diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.984988524285094,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.4558,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999987833918285e-05,
+      "loss": 1.4479,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999951335702733e-05,
+      "loss": 1.3372,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999890505442158e-05,
+      "loss": 1.3914,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999980534328457e-05,
+      "loss": 1.4222,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999969584943719e-05,
+      "loss": 1.4072,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999956202416644e-05,
+      "loss": 1.3195,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999940386779794e-05,
+      "loss": 1.348,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999221380716526e-05,
+      "loss": 1.3456,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999014563366226e-05,
+      "loss": 1.3545,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999878341625027e-05,
+      "loss": 1.2812,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998527939931088e-05,
+      "loss": 1.3375,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998248135030315e-05,
+      "loss": 1.323,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999794400222877e-05,
+      "loss": 1.4263,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997615542266482e-05,
+      "loss": 1.3663,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997262755942653e-05,
+      "loss": 1.3438,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99968856441157e-05,
+      "loss": 1.2855,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999648420770321e-05,
+      "loss": 1.2576,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996058447681973e-05,
+      "loss": 1.2811,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995608365087945e-05,
+      "loss": 1.2616,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999513396101628e-05,
+      "loss": 1.2319,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994635236621306e-05,
+      "loss": 1.3689,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994112193116527e-05,
+      "loss": 1.1436,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999356483177462e-05,
+      "loss": 1.2203,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992993153927432e-05,
+      "loss": 1.1449,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992397160965984e-05,
+      "loss": 1.2228,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999177685434045e-05,
+      "loss": 1.2316,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991132235560174e-05,
+      "loss": 1.3092,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990463306193652e-05,
+      "loss": 1.2577,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989770067868536e-05,
+      "loss": 1.1647,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998905252227162e-05,
+      "loss": 1.1908,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998831067114885e-05,
+      "loss": 1.314,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987544516305313e-05,
+      "loss": 1.1445,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998675405960522e-05,
+      "loss": 1.1633,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985939302971936e-05,
+      "loss": 1.2699,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985100248387936e-05,
+      "loss": 1.2089,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984236897894815e-05,
+      "loss": 1.2043,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9983349253593302e-05,
+      "loss": 1.2562,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9982437317643218e-05,
+      "loss": 1.2421,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9981501092263502e-05,
+      "loss": 1.2917,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9980540579732198e-05,
+      "loss": 1.3397,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9979555782386435e-05,
+      "loss": 1.2327,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9978546702622443e-05,
+      "loss": 1.2871,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.997751334289553e-05,
+      "loss": 1.2185,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9976455705720084e-05,
+      "loss": 1.3285,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997537379366956e-05,
+      "loss": 1.2261,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974267609376494e-05,
+      "loss": 1.2092,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973137155532462e-05,
+      "loss": 1.1791,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971982434888106e-05,
+      "loss": 1.2968,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970803450253116e-05,
+      "loss": 1.24,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996960020449621e-05,
+      "loss": 1.2806,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968372700545145e-05,
+      "loss": 1.2578,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967120941386707e-05,
+      "loss": 1.2295,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99658449300667e-05,
+      "loss": 1.1964,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996454466968993e-05,
+      "loss": 1.2623,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963220163420215e-05,
+      "loss": 1.1747,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996187141448036e-05,
+      "loss": 1.1561,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996049842615217e-05,
+      "loss": 1.2927,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995910120177642e-05,
+      "loss": 1.2256,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957679744752858e-05,
+      "loss": 1.2709,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956234058540195e-05,
+      "loss": 1.2031,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954764146656103e-05,
+      "loss": 1.1953,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953270012677195e-05,
+      "loss": 1.1442,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951751660239015e-05,
+      "loss": 1.1697,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995020909303605e-05,
+      "loss": 1.2153,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99486423148217e-05,
+      "loss": 1.1575,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9947051329408274e-05,
+      "loss": 1.2231,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9945436140666984e-05,
+      "loss": 1.171,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9943796752527932e-05,
+      "loss": 1.2231,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9942133168980105e-05,
+      "loss": 1.2854,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9940445394071358e-05,
+      "loss": 1.2368,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993873343190842e-05,
+      "loss": 1.2302,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936997286656855e-05,
+      "loss": 1.2982,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935236962541092e-05,
+      "loss": 1.2573,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933452463844377e-05,
+      "loss": 1.2922,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9931643794908774e-05,
+      "loss": 1.1661,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992981096013517e-05,
+      "loss": 1.2003,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992795396398325e-05,
+      "loss": 1.1262,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9926072810971494e-05,
+      "loss": 1.1949,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992416750567714e-05,
+      "loss": 1.1951,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9922238052736215e-05,
+      "loss": 1.2822,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99202844568435e-05,
+      "loss": 1.1666,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9918306722752507e-05,
+      "loss": 1.0392,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9916304855275498e-05,
+      "loss": 1.2428,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9914278859283446e-05,
+      "loss": 1.2222,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.991222873970604e-05,
+      "loss": 1.1914,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9910154501531666e-05,
+      "loss": 1.1314,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990805614980739e-05,
+      "loss": 1.1499,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9905933689638958e-05,
+      "loss": 1.1887,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9903787126190774e-05,
+      "loss": 1.302,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990161646468589e-05,
+      "loss": 1.2144,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9899421710405996e-05,
+      "loss": 1.2339,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.989720286869141e-05,
+      "loss": 1.1867,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9894959944941038e-05,
+      "loss": 1.259,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.989269294461242e-05,
+      "loss": 1.2476,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9890401873221642e-05,
+      "loss": 1.1847,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9888086736343385e-05,
+      "loss": 1.2072,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.988574753961087e-05,
+      "loss": 1.17,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9883384288715876e-05,
+      "loss": 1.1533,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.98809969894087e-05,
+      "loss": 1.0659,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.987858564749816e-05,
+      "loss": 1.1814,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9876150268851572e-05,
+      "loss": 1.1722,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9873690859394738e-05,
+      "loss": 1.2267,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.987120742511193e-05,
+      "loss": 1.1396,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9868699972045892e-05,
+      "loss": 1.0883,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986616850629779e-05,
+      "loss": 1.0839,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 1.2437,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9861033561452223e-05,
+      "loss": 1.1318,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9858430094849196e-05,
+      "loss": 1.1162,
+      "step": 218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9855802640552938e-05,
+      "loss": 1.0525,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9853151204956617e-05,
+      "loss": 1.136,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.985047579451175e-05,
+      "loss": 1.0522,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9847776415728186e-05,
+      "loss": 1.2369,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9845053075174102e-05,
+      "loss": 1.1296,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984230577947597e-05,
+      "loss": 1.2401,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.983953453531856e-05,
+      "loss": 1.2585,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.98367393494449e-05,
+      "loss": 1.1647,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9833920228656295e-05,
+      "loss": 1.1434,
+      "step": 236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.983107717981226e-05,
+      "loss": 1.0576,
+      "step": 238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9828210209830564e-05,
+      "loss": 1.124,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9825319325687157e-05,
+      "loss": 1.129,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9822404534416183e-05,
+      "loss": 1.0831,
+      "step": 244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9819465843109965e-05,
+      "loss": 1.1666,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981650325891897e-05,
+      "loss": 1.1545,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981351678905181e-05,
+      "loss": 1.1523,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.981050644077521e-05,
+      "loss": 1.178,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9807472221414002e-05,
+      "loss": 1.1373,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9804414138351097e-05,
+      "loss": 1.2085,
+      "step": 256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9801332199027468e-05,
+      "loss": 1.1745,
+      "step": 258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9798226410942148e-05,
+      "loss": 1.1953,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9795096781652184e-05,
+      "loss": 1.1445,
+      "step": 262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9791943318772644e-05,
+      "loss": 1.2227,
+      "step": 264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9788766029976587e-05,
+      "loss": 1.1688,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9785564922995042e-05,
+      "loss": 1.1893,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9782340005616995e-05,
+      "loss": 1.1779,
+      "step": 270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.977909128568937e-05,
+      "loss": 1.1489,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9775818771117e-05,
+      "loss": 1.1362,
+      "step": 274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9772522469862628e-05,
+      "loss": 1.1897,
+      "step": 276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9769202389946863e-05,
+      "loss": 1.167,
+      "step": 278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.976585853944818e-05,
+      "loss": 1.1824,
+      "step": 280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.976249092650289e-05,
+      "loss": 1.1413,
+      "step": 282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9759099559305124e-05,
+      "loss": 1.1875,
+      "step": 284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9755684446106813e-05,
+      "loss": 1.1451,
+      "step": 286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9752245595217662e-05,
+      "loss": 1.0607,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9748783015005144e-05,
+      "loss": 1.2404,
+      "step": 290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9745296713894463e-05,
+      "loss": 1.1807,
+      "step": 292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.974178670036855e-05,
+      "loss": 1.0758,
+      "step": 294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.973825298296802e-05,
+      "loss": 1.0989,
+      "step": 296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.973469557029117e-05,
+      "loss": 1.1169,
+      "step": 298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.973111447099396e-05,
+      "loss": 1.1579,
+      "step": 300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.972750969378998e-05,
+      "loss": 1.1099,
+      "step": 302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9723881247450434e-05,
+      "loss": 1.2141,
+      "step": 304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.972022914080411e-05,
+      "loss": 1.2618,
+      "step": 306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.971655338273738e-05,
+      "loss": 1.2097,
+      "step": 308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9712853982194154e-05,
+      "loss": 1.2092,
+      "step": 310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9709130948175877e-05,
+      "loss": 1.0878,
+      "step": 312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9705384289741493e-05,
+      "loss": 1.1637,
+      "step": 314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9701614016007435e-05,
+      "loss": 1.1152,
+      "step": 316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9697820136147598e-05,
+      "loss": 1.0662,
+      "step": 318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9694002659393306e-05,
+      "loss": 1.2159,
+      "step": 320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.969016159503331e-05,
+      "loss": 1.1588,
+      "step": 322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9686296952413747e-05,
+      "loss": 1.1617,
+      "step": 324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9682408740938132e-05,
+      "loss": 1.1422,
+      "step": 326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9678496970067326e-05,
+      "loss": 1.1713,
+      "step": 328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.967456164931951e-05,
+      "loss": 1.0971,
+      "step": 330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9670602788270172e-05,
+      "loss": 1.1238,
+      "step": 332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9666620396552075e-05,
+      "loss": 1.2025,
+      "step": 334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9662614483855247e-05,
+      "loss": 1.0267,
+      "step": 336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9658585059926935e-05,
+      "loss": 1.1763,
+      "step": 338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9654532134571594e-05,
+      "loss": 1.1421,
+      "step": 340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9650455717650878e-05,
+      "loss": 1.1594,
+      "step": 342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.964635581908359e-05,
+      "loss": 1.1858,
+      "step": 344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9642232448845663e-05,
+      "loss": 1.2112,
+      "step": 346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9638085616970152e-05,
+      "loss": 1.1302,
+      "step": 348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9633915333547202e-05,
+      "loss": 1.1993,
+      "step": 350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9629721608724006e-05,
+      "loss": 1.0864,
+      "step": 352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.962550445270481e-05,
+      "loss": 1.0865,
+      "step": 354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9621263875750864e-05,
+      "loss": 1.0852,
+      "step": 356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9616999888180408e-05,
+      "loss": 1.2186,
+      "step": 358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9612712500368653e-05,
+      "loss": 1.092,
+      "step": 360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.960840172274773e-05,
+      "loss": 1.1371,
+      "step": 362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9604067565806707e-05,
+      "loss": 1.1527,
+      "step": 364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.959971004009151e-05,
+      "loss": 1.1696,
+      "step": 366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9595329156204955e-05,
+      "loss": 1.1942,
+      "step": 368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9590924924806677e-05,
+      "loss": 1.2083,
+      "step": 370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.958649735661312e-05,
+      "loss": 1.1935,
+      "step": 372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9582046462397514e-05,
+      "loss": 1.1874,
+      "step": 374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9577572252989856e-05,
+      "loss": 1.1034,
+      "step": 376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9573074739276858e-05,
+      "loss": 1.175,
+      "step": 378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9568553932201947e-05,
+      "loss": 1.1102,
+      "step": 380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9564009842765225e-05,
+      "loss": 1.1398,
+      "step": 382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9559442482023444e-05,
+      "loss": 1.0146,
+      "step": 384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.955485186108998e-05,
+      "loss": 1.1453,
+      "step": 386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9550237991134806e-05,
+      "loss": 1.1453,
+      "step": 388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.954560088338447e-05,
+      "loss": 1.1533,
+      "step": 390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.954094054912205e-05,
+      "loss": 1.0356,
+      "step": 392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9536256999687157e-05,
+      "loss": 1.0178,
+      "step": 394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9531550246475875e-05,
+      "loss": 1.087,
+      "step": 396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9526820300940756e-05,
+      "loss": 1.0517,
+      "step": 398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9522067174590777e-05,
+      "loss": 1.0938,
+      "step": 400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9517290878991324e-05,
+      "loss": 1.0381,
+      "step": 402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.951249142576416e-05,
+      "loss": 1.0585,
+      "step": 404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.950766882658739e-05,
+      "loss": 1.1217,
+      "step": 406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.950282309319544e-05,
+      "loss": 1.2144,
+      "step": 408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.949795423737903e-05,
+      "loss": 1.1815,
+      "step": 410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9493062270985143e-05,
+      "loss": 1.167,
+      "step": 412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9488147205916985e-05,
+      "loss": 1.1621,
+      "step": 414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9483209054133975e-05,
+      "loss": 1.1437,
+      "step": 416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.947824782765171e-05,
+      "loss": 1.1501,
+      "step": 418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9473263538541916e-05,
+      "loss": 1.1792,
+      "step": 420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9468256198932455e-05,
+      "loss": 1.1888,
+      "step": 422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.946322582100727e-05,
+      "loss": 1.2405,
+      "step": 424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 1.1984,
+      "step": 426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9453095999225725e-05,
+      "loss": 1.1406,
+      "step": 428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9447996580017418e-05,
+      "loss": 1.2371,
+      "step": 430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9442874171789417e-05,
+      "loss": 1.0566,
+      "step": 432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9437728787005655e-05,
+      "loss": 1.0957,
+      "step": 434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9432560438185964e-05,
+      "loss": 1.1853,
+      "step": 436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9427369137906046e-05,
+      "loss": 1.0938,
+      "step": 438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.942215489879747e-05,
+      "loss": 1.1975,
+      "step": 440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9416917733547605e-05,
+      "loss": 1.1953,
+      "step": 442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9411657654899598e-05,
+      "loss": 1.1901,
+      "step": 444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.940637467565237e-05,
+      "loss": 1.1848,
