 3332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1392764824367706e-06,
+      "loss": 0.4325,
+      "step": 3333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.134199014042274e-06,
+      "loss": 0.4328,
+      "step": 3334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.1291248919557717e-06,
+      "loss": 0.426,
+      "step": 3335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.124054118650327e-06,
+      "loss": 0.4362,
+      "step": 3336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.118986696597377e-06,
+      "loss": 0.4264,
+      "step": 3337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.113922628266718e-06,
+      "loss": 0.4312,
+      "step": 3338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.108861916126518e-06,
+      "loss": 0.4282,
+      "step": 3339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.103804562643302e-06,
+      "loss": 0.4317,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0987505702819687e-06,
+      "loss": 0.4445,
+      "step": 3341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0936999415057712e-06,
+      "loss": 0.4308,
+      "step": 3342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0886526787763237e-06,
+      "loss": 0.4353,
+      "step": 3343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0836087845536e-06,
+      "loss": 0.4295,
+      "step": 3344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0785682612959334e-06,
+      "loss": 0.4381,
+      "step": 3345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.073531111460013e-06,
+      "loss": 0.4354,
+      "step": 3346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0684973375008865e-06,
+      "loss": 0.4321,
+      "step": 3347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.063466941871952e-06,
+      "loss": 0.4187,
+      "step": 3348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.058439927024962e-06,
+      "loss": 0.4292,
+      "step": 3349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0534162954100264e-06,
+      "loss": 0.4331,
+      "step": 3350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0483960494756017e-06,
+      "loss": 0.432,
+      "step": 3351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.043379191668492e-06,
+      "loss": 0.4366,
+      "step": 3352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.038365724433858e-06,
+      "loss": 0.4242,
+      "step": 3353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.033355650215193e-06,
+      "loss": 0.4306,
+      "step": 3354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.028348971454356e-06,
+      "loss": 0.4301,
+      "step": 3355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.023345690591537e-06,
+      "loss": 0.4304,
+      "step": 3356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0183458100652752e-06,
+      "loss": 0.4423,
+      "step": 3357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.013349332312451e-06,
+      "loss": 0.4377,
+      "step": 3358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.008356259768285e-06,
+      "loss": 0.425,
+      "step": 3359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 0.4288,
+      "step": 3360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9983803400385313e-06,
+      "loss": 0.4309,
+      "step": 3361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.993397497715086e-06,
+      "loss": 0.432,
+      "step": 3362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.988418070324577e-06,
+      "loss": 0.4211,
+      "step": 3363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.983442060293926e-06,
+      "loss": 0.4257,
+      "step": 3364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.978469470048376e-06,
+      "loss": 0.428,
+      "step": 3365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9735003020115095e-06,
+      "loss": 0.4344,
+      "step": 3366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.968534558605236e-06,
+      "loss": 0.4342,
+      "step": 3367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.963572242249799e-06,
+      "loss": 0.4304,
+      "step": 3368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9586133553637687e-06,
+      "loss": 0.4228,
+      "step": 3369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.953657900364053e-06,
+      "loss": 0.4281,
+      "step": 3370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9487058796658785e-06,
+      "loss": 0.4404,
+      "step": 3371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9437572956827965e-06,
+      "loss": 0.4283,
+      "step": 3372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.938812150826684e-06,
+      "loss": 0.4276,
+      "step": 3373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9338704475077527e-06,
+      "loss": 0.4337,
+      "step": 3374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.4303,
+      "step": 3375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9239973751138495e-06,
+      "loss": 0.4311,
+      "step": 3376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.919066010850892e-06,
+      "loss": 0.4276,
+      "step": 3377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9141380977491373e-06,
+      "loss": 0.4397,
+      "step": 3378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9092136382103976e-06,
+      "loss": 0.4179,
+      "step": 3379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9042926346347932e-06,
+      "loss": 0.4327,
+      "step": 3380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8993750894207563e-06,
+      "loss": 0.4258,
+      "step": 3381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8944610049650377e-06,
+      "loss": 0.4284,
+      "step": 3382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8895503836627105e-06,
+      "loss": 0.4306,
+      "step": 3383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.884643227907147e-06,
+      "loss": 0.4265,
+      "step": 3384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8797395400900362e-06,
+      "loss": 0.4304,
+      "step": 3385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.874839322601375e-06,
+      "loss": 0.4256,
+      "step": 3386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.869942577829471e-06,
+      "loss": 0.4245,
+      "step": 3387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8650493081609344e-06,
+      "loss": 0.4406,
+      "step": 3388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.860159515980695e-06,
+      "loss": 0.4387,
+      "step": 3389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.855273203671969e-06,
+      "loss": 0.4321,
+      "step": 3390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8503903736162876e-06,
+      "loss": 0.4304,
+      "step": 3391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8455110281934804e-06,
+      "loss": 0.4234,
+      "step": 3392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.840635169781688e-06,
+      "loss": 0.4257,
+      "step": 3393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8357628007573412e-06,
+      "loss": 0.4306,
+      "step": 3394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.830893923495173e-06,
+      "loss": 0.4355,
+      "step": 3395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8260285403682153e-06,
+      "loss": 0.4342,
+      "step": 3396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.821166653747793e-06,
+      "loss": 0.4315,
+      "step": 3397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.816308266003541e-06,
+      "loss": 0.4321,
+      "step": 3398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8114533795033685e-06,
+      "loss": 0.4234,
+      "step": 3399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8066019966134907e-06,
+      "loss": 0.4287,
+      "step": 3400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8017541196984144e-06,
+      "loss": 0.4264,
+      "step": 3401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.796909751120931e-06,
+      "loss": 0.4304,
+      "step": 3402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7920688932421337e-06,
+      "loss": 0.4323,
+      "step": 3403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7872315484213954e-06,
+      "loss": 0.4215,
+      "step": 3404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7823977190163788e-06,
+      "loss": 0.4234,
+      "step": 3405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7775674073830337e-06,
+      "loss": 0.4344,
+      "step": 3406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7727406158755943e-06,
+      "loss": 0.4358,
+      "step": 3407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7679173468465813e-06,
+      "loss": 0.4349,
+      "step": 3408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.763097602646797e-06,
+      "loss": 0.4315,
+      "step": 3409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7582813856253276e-06,
+      "loss": 0.4396,
+      "step": 3410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7534686981295335e-06,
+      "loss": 0.4386,
+      "step": 3411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7486595425050667e-06,
+      "loss": 0.4362,
+      "step": 3412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7438539210958483e-06,
+      "loss": 0.4343,
+      "step": 3413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.739051836244081e-06,
+      "loss": 0.4196,
+      "step": 3414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7342532902902418e-06,
+      "loss": 0.4329,
+      "step": 3415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7294582855730835e-06,
+      "loss": 0.4319,
+      "step": 3416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7246668244296328e-06,
+      "loss": 0.4253,
+      "step": 3417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7198789091951903e-06,
+      "loss": 0.4319,
+      "step": 3418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.715094542203327e-06,
+      "loss": 0.4361,
+      "step": 3419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 0.4237,
+      "step": 3420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7055364622729772e-06,
+      "loss": 0.4293,
+      "step": 3421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.7007627539929847e-06,
+      "loss": 0.4323,
+      "step": 3422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6959926032725537e-06,
+      "loss": 0.4418,
+      "step": 3423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6912260124366007e-06,
+      "loss": 0.4246,
+      "step": 3424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6864629838082957e-06,
+      "loss": 0.4339,
+      "step": 3425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6817035197090892e-06,
+      "loss": 0.4251,
+      "step": 3426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.676947622458683e-06,
+      "loss": 0.4309,
+      "step": 3427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.672195294375045e-06,
+      "loss": 0.4304,
+      "step": 3428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.667446537774402e-06,
+      "loss": 0.4239,
+      "step": 3429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6627013549712355e-06,
+      "loss": 0.4276,
+      "step": 3430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6579597482782972e-06,
+      "loss": 0.4366,
+      "step": 3431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6532217200065856e-06,
+      "loss": 0.4368,
+      "step": 3432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.648487272465361e-06,
+      "loss": 0.423,
+      "step": 3433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.643756407962127e-06,
+      "loss": 0.4306,
+      "step": 3434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.639029128802657e-06,
+      "loss": 0.4342,
+      "step": 3435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.634305437290968e-06,
+      "loss": 0.4354,
+      "step": 3436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.62958533572933e-06,
+      "loss": 0.4399,
+      "step": 3437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.624868826418262e-06,
+      "loss": 0.428,
+      "step": 3438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6201559116565346e-06,
+      "loss": 0.4298,
+      "step": 3439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.615446593741161e-06,
+      "loss": 0.4238,
+      "step": 3440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6107408749674125e-06,
+      "loss": 0.4402,
+      "step": 3441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6060387576287983e-06,
+      "loss": 0.4355,
+      "step": 3442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6013402440170676e-06,
+      "loss": 0.4248,
+      "step": 3443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.596645336422219e-06,
+      "loss": 0.4337,
+      "step": 3444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5919540371325005e-06,
+      "loss": 0.4262,
+      "step": 3445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5872663484343887e-06,
+      "loss": 0.4313,
+      "step": 3446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5825822726126095e-06,
+      "loss": 0.435,
+      "step": 3447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.577901811950121e-06,
+      "loss": 0.4288,
+      "step": 3448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5732249687281228e-06,
+      "loss": 0.4336,
+      "step": 3449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 0.4174,
+      "step": 3450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5638821437215944e-06,
+      "loss": 0.4342,
+      "step": 3451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5592161664906366e-06,
+      "loss": 0.438,
+      "step": 3452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5545538158073278e-06,
+      "loss": 0.4238,
+      "step": 3453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.549895093944039e-06,
+      "loss": 0.4313,
+      "step": 3454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5452400031713786e-06,
+      "loss": 0.4353,
+      "step": 3455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5405885457581793e-06,
+      "loss": 0.4317,
+      "step": 3456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.535940723971505e-06,
+      "loss": 0.4296,
+      "step": 3457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5312965400766475e-06,
+      "loss": 0.4281,
+      "step": 3458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5266559963371216e-06,
+      "loss": 0.4336,
+      "step": 3459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.522019095014683e-06,
+      "loss": 0.4299,
+      "step": 3460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5173858383692906e-06,
+      "loss": 0.4296,
+      "step": 3461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.512756228659141e-06,
+      "loss": 0.4179,
+      "step": 3462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5081302681406463e-06,
+      "loss": 0.4261,
+      "step": 3463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5035079590684496e-06,
+      "loss": 0.425,
+      "step": 3464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4988893036954045e-06,
+      "loss": 0.4322,
+      "step": 3465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.494274304272589e-06,
+      "loss": 0.4245,
+      "step": 3466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4896629630492974e-06,
+      "loss": 0.4217,
+      "step": 3467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.48505528227304e-06,
+      "loss": 0.4244,
+      "step": 3468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.480451264189546e-06,
+      "loss": 0.4227,
+      "step": 3469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4758509110427576e-06,
+      "loss": 0.423,
+      "step": 3470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4712542250748305e-06,
+      "loss": 0.4279,
+      "step": 3471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4666612085261344e-06,
+      "loss": 0.4216,
+      "step": 3472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4620718636352457e-06,
+      "loss": 0.432,
+      "step": 3473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4574861926389615e-06,
+      "loss": 0.4308,
+      "step": 3474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.45290419777228e-06,
+      "loss": 0.4326,
+      "step": 3475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4483258812684096e-06,
+      "loss": 0.4311,
+      "step": 3476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4437512453587653e-06,
+      "loss": 0.4356,
+      "step": 3477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4391802922729703e-06,
+      "loss": 0.4267,
+      "step": 3478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.43461302423885e-06,
+      "loss": 0.4298,
+      "step": 3479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4300494434824373e-06,
+      "loss": 0.4311,
+      "step": 3480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4254895522279642e-06,
+      "loss": 0.4262,
+      "step": 3481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.420933352697865e-06,
+      "loss": 0.4234,
+      "step": 3482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4163808471127815e-06,
+      "loss": 0.4334,
+      "step": 3483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.411832037691545e-06,
+      "loss": 0.4241,
+      "step": 3484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.407286926651192e-06,
+      "loss": 0.4384,
+      "step": 3485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4027455162069567e-06,
+      "loss": 0.4232,
+      "step": 3486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.398207808572258e-06,
+      "loss": 0.4318,
+      "step": 3487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3936738059587284e-06,
+      "loss": 0.4212,
+      "step": 3488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3891435105761838e-06,
+      "loss": 0.4261,
+      "step": 3489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3846169246326345e-06,
+      "loss": 0.4249,
+      "step": 3490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.380094050334283e-06,
+      "loss": 0.4315,
+      "step": 3491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.37557488988552e-06,
+      "loss": 0.4314,
+      "step": 3492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.371059445488938e-06,
+      "loss": 0.4404,
+      "step": 3493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3665477193453037e-06,
+      "loss": 0.4299,
+      "step": 3494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.362039713653581e-06,
+      "loss": 0.4289,
+      "step": 3495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.35753543061091e-06,
+      "loss": 0.4272,
+      "step": 3496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3530348724126304e-06,
+      "loss": 0.4266,
+      "step": 3497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3485380412522586e-06,
+      "loss": 0.4381,
+      "step": 3498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3440449393214947e-06,
+      "loss": 0.4228,
+      "step": 3499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.4304,
+      "step": 3500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.335069931906503e-06,
+      "loss": 0.4319,
+      "step": 3501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3305880307965834e-06,
+      "loss": 0.4223,
+      "step": 3502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3261098676648908e-06,
+      "loss": 0.4367,
+      "step": 3503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.321635444694028e-06,
+      "loss": 0.4275,
+      "step": 3504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.317164764064769e-06,
+      "loss": 0.431,
+      "step": 3505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3126978279560687e-06,
+      "loss": 0.4268,
+      "step": 3506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.308234638545064e-06,
+      "loss": 0.4304,
+      "step": 3507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.3037751980070557e-06,
+      "loss": 0.4328,
+      "step": 3508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2993195085155205e-06,
+      "loss": 0.4349,
+      "step": 3509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2948675722421086e-06,
+      "loss": 0.4183,
+      "step": 3510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2904193913566363e-06,
+      "loss": 0.4332,
+      "step": 3511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2859749680270983e-06,
+      "loss": 0.4359,
+      "step": 3512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2815343044196523e-06,
+      "loss": 0.4304,
+      "step": 3513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.277097402698619e-06,
+      "loss": 0.4279,
+      "step": 3514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.27266426502649e-06,
+      "loss": 0.4339,
+      "step": 3515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2682348935639274e-06,
+      "loss": 0.4284,
+      "step": 3516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2638092904697516e-06,
+      "loss": 0.4299,
+      "step": 3517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.259387457900948e-06,
+      "loss": 0.4242,
+      "step": 3518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.254969398012663e-06,
+      "loss": 0.4186,
+      "step": 3519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2505551129582047e-06,
+      "loss": 0.4277,
+      "step": 3520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2461446048890424e-06,
+      "loss": 0.4221,
+      "step": 3521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.241737875954808e-06,
+      "loss": 0.4281,
+      "step": 3522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.237334928303283e-06,
+      "loss": 0.4264,
+      "step": 3523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2329357640804118e-06,
+      "loss": 0.4194,
+      "step": 3524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.2285403854302912e-06,
+      "loss": 0.4428,
+      "step": 3525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.22414879449518e-06,
+      "loss": 0.4207,
+      "step": 3526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.219760993415485e-06,
+      "loss": 0.4313,
+      "step": 3527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.215376984329767e-06,
+      "loss": 0.4297,
+      "step": 3528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.210996769374737e-06,
+      "loss": 0.427,
+      "step": 3529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.206620350685257e-06,
+      "loss": 0.4358,
+      "step": 3530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.202247730394349e-06,
+      "loss": 0.4257,
+      "step": 3531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.1978789106331666e-06,
+      "loss": 0.4243,
+      "step": 3532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.1935138935310208e-06,
+      "loss": 0.4306,
+      "step": 3533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.1891526812153674e-06,
+      "loss": 0.4316,
+      "step": 3534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.1847952758118118e-06,
+      "loss": 0.4328,
+      "step": 3535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.1804416794441e-06,
+      "loss": 0.4328,
+      "step": 3536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.1760918942341193e-06,
+      "loss": 0.4239,
+      "step": 3537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.171745922301903e-06,
+      "loss": 0.4274,
+      "step": 3538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1674037657656265e-06,
+      "loss": 0.4192,
+      "step": 3539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.163065426741603e-06,
+      "loss": 0.4184,
+      "step": 3540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1587309073442865e-06,
+      "loss": 0.4272,
+      "step": 3541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.154400209686268e-06,
+      "loss": 0.4334,
+      "step": 3542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1500733358782786e-06,
+      "loss": 0.434,
+      "step": 3543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1457502880291815e-06,
+      "loss": 0.4288,
+      "step": 3544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1414310682459805e-06,
+      "loss": 0.4293,
+      "step": 3545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1371156786338108e-06,
+      "loss": 0.4332,
+      "step": 3546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1328041212959403e-06,
+      "loss": 0.4244,
+      "step": 3547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.128496398333768e-06,
+      "loss": 0.4291,
+      "step": 3548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1241925118468288e-06,
+      "loss": 0.4396,
+      "step": 3549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.119892463932781e-06,
+      "loss": 0.4267,
+      "step": 3550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.115596256687419e-06,
+      "loss": 0.4259,
+      "step": 3551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.1113038922046603e-06,
+      "loss": 0.4267,
+      "step": 3552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.107015372576552e-06,
+      "loss": 0.4299,
+      "step": 3553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.102730699893263e-06,
+      "loss": 0.422,
+      "step": 3554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.098449876243096e-06,
+      "loss": 0.4279,
+      "step": 3555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.09417290371247e-06,
+      "loss": 0.4309,
+      "step": 3556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0898997843859338e-06,
+      "loss": 0.4303,
+      "step": 3557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0856305203461436e-06,
+      "loss": 0.4416,
+      "step": 3558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0813651136738957e-06,
+      "loss": 0.4242,
+      "step": 3559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0771035664480944e-06,
+      "loss": 0.4317,
+      "step": 3560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.072845880745766e-06,
+      "loss": 0.4188,
+      "step": 3561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0685920586420562e-06,
+      "loss": 0.4292,
+      "step": 3562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0643421022102216e-06,
+      "loss": 0.4249,
+      "step": 3563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0600960135216463e-06,
+      "loss": 0.4268,
+      "step": 3564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0558537946458177e-06,
+      "loss": 0.4199,
+      "step": 3565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.051615447650347e-06,
+      "loss": 0.4268,
+      "step": 3566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0473809746009444e-06,
+      "loss": 0.4285,
+      "step": 3567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0431503775614457e-06,
+      "loss": 0.433,
+      "step": 3568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0389236585937944e-06,
+      "loss": 0.4228,
+      "step": 3569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0347008197580376e-06,
+      "loss": 0.4289,
+      "step": 3570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0304818631123393e-06,
+      "loss": 0.4221,
+      "step": 3571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.026266790712965e-06,
+      "loss": 0.4332,
+      "step": 3572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.022055604614289e-06,
+      "loss": 0.4257,
+      "step": 3573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.017848306868797e-06,
+      "loss": 0.4285,
+      "step": 3574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.013644899527074e-06,
+      "loss": 0.4378,
+      "step": 3575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.009445384637805e-06,
+      "loss": 0.4348,
+      "step": 3576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.005249764247783e-06,
+      "loss": 0.4348,
+      "step": 3577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.0010580404019066e-06,
+      "loss": 0.4242,
+      "step": 3578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9968702151431697e-06,
+      "loss": 0.4224,
+      "step": 3579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9926862905126663e-06,
+      "loss": 0.4274,
+      "step": 3580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9885062685495905e-06,
+      "loss": 0.4304,
+      "step": 3581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.984330151291233e-06,
+      "loss": 0.4231,
+      "step": 3582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9801579407729866e-06,
+      "loss": 0.4299,
+      "step": 3583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9759896390283362e-06,
+      "loss": 0.4198,
+      "step": 3584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9718252480888567e-06,
+      "loss": 0.4331,
+      "step": 3585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9676647699842246e-06,
+      "loss": 0.4268,
+      "step": 3586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.963508206742202e-06,
+      "loss": 0.4339,
+      "step": 3587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.959355560388654e-06,
+      "loss": 0.4364,
+      "step": 3588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955206832947526e-06,
+      "loss": 0.4221,
+      "step": 3589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95106202644086e-06,
+      "loss": 0.4215,
+      "step": 3590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9469211428887813e-06,
+      "loss": 0.4167,
+      "step": 3591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9427841843095063e-06,
+      "loss": 0.428,
+      "step": 3592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.938651152719344e-06,
+      "loss": 0.4273,
+      "step": 3593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.934522050132678e-06,
+      "loss": 0.4306,
+      "step": 3594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.930396878561983e-06,
+      "loss": 0.4283,
+      "step": 3595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9262756400178163e-06,
+      "loss": 0.4341,
+      "step": 3596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9221583365088246e-06,
+      "loss": 0.4321,
+      "step": 3597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.918044970041729e-06,
+      "loss": 0.4316,
+      "step": 3598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9139355426213346e-06,
+      "loss": 0.4224,
+      "step": 3599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.4282,
+      "step": 3600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9057285129302682e-06,
+      "loss": 0.4286,
+      "step": 3601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9016309146596024e-06,
+      "loss": 0.429,
+      "step": 3602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8975372634356481e-06,
+      "loss": 0.4207,
+      "step": 3603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8934475612536019e-06,
+      "loss": 0.423,
+      "step": 3604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8893618101067357e-06,
+      "loss": 0.4345,
+      "step": 3605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8852800119863912e-06,
+      "loss": 0.4258,
+      "step": 3606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8812021688819914e-06,
+      "loss": 0.4332,
+      "step": 3607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8771282827810278e-06,
+      "loss": 0.4218,
+      "step": 3608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8730583556690607e-06,
+      "loss": 0.4197,
+      "step": 3609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8689923895297247e-06,
+      "loss": 0.4365,
+      "step": 3610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.86493038634472e-06,
+      "loss": 0.4347,
+      "step": 3611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8608723480938207e-06,
+      "loss": 0.4239,
+      "step": 3612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8568182767548626e-06,
+      "loss": 0.4278,
+      "step": 3613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8527681743037518e-06,
+      "loss": 0.4282,
+      "step": 3614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.848722042714457e-06,
+      "loss": 0.4182,
+      "step": 3615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8446798839590186e-06,
+      "loss": 0.4304,
+      "step": 3616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8406417000075327e-06,
+      "loss": 0.4263,
+      "step": 3617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8366074928281608e-06,
+      "loss": 0.4324,
+      "step": 3618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8325772643871264e-06,
+      "loss": 0.4281,
+      "step": 3619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8285510166487154e-06,
+      "loss": 0.4247,
+      "step": 3620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8245287515752708e-06,
+      "loss": 0.427,
+      "step": 3621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.820510471127196e-06,
+      "loss": 0.427,
+      "step": 3622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.816496177262952e-06,
+      "loss": 0.4286,
+      "step": 3623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.812485871939056e-06,
+      "loss": 0.4262,
+      "step": 3624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.808479557110081e-06,
+      "loss": 0.4259,