+      "step": 446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9401068808660544e-05,
+      "loss": 1.1822,
+      "step": 448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.939574006683445e-05,
+      "loss": 1.1888,
+      "step": 450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9390388463140065e-05,
+      "loss": 1.1284,
+      "step": 452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9385014010598998e-05,
+      "loss": 1.1155,
+      "step": 454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9379616722288457e-05,
+      "loss": 1.1359,
+      "step": 456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9374196611341212e-05,
+      "loss": 1.2017,
+      "step": 458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.936875369094556e-05,
+      "loss": 1.1825,
+      "step": 460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9363287974345312e-05,
+      "loss": 1.0658,
+      "step": 462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9357799474839735e-05,
+      "loss": 1.1118,
+      "step": 464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.935228820578354e-05,
+      "loss": 1.1981,
+      "step": 466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9346754180586825e-05,
+      "loss": 1.1582,
+      "step": 468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.934119741271508e-05,
+      "loss": 1.0506,
+      "step": 470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.933561791568913e-05,
+      "loss": 1.1329,
+      "step": 472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9330015703085082e-05,
+      "loss": 1.1007,
+      "step": 474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9324390788534345e-05,
+      "loss": 1.064,
+      "step": 476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9318743185723544e-05,
+      "loss": 1.084,
+      "step": 478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9313072908394524e-05,
+      "loss": 1.148,
+      "step": 480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9307379970344295e-05,
+      "loss": 1.0327,
+      "step": 482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9301664385425006e-05,
+      "loss": 1.0734,
+      "step": 484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9295926167543914e-05,
+      "loss": 1.1937,
+      "step": 486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9290165330663338e-05,
+      "loss": 1.1191,
+      "step": 488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9284381888800646e-05,
+      "loss": 1.1106,
+      "step": 490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9278575856028207e-05,
+      "loss": 0.9996,
+      "step": 492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9272747246473345e-05,
+      "loss": 1.1063,
+      "step": 494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9266896074318335e-05,
+      "loss": 1.0836,
+      "step": 496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9261022353800344e-05,
+      "loss": 1.0874,
+      "step": 498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9255126099211403e-05,
+      "loss": 1.1903,
+      "step": 500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9249207324898377e-05,
+      "loss": 1.2297,
+      "step": 502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.924326604526292e-05,
+      "loss": 1.1173,
+      "step": 504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9237302274761457e-05,
+      "loss": 1.1311,
+      "step": 506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9231316027905128e-05,
+      "loss": 1.2332,
+      "step": 508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.922530731925977e-05,
+      "loss": 0.9706,
+      "step": 510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9219276163445865e-05,
+      "loss": 1.1958,
+      "step": 512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.921322257513852e-05,
+      "loss": 1.0844,
+      "step": 514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9207146569067438e-05,
+      "loss": 1.0564,
+      "step": 516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.920104816001684e-05,
+      "loss": 1.1932,
+      "step": 518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.919492736282548e-05,
+      "loss": 1.0589,
+      "step": 520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9188784192386588e-05,
+      "loss": 1.1334,
+      "step": 522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9182618663647815e-05,
+      "loss": 1.1622,
+      "step": 524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.917643079161124e-05,
+      "loss": 1.1118,
+      "step": 526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9170220591333283e-05,
+      "loss": 1.0058,
+      "step": 528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9163988077924712e-05,
+      "loss": 1.1155,
+      "step": 530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9157733266550577e-05,
+      "loss": 1.078,
+      "step": 532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9151456172430186e-05,
+      "loss": 1.0551,
+      "step": 534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9145156810837072e-05,
+      "loss": 1.1194,
+      "step": 536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.913883519709894e-05,
+      "loss": 1.1368,
+      "step": 538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9132491346597644e-05,
+      "loss": 1.2772,
+      "step": 540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9126125274769148e-05,
+      "loss": 1.2117,
+      "step": 542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9119736997103476e-05,
+      "loss": 1.1841,
+      "step": 544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91133265291447e-05,
+      "loss": 1.0742,
+      "step": 546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9106893886490865e-05,
+      "loss": 1.0796,
+      "step": 548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910043908479399e-05,
+      "loss": 1.0079,
+      "step": 550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909396213976e-05,
+      "loss": 1.088,
+      "step": 552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9087463067148698e-05,
+      "loss": 1.0243,
+      "step": 554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9080941882773744e-05,
+      "loss": 1.2952,
+      "step": 556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9074398602502585e-05,
+      "loss": 1.1602,
+      "step": 558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.906783324225644e-05,
+      "loss": 1.1614,
+      "step": 560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.906124581801025e-05,
+      "loss": 1.1066,
+      "step": 562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.905463634579264e-05,
+      "loss": 1.115,
+      "step": 564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9048004841685888e-05,
+      "loss": 1.1365,
+      "step": 566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9041351321825882e-05,
+      "loss": 1.1183,
+      "step": 568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.903467580240207e-05,
+      "loss": 1.1091,
+      "step": 570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9027978299657436e-05,
+      "loss": 1.1599,
+      "step": 572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9021258829888457e-05,
+      "loss": 1.225,
+      "step": 574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9014517409445053e-05,
+      "loss": 1.0121,
+      "step": 576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9007754054730555e-05,
+      "loss": 1.0556,
+      "step": 578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9000968782201678e-05,
+      "loss": 1.0731,
+      "step": 580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.899416160836845e-05,
+      "loss": 1.1136,
+      "step": 582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8987332549794196e-05,
+      "loss": 1.0963,
+      "step": 584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8980481623095502e-05,
+      "loss": 1.1663,
+      "step": 586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8973608844942147e-05,
+      "loss": 1.1445,
+      "step": 588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8966714232057094e-05,
+      "loss": 1.1259,
+      "step": 590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.895979780121642e-05,
+      "loss": 1.1379,
+      "step": 592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.89528595692493e-05,
+      "loss": 1.0709,
+      "step": 594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8945899553037957e-05,
+      "loss": 1.0895,
+      "step": 596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8938917769517614e-05,
+      "loss": 1.0493,
+      "step": 598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8931914235676457e-05,
+      "loss": 0.9594,
+      "step": 600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.892488896855561e-05,
+      "loss": 1.1143,
+      "step": 602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8917841985249055e-05,
+      "loss": 1.2822,
+      "step": 604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8910773302903632e-05,
+      "loss": 1.0581,
+      "step": 606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8903682938718976e-05,
+      "loss": 1.1096,
+      "step": 608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8896570909947477e-05,
+      "loss": 1.1439,
+      "step": 610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8889437233894234e-05,
+      "loss": 1.0356,
+      "step": 612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.888228192791703e-05,
+      "loss": 1.0795,
+      "step": 614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8875105009426272e-05,
+      "loss": 1.1792,
+      "step": 616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8867906495884955e-05,
+      "loss": 1.1395,
+      "step": 618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8860686404808618e-05,
+      "loss": 1.0511,
+      "step": 620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8853444753765308e-05,
+      "loss": 1.0585,
+      "step": 622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8846181560375525e-05,
+      "loss": 1.1337,
+      "step": 624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.883889684231219e-05,
+      "loss": 1.0908,
+      "step": 626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.88315906173006e-05,
+      "loss": 1.1562,
+      "step": 628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.882426290311838e-05,
+      "loss": 0.9822,
+      "step": 630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8816913717595445e-05,
+      "loss": 1.0973,
+      "step": 632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8809543078613956e-05,
+      "loss": 1.0825,
+      "step": 634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8802151004108266e-05,
+      "loss": 1.1721,
+      "step": 636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 1.0896,
+      "step": 638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8787302620522467e-05,
+      "loss": 1.1183,
+      "step": 640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8779846347571695e-05,
+      "loss": 1.1766,
+      "step": 642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.877236871135529e-05,
+      "loss": 1.08,
+      "step": 644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.876486973006797e-05,
+      "loss": 1.1069,
+      "step": 646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.875734942195637e-05,
+      "loss": 1.1172,
+      "step": 648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.874980780531903e-05,
+      "loss": 1.0455,
+      "step": 650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.874224489850634e-05,
+      "loss": 1.2435,
+      "step": 652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8734660719920477e-05,
+      "loss": 1.1201,
+      "step": 654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.87270552880154e-05,
+      "loss": 0.9779,
+      "step": 656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8719428621296765e-05,
+      "loss": 1.1465,
+      "step": 658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8711780738321897e-05,
+      "loss": 1.1176,
+      "step": 660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8704111657699758e-05,
+      "loss": 1.1217,
+      "step": 662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.869642139809088e-05,
+      "loss": 1.0776,
+      "step": 664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8688709978207323e-05,
+      "loss": 1.02,
+      "step": 666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8680977416812645e-05,
+      "loss": 1.0624,
+      "step": 668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.867322373272184e-05,
+      "loss": 1.1606,
+      "step": 670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.866544894480129e-05,
+      "loss": 1.1472,
+      "step": 672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8657653071968748e-05,
+      "loss": 1.1931,
+      "step": 674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.864983613319325e-05,
+      "loss": 1.1593,
+      "step": 676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8641998147495114e-05,
+      "loss": 1.0349,
+      "step": 678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8634139133945838e-05,
+      "loss": 1.0645,
+      "step": 680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8626259111668107e-05,
+      "loss": 1.0791,
+      "step": 682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8618358099835724e-05,
+      "loss": 1.024,
+      "step": 684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8610436117673557e-05,
+      "loss": 1.0797,
+      "step": 686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8602493184457505e-05,
+      "loss": 1.1108,
+      "step": 688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8594529319514437e-05,
+      "loss": 1.1057,
+      "step": 690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.858654454222217e-05,
+      "loss": 1.0278,
+      "step": 692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8578538872009383e-05,
+      "loss": 0.9993,
+      "step": 694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8570512328355613e-05,
+      "loss": 1.05,
+      "step": 696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.856246493079117e-05,
+      "loss": 1.0813,
+      "step": 698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8554396698897116e-05,
+      "loss": 1.0712,
+      "step": 700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8546307652305207e-05,
+      "loss": 1.0891,
+      "step": 702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8538197810697843e-05,
+      "loss": 1.0363,
+      "step": 704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.853006719380802e-05,
+      "loss": 1.0649,
+      "step": 706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8521915821419285e-05,
+      "loss": 1.156,
+      "step": 708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.85137437133657e-05,
+      "loss": 1.1517,
+      "step": 710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8505550889531766e-05,
+      "loss": 1.0265,
+      "step": 712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8497337369852397e-05,
+      "loss": 1.1985,
+      "step": 714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.848910317431286e-05,
+      "loss": 1.0552,
+      "step": 716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.848084832294874e-05,
+      "loss": 1.253,
+      "step": 718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.847257283584587e-05,
+      "loss": 1.1016,
+      "step": 720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8464276733140307e-05,
+      "loss": 1.1403,
+      "step": 722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.845596003501826e-05,
+      "loss": 1.2197,
+      "step": 724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8447622761716058e-05,
+      "loss": 0.9938,
+      "step": 726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8439264933520086e-05,
+      "loss": 1.0355,
+      "step": 728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.843088657076675e-05,
+      "loss": 1.1461,
+      "step": 730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.842248769384242e-05,
+      "loss": 1.0829,
+      "step": 732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8414068323183378e-05,
+      "loss": 1.1194,
+      "step": 734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8405628479275777e-05,
+      "loss": 1.0746,
+      "step": 736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8397168182655584e-05,
+      "loss": 1.0358,
+      "step": 738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8388687453908527e-05,
+      "loss": 1.1372,
+      "step": 740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.838018631367006e-05,
+      "loss": 1.2092,
+      "step": 742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8371664782625287e-05,
+      "loss": 1.0762,
+      "step": 744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8363122881508947e-05,
+      "loss": 1.0928,
+      "step": 746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8354560631105326e-05,
+      "loss": 1.2407,
+      "step": 748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8345978052248235e-05,
+      "loss": 1.1196,
+      "step": 750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8337375165820943e-05,
+      "loss": 1.0321,
+      "step": 752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.832875199275614e-05,
+      "loss": 1.1741,
+      "step": 754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.832010855403586e-05,
+      "loss": 1.1316,
+      "step": 756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.831144487069147e-05,
+      "loss": 1.0432,
+      "step": 758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8302760963803582e-05,
+      "loss": 1.1426,
+      "step": 760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.829405685450202e-05,
+      "loss": 1.0476,
+      "step": 762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8285332563965765e-05,
+      "loss": 1.0797,
+      "step": 764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8276588113422903e-05,
+      "loss": 0.9968,
+      "step": 766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8267823524150575e-05,
+      "loss": 1.1027,
+      "step": 768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8259038817474923e-05,
+      "loss": 1.0366,
+      "step": 770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.825023401477104e-05,
+      "loss": 1.1276,
+      "step": 772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.824140913746291e-05,
+      "loss": 1.0907,
+      "step": 774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8232564207023377e-05,
+      "loss": 1.1748,
+      "step": 776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8223699244974065e-05,
+      "loss": 1.139,
+      "step": 778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8214814272885344e-05,
+      "loss": 1.0471,
+      "step": 780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8205909312376277e-05,
+      "loss": 1.0496,
+      "step": 782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8196984385114556e-05,
+      "loss": 1.1727,
+      "step": 784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.818803951281646e-05,
+      "loss": 1.0343,
+      "step": 786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.81790747172468e-05,
+      "loss": 1.0835,
+      "step": 788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8170090020218866e-05,
+      "loss": 1.0961,
+      "step": 790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8161085443594367e-05,
+      "loss": 1.0955,
+      "step": 792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.815206100928338e-05,
+      "loss": 1.1442,
+      "step": 794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8143016739244314e-05,
+      "loss": 1.1664,
+      "step": 796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8133952655483834e-05,
+      "loss": 1.045,
+      "step": 798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8124868780056814e-05,
+      "loss": 1.1095,
+      "step": 800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.811576513506629e-05,
+      "loss": 1.0906,
+      "step": 802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8106641742663397e-05,
+      "loss": 1.0254,
+      "step": 804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.809749862504733e-05,
+      "loss": 1.1217,
+      "step": 806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.808833580446526e-05,
+      "loss": 1.1943,
+      "step": 808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.807915330321232e-05,
+      "loss": 1.1309,
+      "step": 810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.806995114363152e-05,
+      "loss": 1.1422,
+      "step": 812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8060729348113706e-05,
+      "loss": 1.0663,
+      "step": 814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8051487939097506e-05,
+      "loss": 1.1041,
+      "step": 816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8042226939069256e-05,
+      "loss": 1.1333,
+      "step": 818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8032946370562983e-05,
+      "loss": 1.1051,
+      "step": 820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8023646256160314e-05,
+      "loss": 1.051,
+      "step": 822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.801432661849044e-05,
+      "loss": 1.0166,
+      "step": 824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.800498748023005e-05,
+      "loss": 1.0703,
+      "step": 826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.79956288641033e-05,
+      "loss": 1.0842,
+      "step": 828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.798625079288172e-05,
+      "loss": 1.1841,
+      "step": 830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7976853289384184e-05,
+      "loss": 1.1714,
+      "step": 832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7967436376476857e-05,
+      "loss": 1.1438,
+      "step": 834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.795800007707312e-05,
+      "loss": 1.0752,
+      "step": 836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7948544414133534e-05,
+      "loss": 1.1434,
+      "step": 838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7939069410665772e-05,
+      "loss": 1.1209,
+      "step": 840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.792957508972457e-05,
+      "loss": 0.9802,
+      "step": 842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.792006147441166e-05,
+      "loss": 1.085,
+      "step": 844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.791052858787573e-05,
+      "loss": 1.2046,
+      "step": 846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.790097645331235e-05,
+      "loss": 1.1376,