+      "step": 3625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.804477234728661e-06,
+      "loss": 0.4329,
+      "step": 3626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8004789067454763e-06,
+      "loss": 0.4311,
+      "step": 3627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7964845751092663e-06,
+      "loss": 0.4233,
+      "step": 3628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7924942417668113e-06,
+      "loss": 0.4251,
+      "step": 3629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7885079086629598e-06,
+      "loss": 0.4183,
+      "step": 3630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7845255777406e-06,
+      "loss": 0.4266,
+      "step": 3631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7805472509406695e-06,
+      "loss": 0.4158,
+      "step": 3632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7765729302021596e-06,
+      "loss": 0.4332,
+      "step": 3633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7726026174621004e-06,
+      "loss": 0.4192,
+      "step": 3634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7686363146555807e-06,
+      "loss": 0.4271,
+      "step": 3635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7646740237157256e-06,
+      "loss": 0.4363,
+      "step": 3636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.760715746573709e-06,
+      "loss": 0.4263,
+      "step": 3637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7567614851587444e-06,
+      "loss": 0.4172,
+      "step": 3638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7528112413980892e-06,
+      "loss": 0.4244,
+      "step": 3639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7488650172170496e-06,
+      "loss": 0.432,
+      "step": 3640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.744922814538964e-06,
+      "loss": 0.4233,
+      "step": 3641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7409846352852144e-06,
+      "loss": 0.4205,
+      "step": 3642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7370504813752232e-06,
+      "loss": 0.4214,
+      "step": 3643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7331203547264452e-06,
+      "loss": 0.4255,
+      "step": 3644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7291942572543806e-06,
+      "loss": 0.4293,
+      "step": 3645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7252721908725633e-06,
+      "loss": 0.4253,
+      "step": 3646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7213541574925551e-06,
+      "loss": 0.4276,
+      "step": 3647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7174401590239587e-06,
+      "loss": 0.4222,
+      "step": 3648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7135301973744122e-06,
+      "loss": 0.4268,
+      "step": 3649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.709624274449584e-06,
+      "loss": 0.4295,
+      "step": 3650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7057223921531706e-06,
+      "loss": 0.4224,
+      "step": 3651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7018245523869038e-06,
+      "loss": 0.429,
+      "step": 3652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6979307570505422e-06,
+      "loss": 0.4258,
+      "step": 3653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6940410080418723e-06,
+      "loss": 0.4258,
+      "step": 3654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6901553072567189e-06,
+      "loss": 0.4259,
+      "step": 3655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.686273656588917e-06,
+      "loss": 0.4202,
+      "step": 3656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6823960579303378e-06,
+      "loss": 0.4295,
+      "step": 3657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6785225131708749e-06,
+      "loss": 0.4272,
+      "step": 3658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6746530241984504e-06,
+      "loss": 0.4263,
+      "step": 3659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6707875928990059e-06,
+      "loss": 0.4292,
+      "step": 3660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.666926221156503e-06,
+      "loss": 0.4197,
+      "step": 3661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6630689108529286e-06,
+      "loss": 0.4314,
+      "step": 3662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6592156638682887e-06,
+      "loss": 0.4252,
+      "step": 3663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6553664820806102e-06,
+      "loss": 0.4249,
+      "step": 3664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.651521367365936e-06,
+      "loss": 0.4366,
+      "step": 3665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6476803215983295e-06,
+      "loss": 0.4151,
+      "step": 3666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.643843346649866e-06,
+      "loss": 0.4184,
+      "step": 3667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6400104443906463e-06,
+      "loss": 0.4251,
+      "step": 3668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6361816166887768e-06,
+      "loss": 0.4283,
+      "step": 3669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6323568654103838e-06,
+      "loss": 0.4353,
+      "step": 3670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6285361924196031e-06,
+      "loss": 0.4279,
+      "step": 3671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6247195995785836e-06,
+      "loss": 0.4229,
+      "step": 3672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6209070887474876e-06,
+      "loss": 0.4271,
+      "step": 3673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6170986617844864e-06,
+      "loss": 0.4244,
+      "step": 3674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6132943205457607e-06,
+      "loss": 0.4214,
+      "step": 3675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6094940668855008e-06,
+      "loss": 0.421,
+      "step": 3676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6056979026559005e-06,
+      "loss": 0.4233,
+      "step": 3677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.601905829707171e-06,
+      "loss": 0.4223,
+      "step": 3678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5981178498875182e-06,
+      "loss": 0.4146,
+      "step": 3679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5943339650431578e-06,
+      "loss": 0.4269,
+      "step": 3680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5905541770183096e-06,
+      "loss": 0.4315,
+      "step": 3681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5867784876551973e-06,
+      "loss": 0.431,
+      "step": 3682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.583006898794044e-06,
+      "loss": 0.4338,
+      "step": 3683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.579239412273078e-06,
+      "loss": 0.4191,
+      "step": 3684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5754760299285255e-06,
+      "loss": 0.4244,
+      "step": 3685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5717167535946142e-06,
+      "loss": 0.4106,
+      "step": 3686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5679615851035669e-06,
+      "loss": 0.425,
+      "step": 3687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5642105262856122e-06,
+      "loss": 0.4247,
+      "step": 3688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.560463578968967e-06,
+      "loss": 0.4215,
+      "step": 3689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5567207449798517e-06,
+      "loss": 0.4167,
+      "step": 3690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.55298202614247e-06,
+      "loss": 0.4266,
+      "step": 3691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5492474242790368e-06,
+      "loss": 0.4173,
+      "step": 3692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.545516941209747e-06,
+      "loss": 0.4186,
+      "step": 3693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5417905787527943e-06,
+      "loss": 0.4167,
+      "step": 3694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.538068338724361e-06,
+      "loss": 0.4193,
+      "step": 3695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5343502229386209e-06,
+      "loss": 0.4261,
+      "step": 3696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.530636233207743e-06,
+      "loss": 0.4323,
+      "step": 3697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.526926371341878e-06,
+      "loss": 0.427,
+      "step": 3698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.52322063914917e-06,
+      "loss": 0.4231,
+      "step": 3699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5195190384357405e-06,
+      "loss": 0.4239,
+      "step": 3700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5158215710057123e-06,
+      "loss": 0.4203,
+      "step": 3701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5121282386611823e-06,
+      "loss": 0.4118,
+      "step": 3702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5084390432022377e-06,
+      "loss": 0.4316,
+      "step": 3703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5047539864269477e-06,
+      "loss": 0.425,
+      "step": 3704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5010730701313626e-06,
+      "loss": 0.4252,
+      "step": 3705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4973962961095135e-06,
+      "loss": 0.4198,
+      "step": 3706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4937236661534227e-06,
+      "loss": 0.4335,
+      "step": 3707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.490055182053083e-06,
+      "loss": 0.4294,
+      "step": 3708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.486390845596466e-06,
+      "loss": 0.4206,
+      "step": 3709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4827306585695234e-06,
+      "loss": 0.4245,
+      "step": 3710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4790746227561925e-06,
+      "loss": 0.425,
+      "step": 3711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4754227399383758e-06,
+      "loss": 0.4236,
+      "step": 3712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4717750118959583e-06,
+      "loss": 0.4376,
+      "step": 3713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.468131440406798e-06,
+      "loss": 0.4261,
+      "step": 3714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4644920272467245e-06,
+      "loss": 0.4249,
+      "step": 3715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4608567741895496e-06,
+      "loss": 0.4305,
+      "step": 3716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4572256830070497e-06,
+      "loss": 0.426,
+      "step": 3717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4535987554689712e-06,
+      "loss": 0.4175,
+      "step": 3718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4499759933430347e-06,
+      "loss": 0.4231,
+      "step": 3719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.446357398394934e-06,
+      "loss": 0.4351,
+      "step": 3720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4427429723883256e-06,
+      "loss": 0.423,
+      "step": 3721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.439132717084839e-06,
+      "loss": 0.4261,
+      "step": 3722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4355266342440678e-06,
+      "loss": 0.4272,
+      "step": 3723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319247256235713e-06,
+      "loss": 0.4223,
+      "step": 3724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4283269929788779e-06,
+      "loss": 0.4303,
+      "step": 3725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247334380634792e-06,
+      "loss": 0.4233,
+      "step": 3726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211440626288286e-06,
+      "loss": 0.423,
+      "step": 3727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175588684243447e-06,
+      "loss": 0.4244,
+      "step": 3728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413977857197405e-06,
+      "loss": 0.4265,
+      "step": 3729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104010306933558e-06,
+      "loss": 0.4227,
+      "step": 3730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4068283906554969e-06,
+      "loss": 0.4217,
+      "step": 3731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40325993882509e-06,
+      "loss": 0.4166,
+      "step": 3732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399695676941354e-06,
+      "loss": 0.4289,
+      "step": 3733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961356067414667e-06,
+      "loss": 0.4303,
+      "step": 3734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925797299605649e-06,
+      "loss": 0.4256,
+      "step": 3735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890280483317375e-06,
+      "loss": 0.4212,
+      "step": 3736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854805635860335e-06,
+      "loss": 0.4206,
+      "step": 3737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381937277452451e-06,
+      "loss": 0.42,
+      "step": 3738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3783981916579448e-06,
+      "loss": 0.4248,
+      "step": 3739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3748633079274254e-06,
+      "loss": 0.4208,
+      "step": 3740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3713326279837502e-06,
+      "loss": 0.4258,
+      "step": 3741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3678061535477305e-06,
+      "loss": 0.4144,
+      "step": 3742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3642838863381258e-06,
+      "loss": 0.433,
+      "step": 3743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3607658280716474e-06,
+      "loss": 0.4203,
+      "step": 3744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3572519804629537e-06,
+      "loss": 0.4145,
+      "step": 3745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3537423452246522e-06,
+      "loss": 0.4274,
+      "step": 3746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3502369240672941e-06,
+      "loss": 0.4283,
+      "step": 3747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3467357186993802e-06,
+      "loss": 0.4265,
+      "step": 3748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3432387308273576e-06,
+      "loss": 0.4278,
+      "step": 3749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.4223,
+      "step": 3750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3362574143864816e-06,
+      "loss": 0.416,
+      "step": 3751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3327730892202384e-06,
+      "loss": 0.4293,
+      "step": 3752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3292929883550998e-06,
+      "loss": 0.4291,
+      "step": 3753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3258171134872267e-06,
+      "loss": 0.4277,
+      "step": 3754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.322345466310717e-06,
+      "loss": 0.4218,
+      "step": 3755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3188780485176089e-06,
+      "loss": 0.4236,
+      "step": 3756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3154148617978813e-06,
+      "loss": 0.4225,
+      "step": 3757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3119559078394462e-06,
+      "loss": 0.4172,
+      "step": 3758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3085011883281606e-06,
+      "loss": 0.4268,
+      "step": 3759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.30505070494781e-06,
+      "loss": 0.4158,
+      "step": 3760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3016044593801202e-06,
+      "loss": 0.4162,
+      "step": 3761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2981624533047432e-06,
+      "loss": 0.4259,
+      "step": 3762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.294724688399278e-06,
+      "loss": 0.4302,
+      "step": 3763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2912911663392468e-06,
+      "loss": 0.4244,
+      "step": 3764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2878618887981064e-06,
+      "loss": 0.4229,
+      "step": 3765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2844368574472454e-06,
+      "loss": 0.4237,
+      "step": 3766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2810160739559797e-06,
+      "loss": 0.4206,
+      "step": 3767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.277599539991563e-06,
+      "loss": 0.4289,
+      "step": 3768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2741872572191684e-06,
+      "loss": 0.4287,
+      "step": 3769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2707792273019049e-06,
+      "loss": 0.415,
+      "step": 3770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2673754519008008e-06,
+      "loss": 0.418,