+      "step": 848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 1.0742,
+      "step": 850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7881814533119678e-05,
+      "loss": 1.029,
+      "step": 852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7872204794115474e-05,
+      "loss": 1.1351,
+      "step": 854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.786257590033391e-05,
+      "loss": 1.1188,
+      "step": 856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7852927875204162e-05,
+      "loss": 1.097,
+      "step": 858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7843260742201964e-05,
+      "loss": 1.1078,
+      "step": 860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7833574524849536e-05,
+      "loss": 1.1975,
+      "step": 862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7823869246715553e-05,
+      "loss": 1.0818,
+      "step": 864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7814144931415043e-05,
+      "loss": 1.0735,
+      "step": 866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.780440160260938e-05,
+      "loss": 1.0704,
+      "step": 868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7794639284006184e-05,
+      "loss": 1.0148,
+      "step": 870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7784857999359292e-05,
+      "loss": 1.1402,
+      "step": 872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.777505777246868e-05,
+      "loss": 1.0834,
+      "step": 874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7765238627180424e-05,
+      "loss": 1.1154,
+      "step": 876
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.775540058738663e-05,
+      "loss": 1.0593,
+      "step": 878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.774554367702538e-05,
+      "loss": 1.061,
+      "step": 880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7735667920080662e-05,
+      "loss": 0.9858,
+      "step": 882
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.772577334058233e-05,
+      "loss": 1.0783,
+      "step": 884
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7715859962606045e-05,
+      "loss": 1.1001,
+      "step": 886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7705927810273188e-05,
+      "loss": 1.1298,
+      "step": 888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7695976907750846e-05,
+      "loss": 1.0907,
+      "step": 890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7686007279251708e-05,
+      "loss": 1.0962,
+      "step": 892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7676018949034045e-05,
+      "loss": 1.017,
+      "step": 894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.766601194140162e-05,
+      "loss": 1.1945,
+      "step": 896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.765598628070365e-05,
+      "loss": 1.0868,
+      "step": 898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7645941991334732e-05,
+      "loss": 0.9812,
+      "step": 900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7635879097734806e-05,
+      "loss": 1.0974,
+      "step": 902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7625797624389055e-05,
+      "loss": 1.1943,
+      "step": 904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7615697595827898e-05,
+      "loss": 1.0878,
+      "step": 906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.760557903662688e-05,
+      "loss": 1.0845,
+      "step": 908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7595441971406647e-05,
+      "loss": 1.0385,
+      "step": 910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7585286424832876e-05,
+      "loss": 1.0956,
+      "step": 912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7575112421616203e-05,
+      "loss": 0.9226,
+      "step": 914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7564919986512183e-05,
+      "loss": 0.9761,
+      "step": 916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7554709144321212e-05,
+      "loss": 1.0568,
+      "step": 918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.754447991988848e-05,
+      "loss": 1.0809,
+      "step": 920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7534232338103904e-05,
+      "loss": 1.028,
+      "step": 922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.752396642390207e-05,
+      "loss": 1.0931,
+      "step": 924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7513682202262163e-05,
+      "loss": 1.0754,
+      "step": 926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.750337969820792e-05,
+      "loss": 1.1067,
+      "step": 928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7493058936807562e-05,
+      "loss": 1.1318,
+      "step": 930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.748271994317374e-05,
+      "loss": 1.131,
+      "step": 932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7472362742463455e-05,
+      "loss": 0.9948,
+      "step": 934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7461987359878022e-05,
+      "loss": 1.0269,
+      "step": 936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7451593820662987e-05,
+      "loss": 1.1522,
+      "step": 938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7441182150108088e-05,
+      "loss": 0.9819,
+      "step": 940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.743075237354716e-05,
+      "loss": 1.0245,
+      "step": 942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7420304516358113e-05,
+      "loss": 1.0467,
+      "step": 944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7409838603962844e-05,
+      "loss": 0.9151,
+      "step": 946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.739935466182718e-05,
+      "loss": 1.0757,
+      "step": 948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.738885271546082e-05,
+      "loss": 1.1047,
+      "step": 950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7378332790417275e-05,
+      "loss": 1.0851,
+      "step": 952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7367794912293794e-05,
+      "loss": 1.0941,
+      "step": 954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.735723910673132e-05,
+      "loss": 1.0764,
+      "step": 956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7346665399414404e-05,
+      "loss": 1.0224,
+      "step": 958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7336073816071167e-05,
+      "loss": 1.1149,
+      "step": 960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7325464382473228e-05,
+      "loss": 1.1105,
+      "step": 962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7314837124435625e-05,
+      "loss": 1.1125,
+      "step": 964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7304192067816782e-05,
+      "loss": 1.0294,
+      "step": 966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7293529238518422e-05,
+      "loss": 1.1923,
+      "step": 968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.728284866248552e-05,
+      "loss": 1.107,
+      "step": 970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7272150365706225e-05,
+      "loss": 1.0266,
+      "step": 972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7261434374211804e-05,
+      "loss": 1.0624,
+      "step": 974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.725070071407659e-05,
+      "loss": 1.1012,
+      "step": 976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.723994941141789e-05,
+      "loss": 1.1323,
+      "step": 978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.722918049239596e-05,
+      "loss": 1.0443,
+      "step": 980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7218393983213902e-05,
+      "loss": 0.9999,
+      "step": 982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7207589910117634e-05,
+      "loss": 1.15,
+      "step": 984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.71967682993958e-05,
+      "loss": 1.058,
+      "step": 986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7185929177379714e-05,
+      "loss": 1.0251,
+      "step": 988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.717507257044331e-05,
+      "loss": 0.9785,
+      "step": 990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7164198505003068e-05,
+      "loss": 0.9978,
+      "step": 992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.715330700751793e-05,
+      "loss": 0.9861,
+      "step": 994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7142398104489274e-05,
+      "loss": 1.1539,
+      "step": 996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7131471822460814e-05,
+      "loss": 1.0574,
+      "step": 998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7120528188018566e-05,
+      "loss": 1.1229,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7109567227790754e-05,
+      "loss": 1.1081,
+      "step": 1002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7098588968447768e-05,
+      "loss": 0.9774,
+      "step": 1004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7087593436702086e-05,
+      "loss": 1.0047,
+      "step": 1006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.707658065930822e-05,
+      "loss": 1.0911,
+      "step": 1008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7065550663062634e-05,
+      "loss": 0.9891,
+      "step": 1010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7054503474803704e-05,
+      "loss": 1.17,
+      "step": 1012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.704343912141162e-05,
+      "loss": 1.106,
+      "step": 1014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.703235762980835e-05,
+      "loss": 1.066,
+      "step": 1016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7021259026957566e-05,
+      "loss": 1.1588,
+      "step": 1018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7010143339864563e-05,
+      "loss": 1.0991,
+      "step": 1020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6999010595576213e-05,
+      "loss": 1.103,
+      "step": 1022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6987860821180896e-05,
+      "loss": 1.196,
+      "step": 1024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.697669404380842e-05,
+      "loss": 1.0892,
+      "step": 1026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6965510290629973e-05,
+      "loss": 1.1086,
+      "step": 1028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6954309588858043e-05,
+      "loss": 1.0873,
+      "step": 1030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.694309196574637e-05,
+      "loss": 1.0697,
+      "step": 1032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6931857448589847e-05,
+      "loss": 0.9517,
+      "step": 1034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6920606064724486e-05,
+      "loss": 1.0223,
+      "step": 1036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6909337841527344e-05,
+      "loss": 1.1926,
+      "step": 1038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6898052806416446e-05,
+      "loss": 1.1018,
+      "step": 1040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6886750986850716e-05,
+      "loss": 1.0406,
+      "step": 1042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6875432410329934e-05,
+      "loss": 0.9755,
+      "step": 1044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.686409710439464e-05,
+      "loss": 1.0135,
+      "step": 1046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.685274509662609e-05,
+      "loss": 1.1118,
+      "step": 1048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.684137641464617e-05,
+      "loss": 1.0095,
+      "step": 1050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.682999108611735e-05,
+      "loss": 1.0496,
+      "step": 1052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6818589138742586e-05,
+      "loss": 1.0831,
+      "step": 1054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6807170600265296e-05,
+      "loss": 1.2248,
+      "step": 1056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6795735498469245e-05,
+      "loss": 1.0273,
+      "step": 1058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6784283861178514e-05,
+      "loss": 1.0812,
+      "step": 1060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 1.0574,
+      "step": 1062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6761331091610418e-05,
+      "loss": 1.1536,
+      "step": 1064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6749830015182106e-05,
+      "loss": 1.0092,
+      "step": 1066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6738312514957087e-05,
+      "loss": 1.0804,
+      "step": 1068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6726778618959928e-05,
+      "loss": 1.09,
+      "step": 1070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6715228355255093e-05,
+      "loss": 1.1343,
+      "step": 1072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6703661751946872e-05,
+      "loss": 1.0558,
+      "step": 1074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6692078837179317e-05,
+      "loss": 1.0913,
+      "step": 1076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6680479639136164e-05,
+      "loss": 1.0627,
+      "step": 1078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.666886418604077e-05,
+      "loss": 1.0714,
+      "step": 1080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.665723250615604e-05,
+      "loss": 1.1603,
+      "step": 1082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6645584627784383e-05,
+      "loss": 1.0104,
+      "step": 1084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.663392057926759e-05,
+      "loss": 1.1624,
+      "step": 1086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6622240388986824e-05,
+      "loss": 1.0206,
+      "step": 1088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6610544085362513e-05,
+      "loss": 1.0537,
+      "step": 1090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.659883169685429e-05,
+      "loss": 1.139,
+      "step": 1092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6587103251960935e-05,
+      "loss": 1.1376,
+      "step": 1094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6575358779220295e-05,
+      "loss": 1.0043,
+      "step": 1096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6563598307209207e-05,
+      "loss": 1.0649,
+      "step": 1098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6551821864543443e-05,
+      "loss": 1.0059,
+      "step": 1100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.654002947987764e-05,
+      "loss": 1.0802,
+      "step": 1102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6528221181905217e-05,
+      "loss": 1.0779,
+      "step": 1104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6516396999358323e-05,
+      "loss": 1.0919,
+      "step": 1106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6504556961007747e-05,
+      "loss": 1.0197,
+      "step": 1108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6492701095662866e-05,
+      "loss": 0.9916,
+      "step": 1110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6480829432171567e-05,
+      "loss": 1.0524,
+      "step": 1112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.646894199942017e-05,
+      "loss": 0.9794,
+      "step": 1114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.645703882633338e-05,
+      "loss": 0.9522,
+      "step": 1116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6445119941874184e-05,
+      "loss": 1.0006,
+      "step": 1118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.643318537504381e-05,
+      "loss": 1.1215,
+      "step": 1120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.642123515488164e-05,
+      "loss": 1.0781,
+      "step": 1122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6409269310465145e-05,
+      "loss": 1.04,
+      "step": 1124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6397287870909814e-05,
+      "loss": 1.0278,
+      "step": 1126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.638529086536908e-05,
+      "loss": 1.0284,
+      "step": 1128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6373278323034254e-05,
+      "loss": 1.0425,
+      "step": 1130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.636125027313445e-05,
+      "loss": 1.0831,
+      "step": 1132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6349206744936518e-05,
+      "loss": 1.2046,
+      "step": 1134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6337147767744966e-05,
+      "loss": 1.1424,
+      "step": 1136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.632507337090189e-05,
+      "loss": 1.0118,
+      "step": 1138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.631298358378692e-05,
+      "loss": 0.9081,
+      "step": 1140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6300878435817115e-05,
+      "loss": 1.1369,
+      "step": 1142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.628875795644692e-05,
+      "loss": 0.9326,
+      "step": 1144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6276622175168083e-05,
+      "loss": 1.0656,
+      "step": 1146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6264471121509587e-05,
+      "loss": 1.0247,
+      "step": 1148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6252304825037576e-05,
+      "loss": 1.0212,
+      "step": 1150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.624012331535528e-05,
+      "loss": 1.0461,
+      "step": 1152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6227926622102947e-05,
+      "loss": 1.1057,
+      "step": 1154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6215714774957775e-05,
+      "loss": 1.1477,
+      "step": 1156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.620348780363382e-05,
+      "loss": 1.0714,
+      "step": 1158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6191245737881956e-05,
+      "loss": 1.0335,
+      "step": 1160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.617898860748978e-05,
+      "loss": 1.1108,
+      "step": 1162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.616671644228153e-05,
+      "loss": 1.0287,
+      "step": 1164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.615442927211805e-05,
+      "loss": 0.9708,
+      "step": 1166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6142127126896682e-05,
+      "loss": 1.037,
+      "step": 1168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.61298100365512e-05,
+      "loss": 1.108,
+      "step": 1170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6117478031051756e-05,
+      "loss": 1.0569,
+      "step": 1172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.610513114040479e-05,
+      "loss": 1.0704,
+      "step": 1174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6092769394652946e-05,
+      "loss": 1.0724,
+      "step": 1176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.608039282387504e-05,
+      "loss": 0.9622,
+      "step": 1178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6068001458185934e-05,
+      "loss": 1.0873,
+      "step": 1180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.605559532773651e-05,
+      "loss": 1.1178,
+      "step": 1182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6043174462713565e-05,
+      "loss": 1.0736,
+      "step": 1184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6030738893339753e-05,
+      "loss": 0.9509,
+      "step": 1186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6018288649873495e-05,
+      "loss": 1.063,
+      "step": 1188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.600582376260894e-05,
+      "loss": 1.0378,
+      "step": 1190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5993344261875847e-05,
+      "loss": 0.9509,
+      "step": 1192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5980850178039547e-05,
+      "loss": 1.1301,
+      "step": 1194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.596834154150084e-05,
+      "loss": 1.0636,
+      "step": 1196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5955818382695953e-05,
+      "loss": 1.0508,
+      "step": 1198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5943280732096437e-05,
+      "loss": 0.9618,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.593072862020911e-05,
+      "loss": 0.9606,
+      "step": 1202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5918162077575976e-05,
+      "loss": 1.1394,
+      "step": 1204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5905581134774154e-05,
+      "loss": 1.0863,
+      "step": 1206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.589298582241579e-05,
+      "loss": 1.1145,
+      "step": 1208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5880376171148014e-05,
+      "loss": 1.0602,
+      "step": 1210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.586775221165283e-05,
+      "loss": 0.9911,
+      "step": 1212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.585511397464707e-05,
+      "loss": 1.1537,
+      "step": 1214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5842461490882288e-05,
+      "loss": 1.0622,
+      "step": 1216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5829794791144723e-05,
+      "loss": 1.1412,
+      "step": 1218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.581711390625519e-05,
+      "loss": 0.9844,
+      "step": 1220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.580441886706903e-05,
+      "loss": 0.9911,
+      "step": 1222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5791709704476016e-05,
+      "loss": 0.9844,
+      "step": 1224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5778986449400294e-05,
+      "loss": 1.092,
+      "step": 1226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5766249132800294e-05,
+      "loss": 0.9114,
+      "step": 1228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5753497785668662e-05,
+      "loss": 1.0616,
+      "step": 1230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5740732439032187e-05,
+      "loss": 1.0112,
+      "step": 1232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.572795312395172e-05,
+      "loss": 1.0562,
+      "step": 1234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.571515987152209e-05,
+      "loss": 0.9613,
+      "step": 1236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5702352712872055e-05,
+      "loss": 1.0254,
+      "step": 1238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5689531679164205e-05,
+      "loss": 1.0329,
+      "step": 1240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5676696801594886e-05,
+      "loss": 1.0865,
+      "step": 1242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.566384811139413e-05,
+      "loss": 1.0754,
+      "step": 1244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5650985639825583e-05,
+      "loss": 1.0884,
+      "step": 1246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5638109418186424e-05,
+      "loss": 1.1105,
+      "step": 1248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.562521947780728e-05,
+      "loss": 0.9587,
+      "step": 1250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5612315850052166e-05,
+      "loss": 1.0646,
+      "step": 1252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5599398566318398e-05,
+      "loss": 0.9582,
+      "step": 1254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5586467658036526e-05,
+      "loss": 0.9924,
+      "step": 1256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5573523156670245e-05,
+      "loss": 1.1182,
+      "step": 1258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5560565093716327e-05,
+      "loss": 1.0079,
+      "step": 1260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5547593500704547e-05,
+      "loss": 1.1699,
+      "step": 1262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5534608409197592e-05,
+      "loss": 1.0251,
+      "step": 1264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5521609850791004e-05,
+      "loss": 1.0481,
+      "step": 1266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5508597857113077e-05,
+      "loss": 1.0609,
+      "step": 1268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.549557245982482e-05,
+      "loss": 1.0378,
+      "step": 1270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5482533690619836e-05,
+      "loss": 0.9945,
+      "step": 1272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 1.0167,
+      "step": 1274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5456416163396732e-05,
+      "loss": 0.9759,
+      "step": 1276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5443337468928207e-05,
+      "loss": 1.1111,
+      "step": 1278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5430245529641987e-05,
+      "loss": 1.0053,
+      "step": 1280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5417140377393596e-05,
+      "loss": 1.0869,
+      "step": 1282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5404022044070705e-05,
+      "loss": 1.0718,
+      "step": 1284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5390890561593054e-05,
+      "loss": 1.0262,
+      "step": 1286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.537774596191238e-05,
+      "loss": 1.0595,
+      "step": 1288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5364588277012343e-05,
+      "loss": 0.9668,
+      "step": 1290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5351417538908436e-05,
+      "loss": 1.161,
+      "step": 1292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.533823377964791e-05,
+      "loss": 1.1035,
+      "step": 1294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5325037031309703e-05,
+      "loss": 1.046,
+      "step": 1296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5311827326004362e-05,
+      "loss": 1.0651,
+      "step": 1298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.529860469587396e-05,
+      "loss": 1.0803,
+      "step": 1300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5285369173092016e-05,
+      "loss": 1.1507,
+      "step": 1302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5272120789863414e-05,
+      "loss": 1.0646,
+      "step": 1304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.525885957842434e-05,
+      "loss": 1.1271,
+      "step": 1306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5245585571042194e-05,
+      "loss": 1.0881,
+      "step": 1308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5232298800015506e-05,
+      "loss": 1.0093,
+      "step": 1310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5218999297673863e-05,
+      "loss": 1.1195,
+      "step": 1312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.520568709637783e-05,
+      "loss": 1.0867,
+      "step": 1314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5192362228518876e-05,
+      "loss": 1.0524,
+      "step": 1316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5179024726519285e-05,
+      "loss": 1.1189,
+      "step": 1318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5165674622832084e-05,
+      "loss": 1.005,
+      "step": 1320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.515231194994097e-05,
+      "loss": 1.0414,
+      "step": 1322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5138936740360208e-05,
+      "loss": 1.0023,
+      "step": 1324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5125549026634585e-05,
+      "loss": 1.011,
+      "step": 1326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5112148841339295e-05,
+      "loss": 0.9053,
+      "step": 1328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.50987362170799e-05,
+      "loss": 1.0596,
+      "step": 1330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5085311186492206e-05,
+      "loss": 1.1282,
+      "step": 1332
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5071873782242223e-05,
+      "loss": 1.109,
+      "step": 1334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.505842403702606e-05,
+      "loss": 1.085,
+      "step": 1336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5044961983569856e-05,
+      "loss": 1.0244,
+      "step": 1338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5031487654629704e-05,
+      "loss": 1.1299,
+      "step": 1340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5018001082991553e-05,
+      "loss": 0.9897,
+      "step": 1342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.500450230147116e-05,
+      "loss": 0.998,
+      "step": 1344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4990991342913973e-05,
+      "loss": 1.035,
+      "step": 1346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4977468240195085e-05,
+      "loss": 1.0583,
+      "step": 1348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4963933026219122e-05,
+      "loss": 0.9951,
+      "step": 1350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.495038573392019e-05,
+      "loss": 1.0424,
+      "step": 1352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4936826396261783e-05,
+      "loss": 1.0204,
+      "step": 1354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4923255046236705e-05,
+      "loss": 0.9811,
+      "step": 1356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4909671716866985e-05,
+      "loss": 1.0521,
+      "step": 1358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4896076441203801e-05,
+      "loss": 0.9512,
+      "step": 1360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.48824692523274e-05,
+      "loss": 1.1036,
+      "step": 1362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.486885018334702e-05,
+      "loss": 1.0224,
+      "step": 1364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4855219267400798e-05,
+      "loss": 1.0679,
+      "step": 1366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4841576537655705e-05,
+      "loss": 1.0703,
+      "step": 1368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.482792202730745e-05,
+      "loss": 1.0549,
+      "step": 1370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4814255769580414e-05,
+      "loss": 1.0344,
+      "step": 1372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4800577797727558e-05,
+      "loss": 1.0001,
+      "step": 1374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4786888145030343e-05,
+      "loss": 0.9752,
+      "step": 1376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4773186844798662e-05,
+      "loss": 1.1156,
+      "step": 1378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4759473930370738e-05,
+      "loss": 1.0112,
+      "step": 1380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4745749435113059e-05,
+      "loss": 1.0115,
+      "step": 1382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4732013392420293e-05,
+      "loss": 1.0137,
+      "step": 1384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4718265835715202e-05,
+      "loss": 1.0115,
+      "step": 1386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4704506798448568e-05,
+      "loss": 0.9592,
+      "step": 1388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.46907363140991e-05,
+      "loss": 1.0084,
+      "step": 1390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4676954416173375e-05,
+      "loss": 1.1322,
+      "step": 1392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4663161138205723e-05,
+      "loss": 0.9445,
+      "step": 1394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4649356513758177e-05,
+      "loss": 1.105,
+      "step": 1396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4635540576420375e-05,
+      "loss": 1.0905,
+      "step": 1398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4621713359809479e-05,
+      "loss": 1.0894,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4607874897570104e-05,
+      "loss": 1.1155,
+      "step": 1402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.459402522337422e-05,
+      "loss": 1.0367,
+      "step": 1404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4580164370921079e-05,
+      "loss": 0.9746,
+      "step": 1406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4566292373937133e-05,
+      "loss": 0.9708,
+      "step": 1408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4552409266175953e-05,
+      "loss": 1.1791,
+      "step": 1410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4538515081418143e-05,
+      "loss": 0.9775,
+      "step": 1412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4524609853471266e-05,
+      "loss": 1.01,
+      "step": 1414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4510693616169742e-05,
+      "loss": 1.0487,
+      "step": 1416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4496766403374791e-05,
+      "loss": 0.9583,
+      "step": 1418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4482828248974335e-05,
+      "loss": 1.1337,
+      "step": 1420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4468879186882916e-05,
+      "loss": 1.0066,
+      "step": 1422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4454919251041624e-05,
+      "loss": 1.0504,
+      "step": 1424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4440948475418001e-05,
+      "loss": 0.9998,
+      "step": 1426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4426966894005967e-05,
+      "loss": 1.0397,
+      "step": 1428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4412974540825732e-05,
+      "loss": 0.979,
+      "step": 1430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4398971449923721e-05,
+      "loss": 1.0442,
+      "step": 1432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4384957655372484e-05,
+      "loss": 1.0746,
+      "step": 1434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4370933191270617e-05,
+      "loss": 1.0306,
+      "step": 1436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.435689809174267e-05,
+      "loss": 1.0168,
+      "step": 1438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.434285239093908e-05,
+      "loss": 0.9451,
+      "step": 1440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4328796123036072e-05,
+      "loss": 1.0496,
+      "step": 1442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.431472932223559e-05,
+      "loss": 1.059,
+      "step": 1444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4300652022765207e-05,
+      "loss": 1.1128,
+      "step": 1446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4286564258878032e-05,
+      "loss": 1.0558,
+      "step": 1448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4272466064852644e-05,
+      "loss": 1.054,
+      "step": 1450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4258357474992998e-05,
+      "loss": 1.0098,
+      "step": 1452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.424423852362835e-05,
+      "loss": 1.09,
+      "step": 1454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.423010924511316e-05,
+      "loss": 1.0351,
+      "step": 1456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.421596967382702e-05,
+      "loss": 0.9583,
+      "step": 1458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4201819844174566e-05,
+      "loss": 0.985,
+      "step": 1460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.418765979058539e-05,
+      "loss": 1.014,
+      "step": 1462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4173489547513975e-05,
+      "loss": 1.0463,
+      "step": 1464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4159309149439585e-05,
+      "loss": 1.0555,
+      "step": 1466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4145118630866187e-05,
+      "loss": 1.1,
+      "step": 1468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.41309180263224e-05,
+      "loss": 1.0573,
+      "step": 1470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.411670737036135e-05,
+      "loss": 0.9699,
+      "step": 1472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4102486697560649e-05,
+      "loss": 0.9899,
+      "step": 1474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4088256042522263e-05,
+      "loss": 1.0142,
+      "step": 1476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.407401543987246e-05,
+      "loss": 1.0445,
+      "step": 1478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4059764924261705e-05,
+      "loss": 1.1021,
+      "step": 1480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4045504530364585e-05,
+      "loss": 0.9518,
+      "step": 1482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4031234292879726e-05,
+      "loss": 0.9748,
+      "step": 1484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 1.037,
+      "step": 1486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4002664426060944e-05,
+      "loss": 1.1053,
+      "step": 1488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3988364866243693e-05,
+      "loss": 0.942,
+      "step": 1490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3974055601871867e-05,
+      "loss": 0.9726,
+      "step": 1492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3959736667762998e-05,
+      "loss": 1.0883,
+      "step": 1494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3945408098758155e-05,
+      "loss": 1.1857,
+      "step": 1496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3931069929721841e-05,
+      "loss": 1.0125,
+      "step": 1498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3916722195541927e-05,
+      "loss": 0.9293,
+      "step": 1500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3902364931129557e-05,
+      "loss": 1.0856,
+      "step": 1502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.388799817141906e-05,
+      "loss": 0.9689,
+      "step": 1504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3873621951367861e-05,
+      "loss": 1.002,
+      "step": 1506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3859236305956425e-05,
+      "loss": 1.0504,
+      "step": 1508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3844841270188134e-05,
+      "loss": 1.1238,
+      "step": 1510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3830436879089228e-05,
+      "loss": 1.1083,
+      "step": 1512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3816023167708706e-05,
+      "loss": 1.0482,
+      "step": 1514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3801600171118243e-05,
+      "loss": 0.9638,
+      "step": 1516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3787167924412113e-05,
+      "loss": 1.0552,
+      "step": 1518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3772726462707091e-05,
+      "loss": 0.9825,
+      "step": 1520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3758275821142382e-05,
+      "loss": 0.9823,
+      "step": 1522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3743816034879522e-05,
+      "loss": 0.9553,
+      "step": 1524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.37293471391023e-05,
+      "loss": 0.9501,
+      "step": 1526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3714869169016666e-05,
+      "loss": 1.0232,
+      "step": 1528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3700382159850656e-05,
+      "loss": 0.9845,
+      "step": 1530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3685886146854297e-05,
+      "loss": 1.2048,
+      "step": 1532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3671381165299524e-05,
+      "loss": 1.1017,
+      "step": 1534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.36568672504801e-05,
+      "loss": 1.0742,
+      "step": 1536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3642344437711513e-05,
+      "loss": 1.0223,
+      "step": 1538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3627812762330913e-05,
+      "loss": 1.0361,
+      "step": 1540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3613272259697007e-05,
+      "loss": 1.0305,
+      "step": 1542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3598722965189985e-05,
+      "loss": 0.9186,
+      "step": 1544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3584164914211428e-05,
+      "loss": 1.0692,
+      "step": 1546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3569598142184225e-05,
+      "loss": 1.0484,
+      "step": 1548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3555022684552484e-05,
+      "loss": 0.9853,
+      "step": 1550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3540438576781441e-05,
+      "loss": 0.9063,
+      "step": 1552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3525845854357392e-05,
+      "loss": 0.9795,
+      "step": 1554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3511244552787583e-05,
+      "loss": 0.993,
+      "step": 1556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3496634707600147e-05,
+      "loss": 1.06,
+      "step": 1558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.348201635434399e-05,
+      "loss": 1.0209,
+      "step": 1560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3467389528588733e-05,
+      "loss": 0.9821,
+      "step": 1562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3452754265924601e-05,
+      "loss": 0.9394,
+      "step": 1564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3438110601962362e-05,
+      "loss": 1.0648,
+      "step": 1566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3423458572333215e-05,
+      "loss": 1.0621,
+      "step": 1568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3408798212688718e-05,
+      "loss": 1.1157,
+      "step": 1570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.33941295587007e-05,
+      "loss": 0.9739,
+      "step": 1572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3379452646061163e-05,
+      "loss": 0.9967,
+      "step": 1574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.336476751048222e-05,
+      "loss": 0.9719,
+      "step": 1576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3350074187695979e-05,
+      "loss": 0.9223,
+      "step": 1578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3335372713454469e-05,
+      "loss": 1.0656,
+      "step": 1580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3320663123529562e-05,
+      "loss": 0.9943,
+      "step": 1582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.330594545371287e-05,
+      "loss": 1.0338,
+      "step": 1584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.329121973981567e-05,
+      "loss": 0.9358,
+      "step": 1586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3276486017668808e-05,
+      "loss": 1.031,
+      "step": 1588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3261744323122621e-05,
+      "loss": 0.9805,
+      "step": 1590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3246994692046837e-05,
+      "loss": 1.0385,
+      "step": 1592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3232237160330499e-05,
+      "loss": 1.0981,
+      "step": 1594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.321747176388188e-05,
+      "loss": 0.9949,
+      "step": 1596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3202698538628376e-05,
+      "loss": 1.0629,
+      "step": 1598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3187917520516448e-05,
+      "loss": 1.0223,
+      "step": 1600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3173128745511508e-05,
+      "loss": 1.0499,
+      "step": 1602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3158332249597843e-05,
+      "loss": 1.027,
+      "step": 1604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3143528068778527e-05,
+      "loss": 1.0735,
+      "step": 1606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3128716239075338e-05,
+      "loss": 1.0389,
+      "step": 1608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3113896796528664e-05,
+      "loss": 1.0096,
+      "step": 1610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3099069777197413e-05,
+      "loss": 0.9816,
+      "step": 1612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3084235217158929e-05,
+      "loss": 0.9918,
+      "step": 1614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3069393152508907e-05,
+      "loss": 1.0969,
+      "step": 1616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3054543619361302e-05,
+      "loss": 0.9449,
+      "step": 1618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.303968665384824e-05,
+      "loss": 0.961,
+      "step": 1620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3024822292119933e-05,
+      "loss": 0.995,
+      "step": 1622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3009950570344589e-05,
+      "loss": 1.0909,
+      "step": 1624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2995071524708324e-05,
+      "loss": 1.0801,
+      "step": 1626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2980185191415074e-05,
+      "loss": 0.8789,
+      "step": 1628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.296529160668651e-05,
+      "loss": 1.0743,
+      "step": 1630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2950390806761943e-05,
+      "loss": 0.9788,
+      "step": 1632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2935482827898249e-05,
+      "loss": 1.0473,
+      "step": 1634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.292056770636976e-05,
+      "loss": 1.041,
+      "step": 1636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2905645478468194e-05,
+      "loss": 1.1584,
+      "step": 1638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2890716180502566e-05,
+      "loss": 1.0537,
+      "step": 1640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2875779848799078e-05,
+      "loss": 1.064,
+      "step": 1642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2860836519701064e-05,
+      "loss": 1.0758,
+      "step": 1644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2845886229568873e-05,
+      "loss": 0.9112,
+      "step": 1646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2830929014779795e-05,
+      "loss": 1.0018,
+      "step": 1648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2815964911727972e-05,
+      "loss": 0.9525,