+      "step": 3771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2639759326748136e-06,
+      "loss": 0.4384,
+      "step": 3772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2605806712808322e-06,
+      "loss": 0.4316,
+      "step": 3773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.257189669373664e-06,
+      "loss": 0.4274,
+      "step": 3774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2538029286060428e-06,
+      "loss": 0.4227,
+      "step": 3775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2504204506286244e-06,
+      "loss": 0.4329,
+      "step": 3776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2470422370899838e-06,
+      "loss": 0.4228,
+      "step": 3777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2436682896366282e-06,
+      "loss": 0.429,
+      "step": 3778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2402986099129765e-06,
+      "loss": 0.4287,
+      "step": 3779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 0.425,
+      "step": 3780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.233572060222057e-06,
+      "loss": 0.418,
+      "step": 3781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.230215193533233e-06,
+      "loss": 0.4282,
+      "step": 3782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2268626011309858e-06,
+      "loss": 0.4383,
+      "step": 3783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.223514284649331e-06,
+      "loss": 0.4284,
+      "step": 3784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2201702457201948e-06,
+      "loss": 0.4173,
+      "step": 3785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2168304859734226e-06,
+      "loss": 0.4292,
+      "step": 3786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2134950070367723e-06,
+      "loss": 0.427,
+      "step": 3787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.210163810535917e-06,
+      "loss": 0.4091,
+      "step": 3788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.206836898094439e-06,
+      "loss": 0.4255,
+      "step": 3789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2035142713338366e-06,
+      "loss": 0.4284,
+      "step": 3790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2001959318735158e-06,
+      "loss": 0.4315,
+      "step": 3791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.196881881330798e-06,
+      "loss": 0.4246,
+      "step": 3792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1935721213209106e-06,
+      "loss": 0.4307,
+      "step": 3793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1902666534569884e-06,
+      "loss": 0.4194,
+      "step": 3794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1869654793500784e-06,
+      "loss": 0.4276,
+      "step": 3795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1836686006091313e-06,
+      "loss": 0.4286,
+      "step": 3796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1803760188410074e-06,
+      "loss": 0.4226,
+      "step": 3797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1770877356504684e-06,
+      "loss": 0.4242,
+      "step": 3798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1738037526401857e-06,
+      "loss": 0.4246,
+      "step": 3799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1705240714107301e-06,
+      "loss": 0.4325,
+      "step": 3800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.167248693560583e-06,
+      "loss": 0.4269,
+      "step": 3801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1639776206861197e-06,
+      "loss": 0.4309,
+      "step": 3802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1607108543816247e-06,
+      "loss": 0.4311,
+      "step": 3803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1574483962392768e-06,
+      "loss": 0.4277,
+      "step": 3804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1541902478491607e-06,
+      "loss": 0.4249,
+      "step": 3805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1509364107992582e-06,
+      "loss": 0.4226,
+      "step": 3806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1476868866754488e-06,
+      "loss": 0.424,
+      "step": 3807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1444416770615118e-06,
+      "loss": 0.4159,
+      "step": 3808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1412007835391237e-06,
+      "loss": 0.4322,
+      "step": 3809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1379642076878528e-06,
+      "loss": 0.424,
+      "step": 3810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1347319510851718e-06,
+      "loss": 0.4202,
+      "step": 3811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1315040153064416e-06,
+      "loss": 0.4264,
+      "step": 3812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1282804019249183e-06,
+      "loss": 0.4329,
+      "step": 3813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1250611125117527e-06,
+      "loss": 0.4243,
+      "step": 3814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1218461486359878e-06,
+      "loss": 0.4226,
+      "step": 3815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1186355118645552e-06,
+      "loss": 0.4194,
+      "step": 3816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1154292037622838e-06,
+      "loss": 0.4223,
+      "step": 3817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1122272258918864e-06,
+      "loss": 0.4242,
+      "step": 3818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1090295798139672e-06,
+      "loss": 0.4226,
+      "step": 3819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.1058362670870248e-06,
+      "loss": 0.4257,
+      "step": 3820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.102647289267438e-06,
+      "loss": 0.4274,
+      "step": 3821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0994626479094749e-06,
+      "loss": 0.4296,
+      "step": 3822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.096282344565296e-06,
+      "loss": 0.4232,
+      "step": 3823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.093106380784934e-06,
+      "loss": 0.4247,
+      "step": 3824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 0.4317,
+      "step": 3825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0867674781052683e-06,
+      "loss": 0.4083,
+      "step": 3826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0836045422954665e-06,
+      "loss": 0.4167,
+      "step": 3827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0804459522284927e-06,
+      "loss": 0.4309,
+      "step": 3828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0772917094438052e-06,
+      "loss": 0.4201,
+      "step": 3829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0741418154787443e-06,
+      "loss": 0.4345,
+      "step": 3830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0709962718685318e-06,
+      "loss": 0.4208,
+      "step": 3831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0678550801462662e-06,
+      "loss": 0.434,
+      "step": 3832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0647182418429224e-06,
+      "loss": 0.4196,
+      "step": 3833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0615857584873624e-06,
+      "loss": 0.4194,
+      "step": 3834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.058457631606319e-06,
+      "loss": 0.4203,
+      "step": 3835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0553338627244026e-06,
+      "loss": 0.4259,
+      "step": 3836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0522144533641e-06,
+      "loss": 0.4269,
+      "step": 3837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0490994050457748e-06,
+      "loss": 0.4211,
+      "step": 3838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0459887192876595e-06,
+      "loss": 0.4264,
+      "step": 3839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.042882397605871e-06,
+      "loss": 0.427,
+      "step": 3840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.039780441514391e-06,
+      "loss": 0.4331,
+      "step": 3841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0366828525250728e-06,
+      "loss": 0.4111,
+      "step": 3842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0335896321476413e-06,
+      "loss": 0.4205,
+      "step": 3843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0305007818897006e-06,
+      "loss": 0.4127,
+      "step": 3844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0274163032567165e-06,
+      "loss": 0.4235,
+      "step": 3845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.024336197752025e-06,
+      "loss": 0.4164,
+      "step": 3846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0212604668768343e-06,
+      "loss": 0.4331,
+      "step": 3847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0181891121302145e-06,
+      "loss": 0.4189,
+      "step": 3848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0151221350091134e-06,
+      "loss": 0.4226,
+      "step": 3849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.012059537008332e-06,
+      "loss": 0.424,
+      "step": 3850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.009001319620545e-06,
+      "loss": 0.4147,
+      "step": 3851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0059474843362893e-06,
+      "loss": 0.4188,
+      "step": 3852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0028980326439708e-06,
+      "loss": 0.418,
+      "step": 3853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.99852966029854e-07,
+      "loss": 0.4161,
+      "step": 3854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.968122859780648e-07,
+      "loss": 0.4215,
+      "step": 3855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.93775993970597e-07,
+      "loss": 0.412,
+      "step": 3856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.907440914873e-07,
+      "loss": 0.4161,
+      "step": 3857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.877165800058874e-07,
+      "loss": 0.4257,
+      "step": 3858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.84693461001932e-07,
+      "loss": 0.4197,
+      "step": 3859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.816747359488632e-07,
+      "loss": 0.426,
+      "step": 3860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.786604063179728e-07,
+      "loss": 0.4194,
+      "step": 3861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.756504735784067e-07,
+      "loss": 0.4268,
+      "step": 3862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.726449391971716e-07,
+      "loss": 0.4214,
+      "step": 3863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.696438046391288e-07,
+      "loss": 0.4162,
+      "step": 3864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.666470713669918e-07,
+      "loss": 0.4233,
+      "step": 3865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.636547408413355e-07,
+      "loss": 0.4252,
+      "step": 3866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.606668145205833e-07,
+      "loss": 0.425,
+      "step": 3867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.576832938610137e-07,
+      "loss": 0.4213,
+      "step": 3868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.547041803167601e-07,
+      "loss": 0.4238,
+      "step": 3869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.517294753398066e-07,
+      "loss": 0.4246,
+      "step": 3870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.487591803799856e-07,
+      "loss": 0.432,
+      "step": 3871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.457932968849826e-07,
+      "loss": 0.4215,
+      "step": 3872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.428318263003378e-07,
+      "loss": 0.4248,
+      "step": 3873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.398747700694322e-07,
+      "loss": 0.4306,
+      "step": 3874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.369221296335007e-07,
+      "loss": 0.417,
+      "step": 3875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.339739064316233e-07,
+      "loss": 0.4222,
+      "step": 3876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.310301019007284e-07,
+      "loss": 0.4231,
+      "step": 3877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.280907174755916e-07,
+      "loss": 0.4163,
+      "step": 3878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.251557545888312e-07,
+      "loss": 0.4293,
+      "step": 3879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.222252146709143e-07,
+      "loss": 0.4248,
+      "step": 3880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.192990991501483e-07,
+      "loss": 0.4304,
+      "step": 3881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.16377409452689e-07,
+      "loss": 0.4256,
+      "step": 3882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.134601470025306e-07,
+      "loss": 0.4189,
+      "step": 3883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.105473132215126e-07,
+      "loss": 0.4251,
+      "step": 3884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.076389095293148e-07,
+      "loss": 0.4227,
+      "step": 3885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.047349373434566e-07,
+      "loss": 0.4188,
+      "step": 3886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.018353980792993e-07,
+      "loss": 0.4212,
+      "step": 3887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.989402931500434e-07,
+      "loss": 0.4204,
+      "step": 3888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.960496239667282e-07,
+      "loss": 0.4253,
+      "step": 3889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.931633919382299e-07,
+      "loss": 0.4242,
+      "step": 3890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.902815984712621e-07,
+      "loss": 0.4175,
+      "step": 3891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.874042449703779e-07,
+      "loss": 0.4324,
+      "step": 3892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.845313328379635e-07,
+      "loss": 0.4147,
+      "step": 3893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.816628634742441e-07,
+      "loss": 0.4215,
+      "step": 3894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.787988382772705e-07,
+      "loss": 0.4269,
+      "step": 3895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.759392586429394e-07,
+      "loss": 0.4312,
+      "step": 3896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.730841259649725e-07,
+      "loss": 0.4148,
+      "step": 3897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.702334416349279e-07,
+      "loss": 0.4164,
+      "step": 3898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.67387207042194e-07,
+      "loss": 0.4176,
+      "step": 3899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.426,
+      "step": 3900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.617080926153698e-07,
+      "loss": 0.4275,
+      "step": 3901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.58875215549212e-07,
+      "loss": 0.4227,
+      "step": 3902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.560467937562278e-07,
+      "loss": 0.4312,
+      "step": 3903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.532228286149502e-07,
+      "loss": 0.4262,
+      "step": 3904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.504033215017527e-07,
+      "loss": 0.4154,
+      "step": 3905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.475882737908248e-07,
+      "loss": 0.423,
+      "step": 3906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.447776868541879e-07,
+      "loss": 0.4194,
+      "step": 3907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.419715620616875e-07,
+      "loss": 0.423,
+      "step": 3908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.39169900780995e-07,
+      "loss": 0.4289,
+      "step": 3909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.363727043776037e-07,
+      "loss": 0.4174,
+      "step": 3910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.335799742148387e-07,
+      "loss": 0.4235,
+      "step": 3911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.307917116538378e-07,
+      "loss": 0.4193,
+      "step": 3912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.280079180535672e-07,
+      "loss": 0.4247,
+      "step": 3913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.252285947708139e-07,
+      "loss": 0.4288,
+      "step": 3914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.224537431601886e-07,
+      "loss": 0.4287,
+      "step": 3915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.196833645741187e-07,
+      "loss": 0.4162,
+      "step": 3916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.169174603628538e-07,
+      "loss": 0.4243,
+      "step": 3917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.141560318744601e-07,
+      "loss": 0.4321,
+      "step": 3918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.113990804548244e-07,
+      "loss": 0.4166,
+      "step": 3919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.086466074476562e-07,
+      "loss": 0.4192,
+      "step": 3920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.058986141944724e-07,
+      "loss": 0.4231,
+      "step": 3921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.031551020346129e-07,
+      "loss": 0.4143,
+      "step": 3922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.004160723052312e-07,
+      "loss": 0.4257,
+      "step": 3923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.976815263412963e-07,
+      "loss": 0.415,
+      "step": 3924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.949514654755963e-07,
+      "loss": 0.4192,
+      "step": 3925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.922258910387282e-07,
+      "loss": 0.418,
+      "step": 3926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.895048043591036e-07,
+      "loss": 0.4163,
+      "step": 3927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.867882067629473e-07,
+      "loss": 0.4277,
+      "step": 3928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.840760995742946e-07,
+      "loss": 0.4199,
+      "step": 3929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.81368484114996e-07,
+      "loss": 0.419,
+      "step": 3930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.78665361704708e-07,
+      "loss": 0.421,
+      "step": 3931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.759667336609011e-07,
+      "loss": 0.4269,
+      "step": 3932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.732726012988512e-07,
+      "loss": 0.4274,
+      "step": 3933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.7058296593165e-07,
+      "loss": 0.4236,
+      "step": 3934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.678978288701911e-07,
+      "loss": 0.4243,
+      "step": 3935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.652171914231777e-07,
+      "loss": 0.4223,
+      "step": 3936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.62541054897119e-07,
+      "loss": 0.4126,
+      "step": 3937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.598694205963331e-07,
+      "loss": 0.4165,
+      "step": 3938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.572022898229403e-07,
+      "loss": 0.4203,
+      "step": 3939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.545396638768698e-07,
+      "loss": 0.4161,
+      "step": 3940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.518815440558514e-07,
+      "loss": 0.4233,
+      "step": 3941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.492279316554207e-07,
+      "loss": 0.4198,
+      "step": 3942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.465788279689156e-07,
+      "loss": 0.4209,
+      "step": 3943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.439342342874789e-07,
+      "loss": 0.4268,
+      "step": 3944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.412941519000527e-07,
+      "loss": 0.4254,
+      "step": 3945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.386585820933812e-07,
+      "loss": 0.4215,
+      "step": 3946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.360275261520078e-07,
+      "loss": 0.4148,
+      "step": 3947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.334009853582791e-07,
+      "loss": 0.4224,
+      "step": 3948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.307789609923377e-07,
+      "loss": 0.429,
+      "step": 3949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.281614543321269e-07,
+      "loss": 0.428,
+      "step": 3950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.255484666533874e-07,
+      "loss": 0.4198,
+      "step": 3951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.22939999229657e-07,
+      "loss": 0.4253,
+      "step": 3952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.203360533322734e-07,
+      "loss": 0.4244,
+      "step": 3953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.177366302303667e-07,
+      "loss": 0.4239,
+      "step": 3954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.151417311908648e-07,
+      "loss": 0.4229,
+      "step": 3955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.125513574784904e-07,
+      "loss": 0.4144,
+      "step": 3956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.099655103557557e-07,
+      "loss": 0.4214,
+      "step": 3957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.073841910829771e-07,
+      "loss": 0.4317,
+      "step": 3958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.048074009182548e-07,
+      "loss": 0.4213,
+      "step": 3959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 0.4241,
+      "step": 3960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.996674129343606e-07,
+      "loss": 0.4264,
+      "step": 3961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.971042176203535e-07,
+      "loss": 0.4301,
+      "step": 3962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.945455564247394e-07,
+      "loss": 0.431,
+      "step": 3963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.919914305945774e-07,
+      "loss": 0.4171,
+      "step": 3964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.894418413747183e-07,
+      "loss": 0.4154,
+      "step": 3965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.868967900077972e-07,
+      "loss": 0.4173,
+      "step": 3966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.84356277734245e-07,
+      "loss": 0.4257,
+      "step": 3967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.818203057922756e-07,
+      "loss": 0.4227,
+      "step": 3968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.792888754178906e-07,
+      "loss": 0.4206,
+      "step": 3969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.767619878448783e-07,
+      "loss": 0.4182,
+      "step": 3970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742396443048138e-07,
+      "loss": 0.4151,
+      "step": 3971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.717218460270536e-07,
+      "loss": 0.4174,
+      "step": 3972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.692085942387483e-07,
+      "loss": 0.4254,
+      "step": 3973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.666998901648203e-07,
+      "loss": 0.4219,
+      "step": 3974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.641957350279838e-07,
+      "loss": 0.4282,
+      "step": 3975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616961300487323e-07,
+      "loss": 0.4179,
+      "step": 3976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.592010764453449e-07,
+      "loss": 0.4308,
+      "step": 3977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.567105754338798e-07,
+      "loss": 0.4184,
+      "step": 3978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.542246282281772e-07,
+      "loss": 0.4212,
+      "step": 3979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.517432360398556e-07,
+      "loss": 0.4155,
+      "step": 3980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.492664000783166e-07,
+      "loss": 0.4228,
+      "step": 3981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.467941215507434e-07,
+      "loss": 0.4274,
+      "step": 3982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.443264016620887e-07,
+      "loss": 0.4252,
+      "step": 3983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.418632416150927e-07,
+      "loss": 0.4206,
+      "step": 3984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.394046426102673e-07,
+      "loss": 0.4275,
+      "step": 3985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.369506058459063e-07,
+      "loss": 0.4239,
+      "step": 3986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.345011325180772e-07,
+      "loss": 0.4278,
+      "step": 3987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.320562238206218e-07,
+      "loss": 0.4292,
+      "step": 3988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.296158809451602e-07,
+      "loss": 0.414,
+      "step": 3989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.271801050810856e-07,
+      "loss": 0.4159,
+      "step": 3990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.247488974155657e-07,
+      "loss": 0.4223,
+      "step": 3991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.223222591335409e-07,
+      "loss": 0.4219,
+      "step": 3992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.199001914177261e-07,
+      "loss": 0.4265,
+      "step": 3993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.174826954486069e-07,
+      "loss": 0.423,
+      "step": 3994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.150697724044407e-07,
+      "loss": 0.4195,
+      "step": 3995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.126614234612593e-07,
+      "loss": 0.4209,
+      "step": 3996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.102576497928614e-07,
+      "loss": 0.4304,
+      "step": 3997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.078584525708175e-07,
+      "loss": 0.4206,
+      "step": 3998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.054638329644658e-07,
+      "loss": 0.4248,
+      "step": 3999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.4207,
+      "step": 4000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.006883312650458e-07,
+      "loss": 0.4268,
+      "step": 4001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.98307451499498e-07,
+      "loss": 0.4245,
+      "step": 4002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.959311540046863e-07,
+      "loss": 0.4177,
+      "step": 4003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.935594399387856e-07,
+      "loss": 0.4339,
+      "step": 4004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.911923104577455e-07,
+      "loss": 0.419,
+      "step": 4005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.888297667152731e-07,
+      "loss": 0.4235,
+      "step": 4006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.864718098628441e-07,
+      "loss": 0.42,
+      "step": 4007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.841184410496992e-07,
+      "loss": 0.423,
+      "step": 4008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.817696614228396e-07,
+      "loss": 0.4189,
+      "step": 4009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.794254721270331e-07,
+      "loss": 0.4278,
+      "step": 4010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.770858743048091e-07,
+      "loss": 0.4244,
+      "step": 4011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.747508690964599e-07,
+      "loss": 0.4162,
+      "step": 4012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.724204576400372e-07,
+      "loss": 0.4355,
+      "step": 4013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.700946410713548e-07,
+      "loss": 0.4231,
+      "step": 4014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.677734205239904e-07,
+      "loss": 0.4251,
+      "step": 4015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.654567971292757e-07,
+      "loss": 0.4197,
+      "step": 4016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.631447720163074e-07,
+      "loss": 0.4196,
+      "step": 4017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.608373463119354e-07,
+      "loss": 0.4155,
+      "step": 4018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.585345211407734e-07,
+      "loss": 0.4181,
+      "step": 4019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.562362976251901e-07,
+      "loss": 0.4147,
+      "step": 4020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.539426768853107e-07,
+      "loss": 0.4285,
+      "step": 4021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.516536600390188e-07,
+      "loss": 0.4257,
+      "step": 4022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.49369248201953e-07,
+      "loss": 0.4194,
+      "step": 4023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.470894424875062e-07,
+      "loss": 0.4247,
+      "step": 4024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.448142440068316e-07,
+      "loss": 0.4283,
+      "step": 4025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.425436538688322e-07,
+      "loss": 0.4161,
+      "step": 4026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.402776731801662e-07,
+      "loss": 0.4253,
+      "step": 4027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.380163030452412e-07,
+      "loss": 0.4186,
+      "step": 4028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.357595445662267e-07,
+      "loss": 0.4157,
+      "step": 4029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.335073988430373e-07,
+      "loss": 0.4187,
+      "step": 4030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.312598669733404e-07,
+      "loss": 0.4231,
+      "step": 4031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.290169500525577e-07,
+      "loss": 0.4132,
+      "step": 4032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.267786491738569e-07,
+      "loss": 0.4227,
+      "step": 4033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.245449654281632e-07,
+      "loss": 0.4266,
+      "step": 4034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.223158999041444e-07,
+      "loss": 0.42,
+      "step": 4035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.200914536882184e-07,
+      "loss": 0.4306,
+      "step": 4036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.178716278645534e-07,
+      "loss": 0.4197,
+      "step": 4037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.156564235150686e-07,
+      "loss": 0.421,
+      "step": 4038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.134458417194255e-07,
+      "loss": 0.4238,
+      "step": 4039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.112398835550348e-07,
+      "loss": 0.422,
+      "step": 4040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.090385500970551e-07,
+      "loss": 0.4273,
+      "step": 4041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.068418424183874e-07,
+      "loss": 0.4271,
+      "step": 4042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.046497615896806e-07,
+      "loss": 0.4173,
+      "step": 4043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.024623086793323e-07,
+      "loss": 0.4232,
+      "step": 4044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.002794847534765e-07,
+      "loss": 0.4258,
+      "step": 4045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.981012908759941e-07,
+      "loss": 0.4233,
+      "step": 4046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.959277281085128e-07,
+      "loss": 0.4214,
+      "step": 4047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.937587975103997e-07,
+      "loss": 0.4273,
+      "step": 4048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.915945001387668e-07,
+      "loss": 0.4186,
+      "step": 4049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.4241,
+      "step": 4050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.872798092920871e-07,
+      "loss": 0.4195,
+      "step": 4051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.851294179199673e-07,
+      "loss": 0.4133,
+      "step": 4052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.829836639801844e-07,
+      "loss": 0.4245,
+      "step": 4053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.808425485185486e-07,
+      "loss": 0.4133,
+      "step": 4054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.787060725786141e-07,
+      "loss": 0.4221,
+      "step": 4055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.765742372016735e-07,
+      "loss": 0.4276,
+      "step": 4056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7444704342675673e-07,
+      "loss": 0.4223,
+      "step": 4057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.723244922906356e-07,
+      "loss": 0.4164,
+      "step": 4058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.702065848278126e-07,
+      "loss": 0.4273,
+      "step": 4059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6809332207053083e-07,
+      "loss": 0.4152,
+      "step": 4060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.659847050487687e-07,
+      "loss": 0.4226,
+      "step": 4061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.638807347902408e-07,
+      "loss": 0.4279,
+      "step": 4062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6178141232039676e-07,
+      "loss": 0.4186,
+      "step": 4063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.596867386624215e-07,
+      "loss": 0.4193,
+      "step": 4064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.575967148372318e-07,