+      "step": 1650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2800993956824303e-05,
+      "loss": 1.0736,
+      "step": 1652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.278601618649636e-05,
+      "loss": 0.9452,
+      "step": 1654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.27710316371883e-05,
+      "loss": 0.9938,
+      "step": 1656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2756040345360771e-05,
+      "loss": 1.0259,
+      "step": 1658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.274104234749083e-05,
+      "loss": 0.988,
+      "step": 1660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2726037680071851e-05,
+      "loss": 1.0983,
+      "step": 1662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2711026379613437e-05,
+      "loss": 1.0379,
+      "step": 1664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2696008482641324e-05,
+      "loss": 1.0761,
+      "step": 1666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2680984025697313e-05,
+      "loss": 1.1011,
+      "step": 1668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2665953045339151e-05,
+      "loss": 1.0651,
+      "step": 1670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.265091557814047e-05,
+      "loss": 1.0795,
+      "step": 1672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2635871660690677e-05,
+      "loss": 0.9955,
+      "step": 1674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2620821329594879e-05,
+      "loss": 0.9747,
+      "step": 1676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2605764621473793e-05,
+      "loss": 0.9265,
+      "step": 1678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2590701572963642e-05,
+      "loss": 1.0082,
+      "step": 1680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2575632220716079e-05,
+      "loss": 1.0273,
+      "step": 1682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2560556601398101e-05,
+      "loss": 1.0197,
+      "step": 1684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2545474751691953e-05,
+      "loss": 0.924,
+      "step": 1686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2530386708295037e-05,
+      "loss": 0.9471,
+      "step": 1688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2515292507919829e-05,
+      "loss": 1.0104,
+      "step": 1690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.250019218729378e-05,
+      "loss": 0.9895,
+      "step": 1692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2485085783159239e-05,
+      "loss": 1.0076,
+      "step": 1694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2469973332273355e-05,
+      "loss": 0.9243,
+      "step": 1696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.9458,
+      "step": 1698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2439730437349636e-05,
+      "loss": 1.0546,
+      "step": 1700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2424600066899304e-05,
+      "loss": 1.0387,
+      "step": 1702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2409463796872463e-05,
+      "loss": 0.9942,
+      "step": 1704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2394321664098932e-05,
+      "loss": 1.0612,
+      "step": 1706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2379173705422795e-05,
+      "loss": 0.9398,
+      "step": 1708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2364019957702316e-05,
+      "loss": 0.9349,
+      "step": 1710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.234886045780984e-05,
+      "loss": 1.0658,
+      "step": 1712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2333695242631705e-05,
+      "loss": 1.0601,
+      "step": 1714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2318524349068171e-05,
+      "loss": 1.0399,
+      "step": 1716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2303347814033292e-05,
+      "loss": 1.0008,
+      "step": 1718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2288165674454871e-05,
+      "loss": 1.0375,
+      "step": 1720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.227297796727433e-05,
+      "loss": 1.036,
+      "step": 1722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2257784729446655e-05,
+      "loss": 1.0603,
+      "step": 1724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2242585997940275e-05,
+      "loss": 1.0761,
+      "step": 1726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2227381809736991e-05,
+      "loss": 1.0275,
+      "step": 1728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2212172201831885e-05,
+      "loss": 1.0424,
+      "step": 1730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2196957211233223e-05,
+      "loss": 0.9656,
+      "step": 1732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.218173687496237e-05,
+      "loss": 1.0192,
+      "step": 1734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2166511230053697e-05,
+      "loss": 0.9529,
+      "step": 1736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2151280313554486e-05,
+      "loss": 1.0015,
+      "step": 1738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2136044162524857e-05,
+      "loss": 0.98,
+      "step": 1740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2120802814037664e-05,
+      "loss": 1.0416,
+      "step": 1742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2105556305178398e-05,
+      "loss": 0.9874,
+      "step": 1744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2090304673045124e-05,
+      "loss": 1.0901,
+      "step": 1746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2075047954748354e-05,
+      "loss": 1.0148,
+      "step": 1748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2059786187410984e-05,
+      "loss": 0.9968,
+      "step": 1750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2044519408168201e-05,
+      "loss": 1.0267,
+      "step": 1752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2029247654167379e-05,
+      "loss": 0.9508,
+      "step": 1754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2013970962568003e-05,
+      "loss": 1.0579,
+      "step": 1756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1998689370541563e-05,
+      "loss": 1.0135,
+      "step": 1758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1983402915271478e-05,
+      "loss": 0.9435,
+      "step": 1760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1968111633953009e-05,
+      "loss": 1.0969,
+      "step": 1762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.195281556379314e-05,
+      "loss": 1.1063,
+      "step": 1764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.193751474201053e-05,
+      "loss": 0.8378,
+      "step": 1766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1922209205835381e-05,
+      "loss": 1.0479,
+      "step": 1768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.190689899250938e-05,
+      "loss": 0.8978,
+      "step": 1770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.189158413928558e-05,
+      "loss": 0.9973,
+      "step": 1772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1876264683428345e-05,
+      "loss": 0.9548,
+      "step": 1774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1860940662213212e-05,
+      "loss": 0.985,
+      "step": 1776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1845612112926843e-05,
+      "loss": 1.0125,
+      "step": 1778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.183027907286692e-05,
+      "loss": 0.9479,
+      "step": 1780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1814941579342045e-05,
+      "loss": 1.1257,
+      "step": 1782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1799599669671655e-05,
+      "loss": 0.9444,
+      "step": 1784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1784253381185937e-05,
+      "loss": 0.9679,
+      "step": 1786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.176890275122573e-05,
+      "loss": 1.1349,
+      "step": 1788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.175354781714244e-05,
+      "loss": 1.0984,
+      "step": 1790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1738188616297941e-05,
+      "loss": 1.0563,
+      "step": 1792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1722825186064494e-05,
+      "loss": 0.996,
+      "step": 1794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1707457563824647e-05,
+      "loss": 1.023,
+      "step": 1796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.169208578697115e-05,
+      "loss": 0.9333,
+      "step": 1798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1676709892906858e-05,
+      "loss": 0.954,
+      "step": 1800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1661329919044657e-05,
+      "loss": 0.925,
+      "step": 1802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.164594590280734e-05,
+      "loss": 1.057,
+      "step": 1804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1630557881627554e-05,
+      "loss": 1.1391,
+      "step": 1806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1615165892947679e-05,
+      "loss": 0.9459,
+      "step": 1808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1599769974219756e-05,
+      "loss": 0.9985,
+      "step": 1810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.158437016290539e-05,
+      "loss": 1.0195,
+      "step": 1812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1568966496475649e-05,
+      "loss": 0.9744,
+      "step": 1814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1553559012410984e-05,
+      "loss": 1.0585,
+      "step": 1816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1538147748201139e-05,
+      "loss": 1.0978,
+      "step": 1818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1522732741345053e-05,
+      "loss": 1.0574,
+      "step": 1820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1507314029350775e-05,
+      "loss": 1.096,
+      "step": 1822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1491891649735367e-05,
+      "loss": 0.9316,
+      "step": 1824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1476465640024814e-05,
+      "loss": 1.0443,
+      "step": 1826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1461036037753935e-05,
+      "loss": 0.9713,
+      "step": 1828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.144560288046629e-05,
+      "loss": 1.0231,
+      "step": 1830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1430166205714089e-05,
+      "loss": 1.0991,
+      "step": 1832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1414726051058103e-05,
+      "loss": 1.0327,
+      "step": 1834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.139928245406757e-05,
+      "loss": 0.9886,
+      "step": 1836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1383835452320098e-05,
+      "loss": 0.9413,
+      "step": 1838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1368385083401585e-05,
+      "loss": 0.9911,
+      "step": 1840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1352931384906126e-05,
+      "loss": 0.9354,
+      "step": 1842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1337474394435908e-05,
+      "loss": 1.034,
+      "step": 1844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1322014149601137e-05,
+      "loss": 0.9278,
+      "step": 1846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1306550688019926e-05,
+      "loss": 1.0181,
+      "step": 1848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.129108404731823e-05,
+      "loss": 1.0501,
+      "step": 1850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1275614265129731e-05,
+      "loss": 0.961,
+      "step": 1852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.126014137909575e-05,
+      "loss": 1.0528,
+      "step": 1854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1244665426865173e-05,
+      "loss": 1.1379,
+      "step": 1856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1229186446094338e-05,
+      "loss": 0.8832,
+      "step": 1858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1213704474446952e-05,
+      "loss": 0.998,
+      "step": 1860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1198219549594e-05,
+      "loss": 1.0951,
+      "step": 1862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1182731709213658e-05,
+      "loss": 0.9905,
+      "step": 1864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1167240990991192e-05,
+      "loss": 0.9449,
+      "step": 1866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1151747432618871e-05,
+      "loss": 1.0222,
+      "step": 1868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1136251071795871e-05,
+      "loss": 0.9773,
+      "step": 1870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1120751946228196e-05,
+      "loss": 0.9687,
+      "step": 1872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1105250093628566e-05,
+      "loss": 0.9521,
+      "step": 1874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1089745551716345e-05,
+      "loss": 1.0412,
+      "step": 1876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1074238358217437e-05,
+      "loss": 1.0068,
+      "step": 1878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1058728550864197e-05,
+      "loss": 0.9283,
+      "step": 1880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1043216167395345e-05,
+      "loss": 1.0227,
+      "step": 1882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1027701245555867e-05,
+      "loss": 1.0745,
+      "step": 1884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1012183823096919e-05,
+      "loss": 0.9773,
+      "step": 1886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0996663937775751e-05,
+      "loss": 0.9932,
+      "step": 1888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.09811416273556e-05,
+      "loss": 1.0579,
+      "step": 1890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.096561692960561e-05,
+      "loss": 0.9958,
+      "step": 1892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0950089882300721e-05,
+      "loss": 1.0588,
+      "step": 1894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0934560523221602e-05,
+      "loss": 1.0609,
+      "step": 1896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0919028890154544e-05,
+      "loss": 0.9774,
+      "step": 1898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0903495020891374e-05,
+      "loss": 0.9962,
+      "step": 1900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0887958953229349e-05,
+      "loss": 0.9606,
+      "step": 1902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0872420724971089e-05,
+      "loss": 0.8738,
+      "step": 1904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0856880373924461e-05,
+      "loss": 1.0137,
+      "step": 1906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0841337937902501e-05,
+      "loss": 1.0548,
+      "step": 1908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.9973,
+      "step": 1910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0810246962210019e-05,
+      "loss": 1.0262,
+      "step": 1912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0794698498190557e-05,
+      "loss": 1.0173,
+      "step": 1914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0779148100497723e-05,
+      "loss": 0.9807,
+      "step": 1916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0763595806968996e-05,
+      "loss": 1.0609,
+      "step": 1918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0748041655446473e-05,
+      "loss": 1.0371,
+      "step": 1920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0732485683776767e-05,
+      "loss": 1.0745,
+      "step": 1922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0716927929810925e-05,
+      "loss": 1.0435,
+      "step": 1924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0701368431404327e-05,
+      "loss": 0.9273,
+      "step": 1926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0685807226416599e-05,
+      "loss": 0.99,
+      "step": 1928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0670244352711518e-05,
+      "loss": 0.9298,
+      "step": 1930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0654679848156925e-05,
+      "loss": 1.0696,
+      "step": 1932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0639113750624626e-05,
+      "loss": 0.9104,
+      "step": 1934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0623546097990302e-05,
+      "loss": 0.9689,
+      "step": 1936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0607976928133423e-05,
+      "loss": 1.0285,
+      "step": 1938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0592406278937143e-05,
+      "loss": 1.0515,
+      "step": 1940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0576834188288226e-05,
+      "loss": 1.0184,
+      "step": 1942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0561260694076936e-05,
+      "loss": 1.0958,
+      "step": 1944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0545685834196948e-05,
+      "loss": 0.9664,
+      "step": 1946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0530109646545272e-05,
+      "loss": 1.017,
+      "step": 1948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.051453216902214e-05,
+      "loss": 0.866,
+      "step": 1950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0498953439530926e-05,
+      "loss": 1.0004,
+      "step": 1952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0483373495978046e-05,
+      "loss": 1.0156,
+      "step": 1954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0467792376272879e-05,
+      "loss": 0.961,
+      "step": 1956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0452210118327652e-05,
+      "loss": 0.9923,
+      "step": 1958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0436626760057379e-05,
+      "loss": 1.0013,
+      "step": 1960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0421042339379733e-05,
+      "loss": 0.9669,
+      "step": 1962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0405456894214987e-05,
+      "loss": 1.0208,
+      "step": 1964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0389870462485903e-05,
+      "loss": 0.9873,
+      "step": 1966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0374283082117635e-05,
+      "loss": 0.9581,
+      "step": 1968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0358694791037654e-05,
+      "loss": 0.9205,
+      "step": 1970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0343105627175645e-05,
+      "loss": 1.031,
+      "step": 1972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0327515628463415e-05,
+      "loss": 1.1879,
+      "step": 1974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0311924832834807e-05,
+      "loss": 0.9777,
+      "step": 1976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0296333278225599e-05,
+      "loss": 1.0311,
+      "step": 1978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0280741002573413e-05,
+      "loss": 1.0438,
+      "step": 1980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0265148043817632e-05,
+      "loss": 0.937,
+      "step": 1982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0249554439899299e-05,
+      "loss": 0.9321,
+      "step": 1984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0233960228761022e-05,
+      "loss": 0.9695,
+      "step": 1986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0218365448346893e-05,
+      "loss": 1.0823,
+      "step": 1988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0202770136602389e-05,
+      "loss": 0.9352,
+      "step": 1990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0187174331474272e-05,
+      "loss": 0.9465,
+      "step": 1992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0171578070910513e-05,
+      "loss": 1.0222,
+      "step": 1994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0155981392860187e-05,
+      "loss": 0.9526,
+      "step": 1996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0140384335273386e-05,
+      "loss": 1.0049,
+      "step": 1998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0124786936101128e-05,
+      "loss": 1.0112,
+      "step": 2000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0109189233295255e-05,
+      "loss": 0.9654,
+      "step": 2002
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0093591264808357e-05,
+      "loss": 1.1046,
+      "step": 2004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0077993068593663e-05,
+      "loss": 1.0173,
+      "step": 2006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0062394682604964e-05,
+      "loss": 0.9919,
+      "step": 2008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0046796144796499e-05,
+      "loss": 0.9841,
+      "step": 2010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.003119749312289e-05,
+      "loss": 0.9512,
+      "step": 2012
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0015598765539032e-05,
+      "loss": 1.1299,
+      "step": 2014
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.7274,
+      "step": 2016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.984401234460971e-06,
+      "loss": 0.6941,
+      "step": 2018
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.968802506877112e-06,
+      "loss": 0.6611,
+      "step": 2020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.953203855203505e-06,
+      "loss": 0.6265,
+      "step": 2022
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.937605317395041e-06,
+      "loss": 0.7557,
+      "step": 2024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.922006931406337e-06,
+      "loss": 0.6676,
+      "step": 2026
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.906408735191644e-06,
+      "loss": 0.7461,
+      "step": 2028
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.890810766704745e-06,
+      "loss": 0.7283,
+      "step": 2030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.875213063898875e-06,
+      "loss": 0.6275,
+      "step": 2032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.859615664726616e-06,
+      "loss": 0.7321,
+      "step": 2034
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.844018607139818e-06,
+      "loss": 0.6703,
+      "step": 2036
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.828421929089494e-06,
+      "loss": 0.7174,
+      "step": 2038
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.812825668525733e-06,
+      "loss": 0.7463,
+      "step": 2040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797229863397616e-06,
+      "loss": 0.645,
+      "step": 2042
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.781634551653108e-06,
+      "loss": 0.6774,
+      "step": 2044
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.766039771238982e-06,
+      "loss": 0.6713,
+      "step": 2046
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.750445560100705e-06,
+      "loss": 0.7031,
+      "step": 2048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.73485195618237e-06,
+      "loss": 0.6404,
+      "step": 2050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.719258997426588e-06,
+      "loss": 0.6305,
+      "step": 2052
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.703666721774403e-06,
+      "loss": 0.7386,
+      "step": 2054
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.688075167165196e-06,
+      "loss": 0.6172,
+      "step": 2056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.672484371536586e-06,
+      "loss": 0.6311,
+      "step": 2058
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.656894372824358e-06,
+      "loss": 0.675,
+      "step": 2060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.64130520896235e-06,
+      "loss": 0.6526,
+      "step": 2062
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.625716917882368e-06,
+      "loss": 0.5974,
+      "step": 2064
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.610129537514102e-06,
+      "loss": 0.5904,
+      "step": 2066
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.594543105785013e-06,
+      "loss": 0.772,
+      "step": 2068
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.578957660620267e-06,
+      "loss": 0.663,
+      "step": 2070
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.563373239942623e-06,
+      "loss": 0.6589,
+      "step": 2072
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.547789881672348e-06,
+      "loss": 0.7155,
+      "step": 2074
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.532207623727126e-06,
+      "loss": 0.6798,
+      "step": 2076
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.516626504021957e-06,
+      "loss": 0.6863,
+      "step": 2078
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.501046560469079e-06,
+      "loss": 0.6884,
+      "step": 2080
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.485467830977864e-06,
+      "loss": 0.6154,
+      "step": 2082
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.469890353454731e-06,
+      "loss": 0.6493,
+      "step": 2084
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.454314165803054e-06,
+      "loss": 0.656,
+      "step": 2086
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.438739305923067e-06,
+      "loss": 0.6525,
+      "step": 2088
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.423165811711777e-06,
+      "loss": 0.6226,
+      "step": 2090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.407593721062858e-06,
+      "loss": 0.6402,
+      "step": 2092
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.39202307186658e-06,
+      "loss": 0.6573,
+      "step": 2094
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.376453902009701e-06,
+      "loss": 0.626,
+      "step": 2096
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.360886249375375e-06,
+      "loss": 0.6027,
+      "step": 2098
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.345320151843078e-06,
+      "loss": 0.6264,
+      "step": 2100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.329755647288484e-06,
+      "loss": 0.6712,
+      "step": 2102
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.314192773583403e-06,
+      "loss": 0.6839,
+      "step": 2104
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.298631568595675e-06,
+      "loss": 0.631,
+      "step": 2106
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.283072070189074e-06,
+      "loss": 0.6935,
+      "step": 2108
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.267514316223234e-06,
+      "loss": 0.6987,
+      "step": 2110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.251958344553529e-06,
+      "loss": 0.5988,
+      "step": 2112
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.23640419303101e-06,
+      "loss": 0.7042,
+      "step": 2114
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.220851899502284e-06,
+      "loss": 0.6016,
+      "step": 2116
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.205301501809448e-06,
+      "loss": 0.6374,
+      "step": 2118
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.189753037789988e-06,
+      "loss": 0.6465,
+      "step": 2120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.6615,
+      "step": 2122
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.1586620620975e-06,
+      "loss": 0.5189,
+      "step": 2124
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.143119626075542e-06,
+      "loss": 0.6731,
+      "step": 2126
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.127579275028914e-06,
+      "loss": 0.806,
+      "step": 2128
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.112041046770653e-06,
+      "loss": 0.6313,
+      "step": 2130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.096504979108629e-06,
+      "loss": 0.6204,
+      "step": 2132
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.080971109845458e-06,
+      "loss": 0.7086,
+      "step": 2134
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.0654394767784e-06,
+      "loss": 0.7032,
+      "step": 2136
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.049910117699282e-06,
+      "loss": 0.662,
+      "step": 2138
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.034383070394394e-06,
+      "loss": 0.6562,
+      "step": 2140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.0188583726444e-06,
+      "loss": 0.6316,
+      "step": 2142
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.00333606222425e-06,
+      "loss": 0.6871,
+      "step": 2144
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.987816176903083e-06,
+      "loss": 0.6765,
+      "step": 2146
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.972298754444135e-06,
+      "loss": 0.6408,
+      "step": 2148
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.956783832604655e-06,
+      "loss": 0.6632,
+      "step": 2150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.941271449135806e-06,
+      "loss": 0.6747,
+      "step": 2152
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.925761641782568e-06,
+      "loss": 0.7084,
+      "step": 2154
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.91025444828366e-06,
+      "loss": 0.6591,
+      "step": 2156
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.89474990637144e-06,
+      "loss": 0.7025,
+      "step": 2158
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.879248053771809e-06,
+      "loss": 0.6193,
+      "step": 2160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.86374892820413e-06,
+      "loss": 0.6767,
+      "step": 2162
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.848252567381132e-06,
+      "loss": 0.698,
+      "step": 2164
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.832759009008811e-06,
+      "loss": 0.7042,
+      "step": 2166
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.817268290786343e-06,
+      "loss": 0.6342,
+      "step": 2168
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.801780450406002e-06,
+      "loss": 0.6638,
+      "step": 2170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.786295525553053e-06,
+      "loss": 0.6281,
+      "step": 2172
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.770813553905666e-06,
+      "loss": 0.7123,
+      "step": 2174
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.755334573134829e-06,
+      "loss": 0.7082,
+      "step": 2176
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.739858620904252e-06,
+      "loss": 0.6898,
+      "step": 2178
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.724385734870272e-06,
+      "loss": 0.6796,
+      "step": 2180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.708915952681771e-06,
+      "loss": 0.7019,
+      "step": 2182
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.693449311980074e-06,
+      "loss": 0.7283,
+      "step": 2184
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.677985850398866e-06,
+      "loss": 0.7167,
+      "step": 2186
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.662525605564094e-06,
+      "loss": 0.7631,
+      "step": 2188
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.647068615093876e-06,
+      "loss": 0.7207,
+      "step": 2190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.631614916598419e-06,
+      "loss": 0.5781,
+      "step": 2192
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.616164547679907e-06,
+      "loss": 0.7202,
+      "step": 2194
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.600717545932435e-06,
+      "loss": 0.7038,
+      "step": 2196
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.5852739489419e-06,
+      "loss": 0.6923,
+      "step": 2198
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.569833794285916e-06,
+      "loss": 0.6168,
+      "step": 2200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.554397119533715e-06,
+      "loss": 0.5952,
+      "step": 2202
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.53896396224607e-06,
+      "loss": 0.7057,
+      "step": 2204
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.52353435997519e-06,
+      "loss": 0.663,
+      "step": 2206
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.508108350264635e-06,
+      "loss": 0.6468,
+      "step": 2208
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.492685970649228e-06,
+      "loss": 0.6709,
+      "step": 2210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.47726725865495e-06,
+      "loss": 0.6867,
+      "step": 2212
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.461852251798865e-06,
+      "loss": 0.7253,
+      "step": 2214
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.446440987589019e-06,
+      "loss": 0.674,
+      "step": 2216
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.431033503524354e-06,
+      "loss": 0.6776,
+      "step": 2218
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.415629837094612e-06,
+      "loss": 0.6784,
+      "step": 2220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.400230025780242e-06,
+      "loss": 0.6327,
+      "step": 2222
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.384834107052321e-06,
+      "loss": 0.7206,
+      "step": 2224
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.369442118372448e-06,
+      "loss": 0.7468,
+      "step": 2226
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.35405409719266e-06,
+      "loss": 0.6357,
+      "step": 2228
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.338670080955348e-06,
+      "loss": 0.6712,
+      "step": 2230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.323290107093143e-06,
+      "loss": 0.6257,
+      "step": 2232
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.307914213028856e-06,
+      "loss": 0.6354,
+      "step": 2234
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.292542436175358e-06,
+      "loss": 0.6703,
+      "step": 2236
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.277174813935509e-06,
+      "loss": 0.663,
+      "step": 2238
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.26181138370206e-06,
+      "loss": 0.7074,
+      "step": 2240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.246452182857561e-06,
+      "loss": 0.6303,
+      "step": 2242
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.231097248774273e-06,
+      "loss": 0.6227,
+      "step": 2244
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.215746618814066e-06,
+      "loss": 0.6354,
+      "step": 2246
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.200400330328347e-06,
+      "loss": 0.7508,
+      "step": 2248
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.185058420657957e-06,
+      "loss": 0.6727,
+      "step": 2250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.16972092713308e-06,
+      "loss": 0.6373,
+      "step": 2252
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.154387887073159e-06,
+      "loss": 0.7155,
+      "step": 2254
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.139059337786793e-06,
+      "loss": 0.6555,
+      "step": 2256
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.12373531657166e-06,
+      "loss": 0.5803,
+      "step": 2258
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.108415860714418e-06,
+      "loss": 0.7269,
+      "step": 2260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.09310100749062e-06,
+      "loss": 0.6762,
+      "step": 2262
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.07779079416462e-06,
+      "loss": 0.6646,
+      "step": 2264
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.062485257989471e-06,
+      "loss": 0.7103,
+      "step": 2266
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.047184436206865e-06,
+      "loss": 0.6179,
+      "step": 2268
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.031888366046998e-06,
+      "loss": 0.6409,
+      "step": 2270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.016597084728525e-06,
+      "loss": 0.6268,
+      "step": 2272
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.001310629458443e-06,
+      "loss": 0.6717,
+      "step": 2274
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.986029037432e-06,
+      "loss": 0.6976,
+      "step": 2276
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.970752345832624e-06,
+      "loss": 0.6657,
+      "step": 2278
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.9554805918318e-06,
+      "loss": 0.6034,
+      "step": 2280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.940213812589018e-06,
+      "loss": 0.6882,
+      "step": 2282
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.924952045251651e-06,
+      "loss": 0.6342,
+      "step": 2284
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.909695326954878e-06,
+      "loss": 0.6281,
+      "step": 2286
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.894443694821604e-06,
+      "loss": 0.6404,
+      "step": 2288
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.879197185962339e-06,
+      "loss": 0.6916,
+      "step": 2290
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.863955837475144e-06,
+      "loss": 0.6599,
+      "step": 2292
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.848719686445516e-06,
+      "loss": 0.6776,
+      "step": 2294
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.833488769946308e-06,
+      "loss": 0.6596,
+      "step": 2296
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.818263125037633e-06,
+      "loss": 0.6694,
+      "step": 2298
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.803042788766779e-06,
+      "loss": 0.7016,
+      "step": 2300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.787827798168115e-06,
+      "loss": 0.6071,
+      "step": 2302
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.772618190263009e-06,
+      "loss": 0.5724,
+      "step": 2304
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.757414002059726e-06,
+      "loss": 0.6416,
+      "step": 2306
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.74221527055335e-06,
+      "loss": 0.6277,
+      "step": 2308
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.727022032725671e-06,
+      "loss": 0.6148,
+      "step": 2310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.711834325545135e-06,
+      "loss": 0.6552,
+      "step": 2312
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.696652185966713e-06,
+      "loss": 0.6757,
+      "step": 2314
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.681475650931834e-06,
+      "loss": 0.6434,
+      "step": 2316
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.666304757368297e-06,
+      "loss": 0.7195,
+      "step": 2318
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.651139542190164e-06,
+      "loss": 0.6691,
+      "step": 2320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.635980042297688e-06,
+      "loss": 0.6192,
+      "step": 2322
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.620826294577208e-06,
+      "loss": 0.7051,
+      "step": 2324
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.605678335901071e-06,
+      "loss": 0.6658,
+      "step": 2326
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.59053620312754e-06,
+      "loss": 0.6588,
+      "step": 2328
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.575399933100698e-06,
+      "loss": 0.6761,
+      "step": 2330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.560269562650368e-06,
+      "loss": 0.7106,
+      "step": 2332
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.6396,
+      "step": 2334
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.530026667726645e-06,
+      "loss": 0.6142,
+      "step": 2336
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.51491421684076e-06,
+      "loss": 0.6932,
+      "step": 2338
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.49980781270622e-06,
+      "loss": 0.7122,
+      "step": 2340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.484707492080172e-06,
+      "loss": 0.6239,
+      "step": 2342
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.469613291704962e-06,
+      "loss": 0.6901,
+      "step": 2344
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.454525248308051e-06,
+      "loss": 0.6955,
+      "step": 2346
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.439443398601903e-06,
+      "loss": 0.6267,
+      "step": 2348
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.424367779283926e-06,
+      "loss": 0.6965,
+      "step": 2350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.409298427036365e-06,
+      "loss": 0.673,
+      "step": 2352
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.3942353785262096e-06,
+      "loss": 0.681,
+      "step": 2354
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.379178670405123e-06,
+      "loss": 0.692,
+      "step": 2356
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.364128339309326e-06,
+      "loss": 0.689,
+      "step": 2358
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.349084421859533e-06,
+      "loss": 0.7263,
+      "step": 2360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.334046954660852e-06,
+      "loss": 0.719,
+      "step": 2362
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.31901597430269e-06,
+      "loss": 0.6281,
+      "step": 2364
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.303991517358678e-06,
+      "loss": 0.631,
+      "step": 2366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.288973620386568e-06,
+      "loss": 0.6206,
+      "step": 2368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.273962319928152e-06,
+      "loss": 0.6568,
+      "step": 2370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.2589576525091706e-06,
+      "loss": 0.7039,
+      "step": 2372
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.2439596546392295e-06,
+      "loss": 0.6154,
+      "step": 2374
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.228968362811702e-06,
+      "loss": 0.7513,
+      "step": 2376
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.21398381350364e-06,
+      "loss": 0.6674,
+      "step": 2378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.199006043175698e-06,
+      "loss": 0.6029,
+      "step": 2380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.1840350882720276e-06,
+      "loss": 0.606,
+      "step": 2382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.169070985220209e-06,
+      "loss": 0.6944,
+      "step": 2384
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.154113770431133e-06,
+      "loss": 0.6939,
+      "step": 2386
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.139163480298941e-06,
+      "loss": 0.6111,
+      "step": 2388
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.124220151200927e-06,
+      "loss": 0.6834,
+      "step": 2390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.10928381949744e-06,
+      "loss": 0.7028,
+      "step": 2392
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.094354521531808e-06,
+      "loss": 0.6278,
+      "step": 2394
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.079432293630244e-06,
+      "loss": 0.6528,
+      "step": 2396
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.064517172101754e-06,
+      "loss": 0.6989,
+      "step": 2398
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.0496091932380595e-06,
+      "loss": 0.6507,
+      "step": 2400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.034708393313494e-06,
+      "loss": 0.7065,
+      "step": 2402
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.019814808584928e-06,
+      "loss": 0.6606,
+      "step": 2404
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.004928475291679e-06,
+      "loss": 0.6786,
+      "step": 2406
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.9900494296554125e-06,
+      "loss": 0.6859,
+      "step": 2408
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.97517770788007e-06,
+      "loss": 0.6581,
+      "step": 2410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.960313346151761e-06,
+      "loss": 0.6455,
+      "step": 2412
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.9454563806387e-06,
+      "loss": 0.7072,
+      "step": 2414
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.930606847491094e-06,
+      "loss": 0.6631,
+      "step": 2416
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.915764782841073e-06,
+      "loss": 0.6231,
+      "step": 2418
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.900930222802589e-06,
+      "loss": 0.6292,
+      "step": 2420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.886103203471337e-06,
+      "loss": 0.6917,
+      "step": 2422
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.871283760924665e-06,
+      "loss": 0.6955,
+      "step": 2424
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.856471931221478e-06,
+      "loss": 0.6725,
+      "step": 2426
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.841667750402163e-06,
+      "loss": 0.6991,
+      "step": 2428
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.8268712544884965e-06,
+      "loss": 0.6314,
+      "step": 2430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.812082479483554e-06,
+      "loss": 0.6324,
+      "step": 2432
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.797301461371626e-06,
+      "loss": 0.592,
+      "step": 2434
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.782528236118124e-06,
+      "loss": 0.634,
+      "step": 2436
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.767762839669503e-06,
+      "loss": 0.6593,
+      "step": 2438
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.7530053079531664e-06,
+      "loss": 0.6776,
+      "step": 2440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.738255676877381e-06,
+      "loss": 0.6862,
+      "step": 2442
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.723513982331194e-06,
+      "loss": 0.6113,
+      "step": 2444
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.708780260184333e-06,
+      "loss": 0.599,
+      "step": 2446
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.694054546287132e-06,
+      "loss": 0.6889,