+      "loss": 0.4179,
+      "step": 4065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5551134186348045e-07,
+      "loss": 0.4185,
+      "step": 4066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.534306207575545e-07,
+      "loss": 0.4255,
+      "step": 4067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5135455253357053e-07,
+      "loss": 0.4145,
+      "step": 4068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.492831382033791e-07,
+      "loss": 0.4221,
+      "step": 4069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4721637877656377e-07,
+      "loss": 0.4181,
+      "step": 4070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.451542752604365e-07,
+      "loss": 0.4119,
+      "step": 4071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4309682866004124e-07,
+      "loss": 0.4211,
+      "step": 4072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4104403997815346e-07,
+      "loss": 0.4275,
+      "step": 4073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3899591021527743e-07,
+      "loss": 0.4239,
+      "step": 4074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3695244036964567e-07,
+      "loss": 0.419,
+      "step": 4075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.349136314372204e-07,
+      "loss": 0.4257,
+      "step": 4076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3287948441169457e-07,
+      "loss": 0.4276,
+      "step": 4077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308500002844862e-07,
+      "loss": 0.4212,
+      "step": 4078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.288251800447396e-07,
+      "loss": 0.4158,
+      "step": 4079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.268050246793276e-07,
+      "loss": 0.4271,
+      "step": 4080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.247895351728504e-07,
+      "loss": 0.4166,
+      "step": 4081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2277871250763327e-07,
+      "loss": 0.417,
+      "step": 4082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.207725576637256e-07,
+      "loss": 0.4268,
+      "step": 4083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1877107161890416e-07,
+      "loss": 0.4266,
+      "step": 4084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.167742553486676e-07,
+      "loss": 0.4139,
+      "step": 4085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1478210982624055e-07,
+      "loss": 0.4193,
+      "step": 4086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.1279463602257207e-07,
+      "loss": 0.4257,
+      "step": 4087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.108118349063306e-07,
+      "loss": 0.4249,
+      "step": 4088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.0883370744390973e-07,
+      "loss": 0.4198,
+      "step": 4089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.068602545994249e-07,
+      "loss": 0.4202,
+      "step": 4090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.0489147733471347e-07,
+      "loss": 0.4188,
+      "step": 4091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.0292737660933335e-07,
+      "loss": 0.4212,
+      "step": 4092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.009679533805633e-07,
+      "loss": 0.4246,
+      "step": 4093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.990132086034026e-07,
+      "loss": 0.4171,
+      "step": 4094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9706314323056936e-07,
+      "loss": 0.4232,
+      "step": 4095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9511775821250206e-07,
+      "loss": 0.4209,
+      "step": 4096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.931770544973601e-07,
+      "loss": 0.4271,
+      "step": 4097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.912410330310157e-07,
+      "loss": 0.4128,
+      "step": 4098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.8930969475706183e-07,
+      "loss": 0.4135,
+      "step": 4099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.8738304061681107e-07,
+      "loss": 0.4142,
+      "step": 4100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.854610715492924e-07,
+      "loss": 0.4232,
+      "step": 4101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.835437884912474e-07,
+      "loss": 0.4343,
+      "step": 4102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.8163119237713877e-07,
+      "loss": 0.4203,
+      "step": 4103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7972328413914074e-07,
+      "loss": 0.4308,
+      "step": 4104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7782006470714614e-07,
+      "loss": 0.4189,
+      "step": 4105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.759215350087619e-07,
+      "loss": 0.4135,
+      "step": 4106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7402769596930567e-07,
+      "loss": 0.4216,
+      "step": 4107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.721385485118123e-07,
+      "loss": 0.4118,
+      "step": 4108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7025409355702977e-07,
+      "loss": 0.4141,
+      "step": 4109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.68374332023419e-07,
+      "loss": 0.4194,
+      "step": 4110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.664992648271526e-07,
+      "loss": 0.4166,
+      "step": 4111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.646288928821151e-07,
+      "loss": 0.4156,
+      "step": 4112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.627632170999029e-07,
+      "loss": 0.4169,
+      "step": 4113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.609022383898242e-07,
+      "loss": 0.4188,
+      "step": 4114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.590459576589e-07,
+      "loss": 0.4249,
+      "step": 4115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.571943758118546e-07,
+      "loss": 0.4271,
+      "step": 4116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.553474937511281e-07,
+      "loss": 0.4226,
+      "step": 4117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5350531237686723e-07,
+      "loss": 0.4145,
+      "step": 4118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.516678325869316e-07,
+      "loss": 0.4174,
+      "step": 4119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.498350552768859e-07,
+      "loss": 0.4131,
+      "step": 4120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.480069813400022e-07,
+      "loss": 0.4199,
+      "step": 4121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4618361166726123e-07,
+      "loss": 0.418,
+      "step": 4122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4436494714735313e-07,
+      "loss": 0.4236,
+      "step": 4123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4255098866667114e-07,
+      "loss": 0.4193,
+      "step": 4124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.4228,
+      "step": 4125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.3893719335709953e-07,
+      "loss": 0.4249,
+      "step": 4126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.3713735828952985e-07,
+      "loss": 0.4209,
+      "step": 4127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.3534223278382405e-07,
+      "loss": 0.4204,
+      "step": 4128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.3355181771490776e-07,
+      "loss": 0.427,
+      "step": 4129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.3176611395540625e-07,
+      "loss": 0.413,
+      "step": 4130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2998512237565005e-07,
+      "loss": 0.4198,
+      "step": 4131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.282088438436715e-07,
+      "loss": 0.4174,
+      "step": 4132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2643727922520905e-07,
+      "loss": 0.4299,
+      "step": 4133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.246704293837011e-07,
+      "loss": 0.425,
+      "step": 4134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2290829518028867e-07,
+      "loss": 0.4188,
+      "step": 4135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.211508774738137e-07,
+      "loss": 0.4103,
+      "step": 4136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.19398177120821e-07,
+      "loss": 0.417,
+      "step": 4137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1765019497555617e-07,
+      "loss": 0.4126,
+      "step": 4138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1590693188996324e-07,
+      "loss": 0.4193,
+      "step": 4139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 0.4153,
+      "step": 4140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1243456629407644e-07,
+      "loss": 0.4249,
+      "step": 4141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.10705465476171e-07,
+      "loss": 0.4257,
+      "step": 4142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0898108710271437e-07,
+      "loss": 0.4188,
+      "step": 4143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.072614320141487e-07,
+      "loss": 0.4226,
+      "step": 4144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0554650104861137e-07,
+      "loss": 0.4221,
+      "step": 4145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0383629504194047e-07,
+      "loss": 0.4217,
+      "step": 4146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0213081482766803e-07,
+      "loss": 0.416,
+      "step": 4147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.00430061237027e-07,
+      "loss": 0.4261,
+      "step": 4148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.987340350989421e-07,
+      "loss": 0.4227,
+      "step": 4149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.970427372400353e-07,
+      "loss": 0.4248,
+      "step": 4150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9535616848462624e-07,
+      "loss": 0.4164,
+      "step": 4151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.936743296547273e-07,
+      "loss": 0.4196,
+      "step": 4152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.919972215700462e-07,
+      "loss": 0.4187,
+      "step": 4153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9032484504798454e-07,
+      "loss": 0.4206,
+      "step": 4154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8865720090364037e-07,
+      "loss": 0.4341,
+      "step": 4155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8699428994980017e-07,
+      "loss": 0.4107,
+      "step": 4156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8533611299694784e-07,
+      "loss": 0.4159,
+      "step": 4157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.836826708532603e-07,
+      "loss": 0.4172,
+      "step": 4158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8203396432460507e-07,
+      "loss": 0.4212,
+      "step": 4159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8038999421453827e-07,
+      "loss": 0.4205,
+      "step": 4160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7875076132431344e-07,
+      "loss": 0.415,
+      "step": 4161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.771162664528726e-07,
+      "loss": 0.4152,
+      "step": 4162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7548651039684847e-07,
+      "loss": 0.419,
+      "step": 4163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7386149395056463e-07,
+      "loss": 0.425,
+      "step": 4164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7224121790603517e-07,
+      "loss": 0.4188,
+      "step": 4165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7062568305295967e-07,
+      "loss": 0.4282,
+      "step": 4166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6901489017873375e-07,
+      "loss": 0.41,
+      "step": 4167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6740884006843826e-07,
+      "loss": 0.4168,
+      "step": 4168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6580753350484044e-07,
+      "loss": 0.4153,
+      "step": 4169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6421097126839714e-07,
+      "loss": 0.4289,
+      "step": 4170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.626191541372558e-07,
+      "loss": 0.4214,
+      "step": 4171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.6103208288724815e-07,
+      "loss": 0.4209,
+      "step": 4172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.59449758291892e-07,
+      "loss": 0.4218,
+      "step": 4173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.57872181122395e-07,
+      "loss": 0.4263,
+      "step": 4174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5629935214764866e-07,
+      "loss": 0.4166,
+      "step": 4175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.547312721342277e-07,
+      "loss": 0.4196,
+      "step": 4176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5316794184640056e-07,
+      "loss": 0.4305,
+      "step": 4177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.516093620461124e-07,
+      "loss": 0.4133,
+      "step": 4178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.500555334929955e-07,
+      "loss": 0.4221,
+      "step": 4179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.4850645694436736e-07,
+      "loss": 0.4227,
+      "step": 4180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.4696213315523074e-07,
+      "loss": 0.4098,
+      "step": 4181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.4542256287826915e-07,
+      "loss": 0.419,
+      "step": 4182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.4388774686385007e-07,
+      "loss": 0.4332,
+      "step": 4183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.423576858600252e-07,
+      "loss": 0.4196,
+      "step": 4184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.4083238061252565e-07,
+      "loss": 0.4241,
+      "step": 4185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3931183186477026e-07,
+      "loss": 0.411,
+      "step": 4186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3779604035785277e-07,
+      "loss": 0.4126,
+      "step": 4187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3628500683055222e-07,
+      "loss": 0.4222,
+      "step": 4188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3477873201932733e-07,
+      "loss": 0.4242,
+      "step": 4189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.332772166583208e-07,
+      "loss": 0.4166,
+      "step": 4190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3178046147935173e-07,
+      "loss": 0.4245,
+      "step": 4191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3028846721191878e-07,
+      "loss": 0.415,
+      "step": 4192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.288012345832047e-07,
+      "loss": 0.4212,
+      "step": 4193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.273187643180652e-07,
+      "loss": 0.4166,
+      "step": 4194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2584105713904126e-07,
+      "loss": 0.4186,
+      "step": 4195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2436811376634893e-07,
+      "loss": 0.4173,
+      "step": 4196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2289993491788065e-07,
+      "loss": 0.4233,
+      "step": 4197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.214365213092118e-07,
+      "loss": 0.42,
+      "step": 4198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1997787365358958e-07,
+      "loss": 0.4184,
+      "step": 4199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.4183,
+      "step": 4200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1707487904287672e-07,
+      "loss": 0.4214,
+      "step": 4201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1563053350266983e-07,
+      "loss": 0.4243,
+      "step": 4202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1419095674527934e-07,
+      "loss": 0.4233,
+      "step": 4203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1275614947233624e-07,
+      "loss": 0.4167,
+      "step": 4204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1132611238315004e-07,
+      "loss": 0.4135,
+      "step": 4205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.0990084617470207e-07,
+      "loss": 0.416,
+      "step": 4206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.0848035154165002e-07,
+      "loss": 0.4204,
+      "step": 4207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.0706462917632676e-07,
+      "loss": 0.4205,
+      "step": 4208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.0565367976873584e-07,
+      "loss": 0.4178,
+      "step": 4209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.0424750400655947e-07,
+      "loss": 0.416,
+      "step": 4210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.0284610257514936e-07,
+      "loss": 0.4204,
+      "step": 4211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.014494761575314e-07,
+      "loss": 0.418,
+      "step": 4212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.0005762543440444e-07,
+      "loss": 0.4241,
+      "step": 4213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9867055108414023e-07,
+      "loss": 0.4137,
+      "step": 4214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9728825378278248e-07,
+      "loss": 0.4168,
+      "step": 4215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9591073420404338e-07,
+      "loss": 0.4172,
+      "step": 4216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9453799301931253e-07,
+      "loss": 0.4262,
+      "step": 4217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9317003089764365e-07,
+      "loss": 0.4294,
+      "step": 4218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9180684850576893e-07,
+      "loss": 0.4158,
+      "step": 4219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9044844650808468e-07,
+      "loss": 0.4262,
+      "step": 4220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8909482556666026e-07,
+      "loss": 0.4178,
+      "step": 4221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.877459863412323e-07,
+      "loss": 0.4145,
+      "step": 4222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8640192948921053e-07,
+      "loss": 0.4279,
+      "step": 4223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8506265566567095e-07,
+      "loss": 0.4214,
+      "step": 4224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8372816552336025e-07,
+      "loss": 0.413,
+      "step": 4225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8239845971269266e-07,
+      "loss": 0.419,
+      "step": 4226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8107353888175083e-07,
+      "loss": 0.416,
+      "step": 4227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7975340367628269e-07,
+      "loss": 0.423,
+      "step": 4228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7843805473970798e-07,
+      "loss": 0.4192,
+      "step": 4229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7712749271311392e-07,
+      "loss": 0.4243,
+      "step": 4230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.758217182352495e-07,
+      "loss": 0.4267,
+      "step": 4231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7452073194253237e-07,
+      "loss": 0.413,
+      "step": 4232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7322453446905084e-07,
+      "loss": 0.4212,
+      "step": 4233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.719331264465529e-07,
+      "loss": 0.4216,
+      "step": 4234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.706465085044584e-07,
+      "loss": 0.4215,
+      "step": 4235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6936468126984573e-07,
+      "loss": 0.4168,
+      "step": 4236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.680876453674629e-07,
+      "loss": 0.4186,
+      "step": 4237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.668154014197243e-07,
+      "loss": 0.4158,
+      "step": 4238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6554795004670389e-07,
+      "loss": 0.4188,
+      "step": 4239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6428529186614195e-07,
+      "loss": 0.4252,
+      "step": 4240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6302742749344292e-07,
+      "loss": 0.4268,
+      "step": 4241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6177435754167413e-07,
+      "loss": 0.4181,
+      "step": 4242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.605260826215682e-07,
+      "loss": 0.4228,
+      "step": 4243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5928260334151847e-07,
+      "loss": 0.431,
+      "step": 4244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.580439203075812e-07,
+      "loss": 0.4158,
+      "step": 4245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5681003412347573e-07,
+      "loss": 0.4257,
+      "step": 4246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.555809453905821e-07,
+      "loss": 0.4155,
+      "step": 4247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.543566547079467e-07,
+      "loss": 0.418,
+      "step": 4248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5313716267226997e-07,
+      "loss": 0.4068,
+      "step": 4249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.4334,
+      "step": 4250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5071257691692153e-07,
+      "loss": 0.4224,
+      "step": 4251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4950748437896235e-07,
+      "loss": 0.4183,
+      "step": 4252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.483071928513913e-07,
+      "loss": 0.4237,
+      "step": 4253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4711170291921485e-07,
+      "loss": 0.4279,
+      "step": 4254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4592101516509916e-07,
+      "loss": 0.4245,
+      "step": 4255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4473513016937223e-07,
+      "loss": 0.4237,
+      "step": 4256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4355404851001953e-07,
+      "loss": 0.4253,
+      "step": 4257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4237777076268723e-07,
+      "loss": 0.4238,
+      "step": 4258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4120629750067672e-07,
+      "loss": 0.4215,
+      "step": 4259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.400396292949513e-07,
+      "loss": 0.4119,
+      "step": 4260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3887776671412943e-07,
+      "loss": 0.4266,
+      "step": 4261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.377207103244904e-07,
+      "loss": 0.417,
+      "step": 4262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3656846068996976e-07,
+      "loss": 0.422,
+      "step": 4263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3542101837215826e-07,
+      "loss": 0.4167,
+      "step": 4264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3427838393030634e-07,
+      "loss": 0.4226,
+      "step": 4265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.3314055792131964e-07,
+      "loss": 0.4158,
+      "step": 4266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.320075408997612e-07,
+      "loss": 0.4199,
+      "step": 4267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.308793334178493e-07,
+      "loss": 0.4257,
+      "step": 4268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2975593602545966e-07,
+      "loss": 0.4226,
+      "step": 4269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2863734927012094e-07,
+      "loss": 0.4315,
+      "step": 4270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.275235736970193e-07,
+      "loss": 0.4138,
+      "step": 4271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.26414609848996e-07,
+      "loss": 0.4258,
+      "step": 4272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531045826654652e-07,
+      "loss": 0.4128,
+      "step": 4273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242111194878215e-07,
+      "loss": 0.4114,
+      "step": 4274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 0.4116,
+      "step": 4275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2202688248241113e-07,
+      "loss": 0.4178,
+      "step": 4276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2094198532029754e-07,
+      "loss": 0.4271,
+      "step": 4277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1986190309104861e-07,
+      "loss": 0.4301,
+      "step": 4278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1878663632108322e-07,
+      "loss": 0.4126,
+      "step": 4279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1771618553447217e-07,
+      "loss": 0.4228,
+      "step": 4280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1665055125294033e-07,
+      "loss": 0.4286,
+      "step": 4281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1558973399586671e-07,
+      "loss": 0.419,
+      "step": 4282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1453373428027992e-07,
+      "loss": 0.4203,
+      "step": 4283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.134825526208605e-07,
+      "loss": 0.4371,
+      "step": 4284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1243618952994195e-07,
+      "loss": 0.4169,
+      "step": 4285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1139464551750857e-07,
+      "loss": 0.4195,
+      "step": 4286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1035792109119758e-07,
+      "loss": 0.4242,
+      "step": 4287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0932601675629595e-07,
+      "loss": 0.4213,
+      "step": 4288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0829893301573913e-07,
+      "loss": 0.4303,
+      "step": 4289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0727667037011668e-07,
+      "loss": 0.4227,
+      "step": 4290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0625922931766786e-07,
+      "loss": 0.4139,
+      "step": 4291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.052466103542793e-07,
+      "loss": 0.4273,
+      "step": 4292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0423881397349067e-07,
+      "loss": 0.4268,
+      "step": 4293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0323584066648795e-07,
+      "loss": 0.419,
+      "step": 4294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0223769092211012e-07,
+      "loss": 0.4231,
+      "step": 4295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.0124436522684244e-07,
+      "loss": 0.4187,
+      "step": 4296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.002558640648199e-07,
+      "loss": 0.4314,
+      "step": 4297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.9272187917826e-08,
+      "loss": 0.4172,
+      "step": 4298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.829333726529056e-08,
+      "loss": 0.4207,
+      "step": 4299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.731931258429638e-08,
+      "loss": 0.4128,
+      "step": 4300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.635011434957153e-08,
+      "loss": 0.4258,
+      "step": 4301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.538574303348813e-08,
+      "loss": 0.4233,
+      "step": 4302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.442619910607131e-08,
+      "loss": 0.4256,
+      "step": 4303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.347148303499143e-08,
+      "loss": 0.4114,
+      "step": 4304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.252159528556404e-08,
+      "loss": 0.4165,
+      "step": 4305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.157653632075435e-08,
+      "loss": 0.417,
+      "step": 4306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.063630660117172e-08,
+      "loss": 0.4231,
+      "step": 4307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.970090658507291e-08,
+      "loss": 0.4292,
+      "step": 4308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.877033672835988e-08,
+      "loss": 0.4179,
+      "step": 4309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.784459748458318e-08,
+      "loss": 0.4292,
+      "step": 4310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.692368930493522e-08,
+      "loss": 0.4245,
+      "step": 4311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.600761263825475e-08,
+      "loss": 0.419,
+      "step": 4312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.509636793102683e-08,
+      "loss": 0.4235,
+      "step": 4313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.418995562738286e-08,
+      "loss": 0.4234,
+      "step": 4314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.328837616909612e-08,
+      "loss": 0.4314,
+      "step": 4315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.239162999558403e-08,
+      "loss": 0.4157,
+      "step": 4316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.149971754391251e-08,
+      "loss": 0.4197,
+      "step": 4317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.061263924878604e-08,
+      "loss": 0.4117,
+      "step": 4318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.973039554255768e-08,
+      "loss": 0.4351,
+      "step": 4319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 0.4252,
+      "step": 4320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.798041361441688e-08,
+      "loss": 0.4178,
+      "step": 4321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.71126762454233e-08,
+      "loss": 0.4279,
+      "step": 4322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.624977517116772e-08,
+      "loss": 0.4233,
+      "step": 4323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.539171081221597e-08,
+      "loss": 0.4236,
+      "step": 4324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.453848358678018e-08,
+      "loss": 0.4226,
+      "step": 4325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.369009391070992e-08,
+      "loss": 0.4214,
+      "step": 4326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.284654219750332e-08,
+      "loss": 0.4231,
+      "step": 4327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.200782885829482e-08,
+      "loss": 0.4197,
+      "step": 4328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.117395430186414e-08,
+      "loss": 0.4229,
+      "step": 4329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.034491893463059e-08,
+      "loss": 0.4223,
+      "step": 4330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.95207231606576e-08,
+      "loss": 0.427,
+      "step": 4331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.870136738164612e-08,
+      "loss": 0.417,
+      "step": 4332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.788685199694222e-08,
+      "loss": 0.4266,
+      "step": 4333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.707717740353059e-08,
+      "loss": 0.42,
+      "step": 4334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.627234399603554e-08,
+      "loss": 0.429,
+      "step": 4335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.547235216672443e-08,
+      "loss": 0.4177,
+      "step": 4336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.4677202305502e-08,
+      "loss": 0.4224,
+      "step": 4337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.388689479991606e-08,
+      "loss": 0.4309,
+      "step": 4338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.310143003515179e-08,
+      "loss": 0.4182,
+      "step": 4339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.232080839403631e-08,
+      "loss": 0.4296,
+      "step": 4340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.154503025703418e-08,
+      "loss": 0.4248,
+      "step": 4341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07740960022507e-08,
+      "loss": 0.4234,
+      "step": 4342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.000800600542977e-08,
+      "loss": 0.4233,
+      "step": 4343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9246760639953824e-08,
+      "loss": 0.4216,
+      "step": 4344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.849036027684607e-08,
+      "loss": 0.4219,
+      "step": 4345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7738805284764945e-08,
+      "loss": 0.4197,
+      "step": 4346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.699209603001077e-08,
+      "loss": 0.4281,
+      "step": 4347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.625023287652021e-08,
+      "loss": 0.4163,
+      "step": 4348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.5513216185867356e-08,
+      "loss": 0.419,
+      "step": 4349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 0.4305,
+      "step": 4350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.4053723627567336e-08,
+      "loss": 0.4313,
+      "step": 4351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.3331248471258926e-08,
+      "loss": 0.4213,
+      "step": 4352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.261362120046687e-08,
+      "loss": 0.4126,
+      "step": 4353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.190084216495361e-08,
+      "loss": 0.4263,
+      "step": 4354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.119291171211793e-08,
+      "loss": 0.4277,
+      "step": 4355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.048983018699827e-08,
+      "loss": 0.4264,
+      "step": 4356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.979159793226718e-08,
+      "loss": 0.4136,
+      "step": 4357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9098215288235776e-08,
+      "loss": 0.4154,