+      "step": 2448
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.679336876470441e-06,
+      "loss": 0.7056,
+      "step": 2450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.664627286545536e-06,
+      "loss": 0.6923,
+      "step": 2452
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.649925812304025e-06,
+      "loss": 0.6434,
+      "step": 2454
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.635232489517782e-06,
+      "loss": 0.6766,
+      "step": 2456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.620547353938837e-06,
+      "loss": 0.728,
+      "step": 2458
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.605870441299302e-06,
+      "loss": 0.6419,
+      "step": 2460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.591201787311286e-06,
+      "loss": 0.6459,
+      "step": 2462
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.5765414276667895e-06,
+      "loss": 0.6458,
+      "step": 2464
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.561889398037643e-06,
+      "loss": 0.6214,
+      "step": 2466
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.547245734075403e-06,
+      "loss": 0.7162,
+      "step": 2468
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.532610471411274e-06,
+      "loss": 0.6957,
+      "step": 2470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.517983645656014e-06,
+      "loss": 0.6705,
+      "step": 2472
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.503365292399857e-06,
+      "loss": 0.6689,
+      "step": 2474
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.488755447212418e-06,
+      "loss": 0.6398,
+      "step": 2476
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.4741541456426115e-06,
+      "loss": 0.651,
+      "step": 2478
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.459561423218561e-06,
+      "loss": 0.6459,
+      "step": 2480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.444977315447521e-06,
+      "loss": 0.6324,
+      "step": 2482
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.4304018578157755e-06,
+      "loss": 0.6501,
+      "step": 2484
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.415835085788574e-06,
+      "loss": 0.6496,
+      "step": 2486
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.401277034810017e-06,
+      "loss": 0.6259,
+      "step": 2488
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.386727740302995e-06,
+      "loss": 0.6057,
+      "step": 2490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.37218723766909e-06,
+      "loss": 0.638,
+      "step": 2492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.357655562288487e-06,
+      "loss": 0.7214,
+      "step": 2494
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.3431327495199025e-06,
+      "loss": 0.6199,
+      "step": 2496
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.328618834700475e-06,
+      "loss": 0.5947,
+      "step": 2498
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.314113853145704e-06,
+      "loss": 0.7102,
+      "step": 2500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.299617840149349e-06,
+      "loss": 0.6289,
+      "step": 2502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.28513083098334e-06,
+      "loss": 0.65,
+      "step": 2504
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.270652860897705e-06,
+      "loss": 0.6168,
+      "step": 2506
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.25618396512048e-06,
+      "loss": 0.6898,
+      "step": 2508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.241724178857621e-06,
+      "loss": 0.7599,
+      "step": 2510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.227273537292911e-06,
+      "loss": 0.6371,
+      "step": 2512
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.212832075587891e-06,
+      "loss": 0.6176,
+      "step": 2514
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.19839982888176e-06,
+      "loss": 0.6578,
+      "step": 2516
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.1839768322912966e-06,
+      "loss": 0.6292,
+      "step": 2518
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.169563120910775e-06,
+      "loss": 0.6349,
+      "step": 2520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.155158729811867e-06,
+      "loss": 0.6743,
+      "step": 2522
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.140763694043578e-06,
+      "loss": 0.6624,
+      "step": 2524
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.12637804863214e-06,
+      "loss": 0.6474,
+      "step": 2526
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.112001828580945e-06,
+      "loss": 0.6349,
+      "step": 2528
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.097635068870445e-06,
+      "loss": 0.6611,
+      "step": 2530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.083277804458072e-06,
+      "loss": 0.6373,
+      "step": 2532
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.068930070278159e-06,
+      "loss": 0.741,
+      "step": 2534
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.054591901241846e-06,
+      "loss": 0.7184,
+      "step": 2536
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.0402633322370015e-06,
+      "loss": 0.6419,
+      "step": 2538
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.025944398128137e-06,
+      "loss": 0.6488,
+      "step": 2540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.011635133756309e-06,
+      "loss": 0.6785,
+      "step": 2542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.99733557393906e-06,
+      "loss": 0.6235,
+      "step": 2544
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.7035,
+      "step": 2546
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.96876570712028e-06,
+      "loss": 0.6098,
+      "step": 2548
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.954495469635418e-06,
+      "loss": 0.6903,
+      "step": 2550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.9402350757382965e-06,
+      "loss": 0.6815,
+      "step": 2552
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.925984560127542e-06,
+      "loss": 0.6078,
+      "step": 2554
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.911743957477739e-06,
+      "loss": 0.658,
+      "step": 2556
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.897513302439355e-06,
+      "loss": 0.7074,
+      "step": 2558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.883292629638651e-06,
+      "loss": 0.7053,
+      "step": 2560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.869081973677604e-06,
+      "loss": 0.6775,
+      "step": 2562
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8548813691338135e-06,
+      "loss": 0.656,
+      "step": 2564
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.84069085056042e-06,
+      "loss": 0.7382,
+      "step": 2566
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.826510452486027e-06,
+      "loss": 0.7037,
+      "step": 2568
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.81234020941461e-06,
+      "loss": 0.7243,
+      "step": 2570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.798180155825437e-06,
+      "loss": 0.6948,
+      "step": 2572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.784030326172981e-06,
+      "loss": 0.6497,
+      "step": 2574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.76989075488684e-06,
+      "loss": 0.6378,
+      "step": 2576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.755761476371653e-06,
+      "loss": 0.6219,
+      "step": 2578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.741642525007003e-06,
+      "loss": 0.5762,
+      "step": 2580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.727533935147359e-06,
+      "loss": 0.662,
+      "step": 2582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.7134357411219755e-06,
+      "loss": 0.6347,
+      "step": 2584
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.699347977234799e-06,
+      "loss": 0.6086,
+      "step": 2586
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.6852706777644115e-06,
+      "loss": 0.64,
+      "step": 2588
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.671203876963931e-06,
+      "loss": 0.6484,
+      "step": 2590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.657147609060924e-06,
+      "loss": 0.6643,
+      "step": 2592
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.643101908257334e-06,
+      "loss": 0.6417,
+      "step": 2594
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.629066808729385e-06,
+      "loss": 0.7175,
+      "step": 2596
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.615042344627515e-06,
+      "loss": 0.6875,
+      "step": 2598
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.601028550076278e-06,
+      "loss": 0.6308,
+      "step": 2600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.587025459174271e-06,
+      "loss": 0.6949,
+      "step": 2602
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.573033105994038e-06,
+      "loss": 0.6179,
+      "step": 2604
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.559051524582002e-06,
+      "loss": 0.6499,
+      "step": 2606
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.545080748958378e-06,
+      "loss": 0.6996,
+      "step": 2608
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.531120813117086e-06,
+      "loss": 0.65,
+      "step": 2610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.517171751025668e-06,
+      "loss": 0.6405,
+      "step": 2612
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.503233596625211e-06,
+      "loss": 0.6725,
+      "step": 2614
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.489306383830258e-06,
+      "loss": 0.7136,
+      "step": 2616
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.475390146528738e-06,
+      "loss": 0.6647,
+      "step": 2618
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.461484918581859e-06,
+      "loss": 0.5684,
+      "step": 2620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.44759073382405e-06,
+      "loss": 0.6602,
+      "step": 2622
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.43370762606287e-06,
+      "loss": 0.691,
+      "step": 2624
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.419835629078928e-06,
+      "loss": 0.6414,
+      "step": 2626
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.405974776625785e-06,
+      "loss": 0.6204,
+      "step": 2628
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.392125102429899e-06,
+      "loss": 0.7113,
+      "step": 2630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.378286640190522e-06,
+      "loss": 0.6489,
+      "step": 2632
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.364459423579629e-06,
+      "loss": 0.6971,
+      "step": 2634
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.350643486241826e-06,
+      "loss": 0.6009,
+      "step": 2636
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.33683886179428e-06,
+      "loss": 0.657,
+      "step": 2638
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.3230455838266275e-06,
+      "loss": 0.6851,
+      "step": 2640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.309263685900899e-06,
+      "loss": 0.6631,
+      "step": 2642
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.295493201551433e-06,
+      "loss": 0.6206,
+      "step": 2644
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.281734164284802e-06,
+      "loss": 0.6186,
+      "step": 2646
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.267986607579711e-06,
+      "loss": 0.6177,
+      "step": 2648
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.254250564886944e-06,
+      "loss": 0.6329,
+      "step": 2650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.240526069629265e-06,
+      "loss": 0.615,
+      "step": 2652
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.22681315520134e-06,
+      "loss": 0.6553,
+      "step": 2654
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.213111854969661e-06,
+      "loss": 0.6016,
+      "step": 2656
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.1994222022724486e-06,
+      "loss": 0.7026,
+      "step": 2658
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.1857442304195895e-06,
+      "loss": 0.7029,
+      "step": 2660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.172077972692553e-06,
+      "loss": 0.7096,
+      "step": 2662
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.158423462344297e-06,
+      "loss": 0.6248,
+      "step": 2664
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.144780732599202e-06,
+      "loss": 0.6935,
+      "step": 2666
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.13114981665298e-06,
+      "loss": 0.6597,
+      "step": 2668
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.117530747672603e-06,
+      "loss": 0.635,
+      "step": 2670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.103923558796203e-06,
+      "loss": 0.6966,
+      "step": 2672
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.090328283133019e-06,
+      "loss": 0.6497,
+      "step": 2674
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.076744953763299e-06,
+      "loss": 0.7003,
+      "step": 2676
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.06317360373822e-06,
+      "loss": 0.6528,
+      "step": 2678
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.049614266079813e-06,
+      "loss": 0.6376,
+      "step": 2680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.036066973780882e-06,
+      "loss": 0.5926,
+      "step": 2682
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.022531759804919e-06,
+      "loss": 0.6111,
+      "step": 2684
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.009008657086025e-06,
+      "loss": 0.6644,
+      "step": 2686
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.99549769852884e-06,
+      "loss": 0.608,
+      "step": 2688
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.981998917008448e-06,
+      "loss": 0.6816,
+      "step": 2690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9685123453703e-06,
+      "loss": 0.6094,
+      "step": 2692
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.955038016430149e-06,
+      "loss": 0.6026,
+      "step": 2694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.941575962973946e-06,
+      "loss": 0.6461,
+      "step": 2696
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.928126217757783e-06,
+      "loss": 0.6004,
+      "step": 2698
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.914688813507798e-06,
+      "loss": 0.6385,
+      "step": 2700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.901263782920105e-06,
+      "loss": 0.7059,
+      "step": 2702
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.8878511586607055e-06,
+      "loss": 0.6244,
+      "step": 2704
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.874450973365419e-06,
+      "loss": 0.6682,
+      "step": 2706
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.861063259639793e-06,
+      "loss": 0.6546,
+      "step": 2708
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.847688050059033e-06,
+      "loss": 0.6224,
+      "step": 2710
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.8343253771679155e-06,
+      "loss": 0.6317,
+      "step": 2712
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.8209752734807205e-06,
+      "loss": 0.5852,
+      "step": 2714
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.8076377714811285e-06,
+      "loss": 0.6472,
+      "step": 2716
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7943129036221735e-06,
+      "loss": 0.6486,
+      "step": 2718
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.781000702326141e-06,
+      "loss": 0.6729,
+      "step": 2720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.767701199984497e-06,
+      "loss": 0.6962,
+      "step": 2722
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.754414428957806e-06,
+      "loss": 0.6928,
+      "step": 2724
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.74114042157566e-06,
+      "loss": 0.625,
+      "step": 2726
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7278792101365864e-06,
+      "loss": 0.6235,
+      "step": 2728
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.714630826907986e-06,
+      "loss": 0.6811,
+      "step": 2730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.701395304126039e-06,
+      "loss": 0.6498,
+      "step": 2732
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.688172673995638e-06,
+      "loss": 0.6704,
+      "step": 2734
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6749629686902984e-06,
+      "loss": 0.6476,
+      "step": 2736
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.661766220352098e-06,
+      "loss": 0.6569,
+      "step": 2738
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.64858246109157e-06,
+      "loss": 0.637,
+      "step": 2740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.63541172298766e-06,
+      "loss": 0.5807,
+      "step": 2742
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.622254038087622e-06,
+      "loss": 0.6747,
+      "step": 2744
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.60910943840695e-06,
+      "loss": 0.5375,
+      "step": 2746
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.595977955929298e-06,
+      "loss": 0.6979,
+      "step": 2748
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.582859622606406e-06,
+      "loss": 0.6505,
+      "step": 2750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.569754470358015e-06,
+      "loss": 0.605,
+      "step": 2752
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.556662531071796e-06,
+      "loss": 0.6785,
+      "step": 2754
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.543583836603271e-06,
+      "loss": 0.5867,
+      "step": 2756
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.5912,
+      "step": 2758
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.517466309380167e-06,
+      "loss": 0.6986,
+      "step": 2760
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.504427540175182e-06,
+      "loss": 0.6569,
+      "step": 2762
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.4914021428869225e-06,
+      "loss": 0.7133,
+      "step": 2764
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.4783901492089985e-06,
+      "loss": 0.5889,
+      "step": 2766
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.465391590802407e-06,
+      "loss": 0.6494,
+      "step": 2768
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.452406499295452e-06,
+      "loss": 0.6486,
+      "step": 2770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.439434906283674e-06,
+      "loss": 0.728,
+      "step": 2772
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.4264768433297565e-06,
+      "loss": 0.6382,
+      "step": 2774
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.413532341963477e-06,
+      "loss": 0.6125,
+      "step": 2776
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.4006014336816035e-06,
+      "loss": 0.6633,
+      "step": 2778
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.387684149947837e-06,
+      "loss": 0.6362,
+      "step": 2780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3747805221927266e-06,
+      "loss": 0.6853,
+      "step": 2782
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.36189058181358e-06,
+      "loss": 0.6875,
+      "step": 2784
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.349014360174417e-06,
+      "loss": 0.6433,
+      "step": 2786
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.336151888605871e-06,
+      "loss": 0.6001,
+      "step": 2788
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.323303198405117e-06,
+      "loss": 0.6404,
+      "step": 2790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.310468320835797e-06,
+      "loss": 0.7064,
+      "step": 2792
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.297647287127946e-06,
+      "loss": 0.5622,
+      "step": 2794
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.284840128477913e-06,
+      "loss": 0.673,
+      "step": 2796
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.272046876048286e-06,
+      "loss": 0.7006,
+      "step": 2798
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.259267560967813e-06,
+      "loss": 0.6462,
+      "step": 2800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.24650221433134e-06,
+      "loss": 0.6858,
+      "step": 2802
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.233750867199708e-06,
+      "loss": 0.5764,
+      "step": 2804
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.221013550599707e-06,
+      "loss": 0.6528,
+      "step": 2806
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.208290295523985e-06,
+      "loss": 0.6805,
+      "step": 2808
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.195581132930975e-06,
+      "loss": 0.629,
+      "step": 2810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.1828860937448135e-06,
+      "loss": 0.7001,
+      "step": 2812
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.170205208855281e-06,
+      "loss": 0.6914,
+      "step": 2814
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.157538509117714e-06,
+      "loss": 0.6342,
+      "step": 2816
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.144886025352934e-06,
+      "loss": 0.671,
+      "step": 2818
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.13224778834717e-06,
+      "loss": 0.6501,
+      "step": 2820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.119623828851987e-06,
+      "loss": 0.6591,
+      "step": 2822
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.107014177584211e-06,
+      "loss": 0.6779,
+      "step": 2824
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.094418865225853e-06,
+      "loss": 0.5933,
+      "step": 2826
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.081837922424027e-06,
+      "loss": 0.6545,
+      "step": 2828
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.069271379790891e-06,
+      "loss": 0.6536,
+      "step": 2830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.056719267903564e-06,
+      "loss": 0.653,
+      "step": 2832
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.044181617304048e-06,
+      "loss": 0.647,
+      "step": 2834
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.0316584584991605e-06,
+      "loss": 0.6862,
+      "step": 2836
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.019149821960455e-06,
+      "loss": 0.6298,
+      "step": 2838
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.0066557381241524e-06,
+      "loss": 0.6679,
+      "step": 2840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.994176237391059e-06,
+      "loss": 0.6181,
+      "step": 2842
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.981711350126501e-06,
+      "loss": 0.5991,
+      "step": 2844
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.969261106660252e-06,