+      "step": 4358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.840968259284817e-08,
+      "loss": 0.4141,
+      "step": 4359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.772600018168816e-08,
+      "loss": 0.4121,
+      "step": 4360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.704716838797363e-08,
+      "loss": 0.4133,
+      "step": 4361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.6373187542561036e-08,
+      "loss": 0.429,
+      "step": 4362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.570405797393762e-08,
+      "loss": 0.4197,
+      "step": 4363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.503978000823028e-08,
+      "loss": 0.4228,
+      "step": 4364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.438035396920004e-08,
+      "loss": 0.4141,
+      "step": 4365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.3725780178243135e-08,
+      "loss": 0.4182,
+      "step": 4366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.3076058954391045e-08,
+      "loss": 0.4204,
+      "step": 4367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2431190614309334e-08,
+      "loss": 0.4344,
+      "step": 4368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.179117547229883e-08,
+      "loss": 0.4251,
+      "step": 4369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.115601384029666e-08,
+      "loss": 0.4113,
+      "step": 4370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.052570602787076e-08,
+      "loss": 0.4244,
+      "step": 4371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.990025234222872e-08,
+      "loss": 0.423,
+      "step": 4372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.927965308820558e-08,
+      "loss": 0.4276,
+      "step": 4373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.866390856827495e-08,
+      "loss": 0.424,
+      "step": 4374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.805301908254455e-08,
+      "loss": 0.4199,
+      "step": 4375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.7446984928753984e-08,
+      "loss": 0.4209,
+      "step": 4376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.684580640227586e-08,
+      "loss": 0.4214,
+      "step": 4377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6249483796116924e-08,
+      "loss": 0.4206,
+      "step": 4378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.565801740092023e-08,
+      "loss": 0.4217,
+      "step": 4379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.50714075049563e-08,
+      "loss": 0.4148,
+      "step": 4380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4489654394134206e-08,
+      "loss": 0.4301,
+      "step": 4381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.391275835199159e-08,
+      "loss": 0.4207,
+      "step": 4382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3340719659701315e-08,
+      "loss": 0.4236,
+      "step": 4383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.2773538596068134e-08,
+      "loss": 0.419,
+      "step": 4384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.22112154375287e-08,
+      "loss": 0.4235,
+      "step": 4385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.165375045815266e-08,
+      "loss": 0.4264,
+      "step": 4386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.110114392964159e-08,
+      "loss": 0.4141,
+      "step": 4387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0553396121330015e-08,
+      "loss": 0.42,
+      "step": 4388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.001050730018218e-08,
+      "loss": 0.4238,
+      "step": 4389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.947247773079753e-08,
+      "loss": 0.4128,
+      "step": 4390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8939307675402983e-08,
+      "loss": 0.4269,
+      "step": 4391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8410997393860663e-08,
+      "loss": 0.4227,
+      "step": 4392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7887547143662375e-08,
+      "loss": 0.4111,
+      "step": 4393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7368957179929602e-08,
+      "loss": 0.4246,
+      "step": 4394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6855227755419046e-08,
+      "loss": 0.4246,
+      "step": 4395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6346359120514863e-08,
+      "loss": 0.4288,
+      "step": 4396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.584235152323422e-08,
+      "loss": 0.4248,
+      "step": 4397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5343205209225062e-08,
+      "loss": 0.4317,
+      "step": 4398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.484892042176279e-08,
+      "loss": 0.4291,
+      "step": 4399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4359497401758026e-08,
+      "loss": 0.42,
+      "step": 4400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3874936387747738e-08,
+      "loss": 0.4208,
+      "step": 4401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.339523761590301e-08,
+      "loss": 0.4191,
+      "step": 4402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.292040132002238e-08,
+      "loss": 0.4143,
+      "step": 4403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2450427731534052e-08,
+      "loss": 0.424,
+      "step": 4404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1985317079500358e-08,
+      "loss": 0.4213,
+      "step": 4405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.152506959060774e-08,
+      "loss": 0.4174,
+      "step": 4406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1069685489176762e-08,
+      "loss": 0.4204,
+      "step": 4407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.061916499715544e-08,
+      "loss": 0.414,
+      "step": 4408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.017350833412146e-08,
+      "loss": 0.4234,
+      "step": 4409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.973271571728441e-08,
+      "loss": 0.4216,
+      "step": 4410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.929678736148022e-08,
+      "loss": 0.4221,
+      "step": 4411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.886572347917337e-08,
+      "loss": 0.4217,
+      "step": 4412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8439524280462474e-08,
+      "loss": 0.4115,
+      "step": 4413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8018189973069144e-08,
+      "loss": 0.4294,
+      "step": 4414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7601720762346895e-08,
+      "loss": 0.4262,
+      "step": 4415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7190116851280024e-08,
+      "loss": 0.4196,
+      "step": 4416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.678337844047695e-08,
+      "loss": 0.4122,
+      "step": 4417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6381505728176872e-08,
+      "loss": 0.4171,
+      "step": 4418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5984498910249778e-08,
+      "loss": 0.4248,
+      "step": 4419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5592358180189782e-08,
+      "loss": 0.4297,
+      "step": 4420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5205083729122883e-08,
+      "loss": 0.423,
+      "step": 4421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.482267574580143e-08,
+      "loss": 0.4216,
+      "step": 4422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.4445134416607442e-08,
+      "loss": 0.4158,
+      "step": 4423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.4072459925548176e-08,
+      "loss": 0.4209,
+      "step": 4424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.370465245426167e-08,
+      "loss": 0.4236,
+      "step": 4425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3341712182012301e-08,
+      "loss": 0.424,
+      "step": 4426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2983639285693018e-08,
+      "loss": 0.4162,
+      "step": 4427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2630433939825326e-08,
+      "loss": 0.4122,
+      "step": 4428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2282096316554858e-08,
+      "loss": 0.411,
+      "step": 4429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1938626585660252e-08,
+      "loss": 0.4246,
+      "step": 4430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1600024914540931e-08,
+      "loss": 0.4113,
+      "step": 4431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.126629146822933e-08,
+      "loss": 0.4171,
+      "step": 4432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0937426409384223e-08,
+      "loss": 0.4239,
+      "step": 4433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0613429898287397e-08,
+      "loss": 0.4193,
+      "step": 4434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0294302092853647e-08,
+      "loss": 0.418,
+      "step": 4435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.980043148619668e-09,
+      "loss": 0.4124,
+      "step": 4436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.670653218752935e-09,
+      "loss": 0.4213,
+      "step": 4437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.366132454046162e-09,
+      "loss": 0.4112,
+      "step": 4438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.066481002918403e-09,
+      "loss": 0.4181,
+      "step": 4439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.771699011416169e-09,
+      "loss": 0.4198,
+      "step": 4440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.481786623214527e-09,
+      "loss": 0.4214,
+      "step": 4441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.196743979610455e-09,
+      "loss": 0.4208,
+      "step": 4442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.916571219531711e-09,
+      "loss": 0.4162,
+      "step": 4443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.641268479531283e-09,
+      "loss": 0.4179,
+      "step": 4444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.370835893788508e-09,
+      "loss": 0.4222,
+      "step": 4445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.105273594107953e-09,
+      "loss": 0.4263,
+      "step": 4446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.844581709921639e-09,
+      "loss": 0.4125,
+      "step": 4447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.588760368287928e-09,
+      "loss": 0.4149,
+      "step": 4448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.3378096938915276e-09,
+      "loss": 0.4213,
+      "step": 4449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.091729809042379e-09,
+      "loss": 0.4195,
+      "step": 4450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.850520833676765e-09,
+      "loss": 0.4236,
+      "step": 4451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.614182885357311e-09,
+      "loss": 0.4259,
+      "step": 4452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.382716079271877e-09,
+      "loss": 0.4251,
+      "step": 4453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.156120528233555e-09,
+      "loss": 0.4269,
+      "step": 4454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9343963426840006e-09,
+      "loss": 0.4215,
+      "step": 4455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.717543630688992e-09,
+      "loss": 0.4168,
+      "step": 4456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.505562497938431e-09,
+      "loss": 0.4194,
+      "step": 4457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.298453047749674e-09,
+      "loss": 0.4227,
+      "step": 4458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.096215381066415e-09,
+      "loss": 0.4173,
+      "step": 4459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.898849596456477e-09,
+      "loss": 0.4343,
+      "step": 4460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.7063557901129144e-09,
+      "loss": 0.4199,
+      "step": 4461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.518734055855122e-09,
+      "loss": 0.428,
+      "step": 4462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3359844851277302e-09,
+      "loss": 0.4114,
+      "step": 4463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.1581071670006013e-09,
+      "loss": 0.4183,
+      "step": 4464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9851021881688314e-09,
+      "loss": 0.4275,
+      "step": 4465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8169696329527484e-09,
+      "loss": 0.4192,
+      "step": 4466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6537095832990247e-09,
+      "loss": 0.4246,
+      "step": 4467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.495322118778454e-09,
+      "loss": 0.412,
+      "step": 4468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.341807316587064e-09,
+      "loss": 0.4144,
+      "step": 4469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.193165251545004e-09,
+      "loss": 0.4156,
+      "step": 4470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.049395996099879e-09,
+      "loss": 0.4256,
+      "step": 4471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.910499620322304e-09,
+      "loss": 0.4167,
+      "step": 4472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.776476191910348e-09,
+      "loss": 0.4222,
+      "step": 4473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.647325776182873e-09,
+      "loss": 0.4251,
+      "step": 4474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5230484360873043e-09,
+      "loss": 0.4244,
+      "step": 4475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.4036442321962995e-09,
+      "loss": 0.4237,
+      "step": 4476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2891132227033087e-09,
+      "loss": 0.4242,
+      "step": 4477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794554634314558e-09,
+      "loss": 0.4234,
+      "step": 4478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0746710078257673e-09,
+      "loss": 0.4176,
+      "step": 4479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.74759906957612e-10,
+      "loss": 0.4152,
+      "step": 4480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.797222095224822e-10,
+      "loss": 0.4277,
+      "step": 4481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.895579618388827e-10,
+      "loss": 0.4113,
+      "step": 4482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.042672078527712e-10,
+      "loss": 0.4092,
+      "step": 4483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.238499891353389e-10,
+      "loss": 0.4275,
+      "step": 4484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.483063448785686e-10,
+      "loss": 0.422,
+      "step": 4485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.77636311903007e-10,
+      "loss": 0.4337,
+      "step": 4486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.118399246522131e-10,
+      "loss": 0.4207,
+      "step": 4487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.509172151938689e-10,
+      "loss": 0.4167,
+      "step": 4488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.948682132208891e-10,
+      "loss": 0.4201,
+      "step": 4489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.436929460525317e-10,
+      "loss": 0.4308,
+      "step": 4490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9739143862884668e-10,
+      "loss": 0.4237,
+      "step": 4491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.559637135173375e-10,
+      "loss": 0.4185,
+      "step": 4492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1940979091074056e-10,
+      "loss": 0.4183,
+      "step": 4493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.772968862369447e-11,
+      "loss": 0.4139,
+      "step": 4494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.092342209607083e-11,
+      "loss": 0.4208,
+      "step": 4495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.899100439408443e-11,
+      "loss": 0.4186,
+      "step": 4496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1932446206962556e-11,
+      "loss": 0.4174,
+      "step": 4497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.74775584916543e-12,
+      "loss": 0.4201,
+      "step": 4498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.4369392592760166e-12,
+      "loss": 0.4172,
+      "step": 4499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0,
+      "loss": 0.4197,
+      "step": 4500
+    },
+    {
+      "epoch": 0.46,
+      "step": 4500,
+      "total_flos": 6.639877034606592e+18,
+      "train_loss": 0.48130192704333197,
+      "train_runtime": 48519.7598,
+      "train_samples_per_second": 23.743,
+      "train_steps_per_second": 0.093
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 10000,
+  "total_flos": 6.639877034606592e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}