+      "loss": 0.6876,
+      "step": 2846
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.956825537286436e-06,
+      "loss": 0.6379,
+      "step": 2848
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.944404672263494e-06,
+      "loss": 0.6842,
+      "step": 2850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.931998541814069e-06,
+      "loss": 0.6142,
+      "step": 2852
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.9196071761249665e-06,
+      "loss": 0.6461,
+      "step": 2854
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.907230605347057e-06,
+      "loss": 0.6672,
+      "step": 2856
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.894868859595216e-06,
+      "loss": 0.7138,
+      "step": 2858
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.882521968948246e-06,
+      "loss": 0.6803,
+      "step": 2860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.870189963448801e-06,
+      "loss": 0.6145,
+      "step": 2862
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.857872873103322e-06,
+      "loss": 0.6199,
+      "step": 2864
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.845570727881951e-06,
+      "loss": 0.6396,
+      "step": 2866
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.833283557718471e-06,
+      "loss": 0.6507,
+      "step": 2868
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.821011392510227e-06,
+      "loss": 0.668,
+      "step": 2870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.808754262118046e-06,
+      "loss": 0.6487,
+      "step": 2872
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7965121963661823e-06,
+      "loss": 0.5782,
+      "step": 2874
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7842852250422293e-06,
+      "loss": 0.6581,
+      "step": 2876
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.772073377897052e-06,
+      "loss": 0.6742,
+      "step": 2878
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7598766846447186e-06,
+      "loss": 0.6115,
+      "step": 2880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7476951749624236e-06,
+      "loss": 0.6028,
+      "step": 2882
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.735528878490412e-06,
+      "loss": 0.6697,
+      "step": 2884
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7233778248319175e-06,
+      "loss": 0.655,
+      "step": 2886
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.711242043553085e-06,
+      "loss": 0.7478,
+      "step": 2888
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.6991215641828903e-06,
+      "loss": 0.6176,
+      "step": 2890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.687016416213084e-06,
+      "loss": 0.6596,
+      "step": 2892
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.674926629098113e-06,
+      "loss": 0.7006,
+      "step": 2894
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6628522322550396e-06,
+      "loss": 0.6353,
+      "step": 2896
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.650793255063485e-06,
+      "loss": 0.6088,
+      "step": 2898
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6387497268655526e-06,
+      "loss": 0.6223,
+      "step": 2900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6267216769657486e-06,
+      "loss": 0.6713,
+      "step": 2902
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.614709134630923e-06,
+      "loss": 0.6149,
+      "step": 2904
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6027121290901888e-06,
+      "loss": 0.6516,
+      "step": 2906
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.590730689534857e-06,
+      "loss": 0.6587,
+      "step": 2908
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.578764845118362e-06,
+      "loss": 0.702,
+      "step": 2910
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5668146249561943e-06,
+      "loss": 0.6854,
+      "step": 2912
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.554880058125819e-06,
+      "loss": 0.6114,
+      "step": 2914
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5429611736666237e-06,
+      "loss": 0.6724,
+      "step": 2916
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.53105800057983e-06,
+      "loss": 0.7208,
+      "step": 2918
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.519170567828435e-06,
+      "loss": 0.6079,
+      "step": 2920
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.507298904337134e-06,
+      "loss": 0.6364,
+      "step": 2922
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4954430389922534e-06,
+      "loss": 0.652,
+      "step": 2924
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4836030006416777e-06,
+      "loss": 0.6373,
+      "step": 2926
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4717788180947855e-06,
+      "loss": 0.6763,
+      "step": 2928
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.459970520122364e-06,
+      "loss": 0.6164,
+      "step": 2930
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4481781354565604e-06,
+      "loss": 0.6251,
+      "step": 2932
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.436401692790797e-06,
+      "loss": 0.686,
+      "step": 2934
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.424641220779711e-06,
+      "loss": 0.6603,
+      "step": 2936
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4128967480390673e-06,
+      "loss": 0.5985,
+      "step": 2938
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4011683031457134e-06,
+      "loss": 0.6048,
+      "step": 2940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.389455914637493e-06,
+      "loss": 0.6333,
+      "step": 2942
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3777596110131805e-06,
+      "loss": 0.5561,
+      "step": 2944
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.366079420732413e-06,
+      "loss": 0.6097,
+      "step": 2946
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3544153722156214e-06,
+      "loss": 0.6303,
+      "step": 2948
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.342767493843959e-06,
+      "loss": 0.6627,
+      "step": 2950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.331135813959232e-06,
+      "loss": 0.6805,
+      "step": 2952
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.319520360863837e-06,
+      "loss": 0.621,
+      "step": 2954
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.307921162820685e-06,
+      "loss": 0.675,
+      "step": 2956
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2963382480531292e-06,
+      "loss": 0.6276,
+      "step": 2958
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2847716447449095e-06,
+      "loss": 0.6127,
+      "step": 2960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2732213810400746e-06,
+      "loss": 0.6163,
+      "step": 2962
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2616874850429148e-06,
+      "loss": 0.5749,
+      "step": 2964
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.250169984817897e-06,
+      "loss": 0.6647,
+      "step": 2966
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2386689083895863e-06,
+      "loss": 0.6217,
+      "step": 2968
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.6624,
+      "step": 2970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2157161388214884e-06,
+      "loss": 0.5726,
+      "step": 2972
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.204264501530756e-06,
+      "loss": 0.5676,
+      "step": 2974
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.192829399734706e-06,
+      "loss": 0.6874,
+      "step": 2976
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1814108612574134e-06,
+      "loss": 0.6216,
+      "step": 2978
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1700089138826564e-06,
+      "loss": 0.6749,
+      "step": 2980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.158623585353833e-06,
+      "loss": 0.6845,
+      "step": 2982
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.147254903373913e-06,
+      "loss": 0.6404,
+      "step": 2984
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1359028956053617e-06,
+      "loss": 0.6706,
+      "step": 2986
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.124567589670069e-06,
+      "loss": 0.6484,
+      "step": 2988
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1132490131492844e-06,
+      "loss": 0.6283,
+      "step": 2990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1019471935835565e-06,
+      "loss": 0.6465,
+      "step": 2992
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.090662158472655e-06,
+      "loss": 0.6932,
+      "step": 2994
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.079393935275513e-06,
+      "loss": 0.6287,
+      "step": 2996
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0681425514101547e-06,
+      "loss": 0.6626,
+      "step": 2998
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.056908034253635e-06,
+      "loss": 0.7158,
+      "step": 3000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.045690411141957e-06,
+      "loss": 0.5933,
+      "step": 3002
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0344897093700333e-06,
+      "loss": 0.6204,
+      "step": 3004
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0233059561915857e-06,
+      "loss": 0.7443,
+      "step": 3006
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.01213917881911e-06,
+      "loss": 0.6204,
+      "step": 3008
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0009894044237907e-06,
+      "loss": 0.6231,
+      "step": 3010
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.9898566601354417e-06,
+      "loss": 0.6244,
+      "step": 3012
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.978740973042438e-06,
+      "loss": 0.6864,
+      "step": 3014
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.96764237019165e-06,
+      "loss": 0.5827,
+      "step": 3016
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.9565608785883817e-06,
+      "loss": 0.5573,
+      "step": 3018
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.9454965251962973e-06,
+      "loss": 0.5724,
+      "step": 3020
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.934449336937364e-06,
+      "loss": 0.7145,
+      "step": 3022
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.9234193406917833e-06,
+      "loss": 0.5912,
+      "step": 3024
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.912406563297916e-06,
+      "loss": 0.5811,
+      "step": 3026
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.901411031552236e-06,
+      "loss": 0.6165,
+      "step": 3028
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.8904327722092495e-06,
+      "loss": 0.6321,
+      "step": 3030
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.879471811981437e-06,
+      "loss": 0.5852,
+      "step": 3032
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.868528177539187e-06,
+      "loss": 0.6749,
+      "step": 3034
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.857601895510729e-06,
+      "loss": 0.6925,
+      "step": 3036
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8466929924820708e-06,
+      "loss": 0.6801,
+      "step": 3038
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8358014949969338e-06,
+      "loss": 0.6904,
+      "step": 3040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8249274295566863e-06,
+      "loss": 0.685,
+      "step": 3042
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8140708226202883e-06,
+      "loss": 0.718,
+      "step": 3044
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.803231700604204e-06,
+      "loss": 0.6328,
+      "step": 3046
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.79241008988237e-06,
+      "loss": 0.6388,
+      "step": 3048
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7816060167861003e-06,
+      "loss": 0.619,
+      "step": 3050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7708195076040445e-06,
+      "loss": 0.7083,
+      "step": 3052
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.760050588582114e-06,
+      "loss": 0.7122,
+      "step": 3054
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.749299285923417e-06,
+      "loss": 0.6874,
+      "step": 3056
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7385656257882e-06,
+      "loss": 0.6636,
+      "step": 3058
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7278496342937788e-06,
+      "loss": 0.5918,
+      "step": 3060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.717151337514482e-06,
+      "loss": 0.6187,
+      "step": 3062
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7064707614815776e-06,
+      "loss": 0.6227,
+      "step": 3064
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.695807932183219e-06,
+      "loss": 0.6801,
+      "step": 3066
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.685162875564378e-06,
+      "loss": 0.6227,
+      "step": 3068
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.674535617526777e-06,
+      "loss": 0.6153,
+      "step": 3070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6639261839288345e-06,
+      "loss": 0.6088,
+      "step": 3072
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6533346005855986e-06,
+      "loss": 0.719,
+      "step": 3074
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.642760893268684e-06,
+      "loss": 0.6458,
+      "step": 3076
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.632205087706207e-06,
+      "loss": 0.6118,
+      "step": 3078
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6216672095827267e-06,
+      "loss": 0.6674,
+      "step": 3080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.611147284539183e-06,
+      "loss": 0.6431,
+      "step": 3082
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6006453381728236e-06,
+      "loss": 0.6543,
+      "step": 3084
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5901613960371584e-06,
+      "loss": 0.6629,
+      "step": 3086
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5796954836418886e-06,
+      "loss": 0.6761,
+      "step": 3088
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.569247626452842e-06,
+      "loss": 0.6178,
+      "step": 3090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.558817849891918e-06,
+      "loss": 0.6512,
+      "step": 3092
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.548406179337015e-06,
+      "loss": 0.6227,
+      "step": 3094
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5380126401219806e-06,
+      "loss": 0.6137,
+      "step": 3096
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.527637257536547e-06,
+      "loss": 0.6823,
+      "step": 3098
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.517280056826262e-06,
+      "loss": 0.6295,
+      "step": 3100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5069410631924383e-06,
+      "loss": 0.6753,
+      "step": 3102
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.496620301792082e-06,
+      "loss": 0.6056,
+      "step": 3104
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4863177977378393e-06,
+      "loss": 0.6788,
+      "step": 3106
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4760335760979313e-06,
+      "loss": 0.5443,
+      "step": 3108
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4657676618960944e-06,
+      "loss": 0.7115,
+      "step": 3110
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.455520080111522e-06,
+      "loss": 0.6888,
+      "step": 3112
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.445290855678791e-06,
+      "loss": 0.619,
+      "step": 3114
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4350800134878207e-06,
+      "loss": 0.5954,
+      "step": 3116
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.424887578383799e-06,
+      "loss": 0.6003,
+      "step": 3118
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.414713575167129e-06,
+      "loss": 0.6596,
+      "step": 3120
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4045580285933555e-06,
+      "loss": 0.6377,
+      "step": 3122
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.394420963373124e-06,
+      "loss": 0.6295,
+      "step": 3124
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3843024041721053e-06,
+      "loss": 0.6763,
+      "step": 3126
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3742023756109457e-06,
+      "loss": 0.6553,
+      "step": 3128
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3641209022651977e-06,
+      "loss": 0.5758,
+      "step": 3130
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3540580086652675e-06,
+      "loss": 0.6313,
+      "step": 3132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.344013719296353e-06,
+      "loss": 0.5763,
+      "step": 3134
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3339880585983844e-06,
+      "loss": 0.66,
+      "step": 3136
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.32398105096596e-06,
+      "loss": 0.6768,
+      "step": 3138
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.313992720748295e-06,
+      "loss": 0.604,
+      "step": 3140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.304023092249159e-06,
+      "loss": 0.6038,
+      "step": 3142
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2940721897268136e-06,
+      "loss": 0.6497,
+      "step": 3144
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2841400373939594e-06,
+      "loss": 0.6327,
+      "step": 3146
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.274226659417671e-06,
+      "loss": 0.6151,
+      "step": 3148
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2643320799193404e-06,
+      "loss": 0.5812,
+      "step": 3150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.254456322974622e-06,
+      "loss": 0.5965,
+      "step": 3152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.244599412613371e-06,
+      "loss": 0.561,
+      "step": 3154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.234761372819577e-06,
+      "loss": 0.6308,
+      "step": 3156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2249422275313214e-06,
+      "loss": 0.6631,
+      "step": 3158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2151420006407144e-06,
+      "loss": 0.6207,
+      "step": 3160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2053607159938195e-06,
+      "loss": 0.5989,
+      "step": 3162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1955983973906234e-06,
+      "loss": 0.6603,
+      "step": 3164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1858550685849577e-06,
+      "loss": 0.6944,
+      "step": 3166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.17613075328445e-06,
+      "loss": 0.6114,
+      "step": 3168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1664254751504643e-06,
+      "loss": 0.7135,
+      "step": 3170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1567392577980393e-06,
+      "loss": 0.6546,
+      "step": 3172
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1470721247958402e-06,
+      "loss": 0.66,
+      "step": 3174
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.137424099666091e-06,
+      "loss": 0.6323,
+      "step": 3176
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1277952058845286e-06,
+      "loss": 0.6107,
+      "step": 3178
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.118185466880327e-06,
+      "loss": 0.6075,
+      "step": 3180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.5804,
+      "step": 3182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.0990235466876516e-06,
+      "loss": 0.5956,
+      "step": 3184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.0894714121242743e-06,
+      "loss": 0.6373,
+      "step": 3186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.079938525588342e-06,
+      "loss": 0.6496,
+      "step": 3188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.0704249102754327e-06,
+      "loss": 0.653,
+      "step": 3190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.060930589334228e-06,
+      "loss": 0.6464,
+      "step": 3192
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0514555858664663e-06,
+      "loss": 0.6891,
+      "step": 3194
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0419999229268807e-06,
+      "loss": 0.6106,
+      "step": 3196
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.032563623523147e-06,
+      "loss": 0.6475,
+      "step": 3198
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0231467106158188e-06,
+      "loss": 0.6835,
+      "step": 3200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0137492071182862e-06,
+      "loss": 0.6393,
+      "step": 3202
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0043711358967045e-06,
+      "loss": 0.6599,
+      "step": 3204
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.995012519769951e-06,
+      "loss": 0.5575,
+      "step": 3206
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.985673381509565e-06,
+      "loss": 0.6775,
+      "step": 3208
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9763537438396894e-06,
+      "loss": 0.6792,
+      "step": 3210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9670536294370203e-06,
+      "loss": 0.6297,
+      "step": 3212
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9577730609307454e-06,
+      "loss": 0.6731,
+      "step": 3214
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9485120609024977e-06,
+      "loss": 0.5955,
+      "step": 3216
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9392706518862935e-06,
+      "loss": 0.5511,
+      "step": 3218
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9300488563684804e-06,
+      "loss": 0.615,
+      "step": 3220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9208466967876838e-06,
+      "loss": 0.6404,
+      "step": 3222
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9116641955347447e-06,
+      "loss": 0.6594,
+      "step": 3224
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9025013749526767e-06,
+      "loss": 0.7208,
+      "step": 3226
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8933582573366037e-06,
+      "loss": 0.5931,
+      "step": 3228
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8842348649337117e-06,
+      "loss": 0.6534,
+      "step": 3230
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.875131219943187e-06,
+      "loss": 0.5661,
+      "step": 3232
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8660473445161664e-06,
+      "loss": 0.6018,
+      "step": 3234
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.856983260755686e-06,
+      "loss": 0.611,
+      "step": 3236
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8479389907166224e-06,
+      "loss": 0.5963,
+      "step": 3238
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8389145564056387e-06,
+      "loss": 0.6733,
+      "step": 3240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.829909979781137e-06,
+      "loss": 0.6799,
+      "step": 3242
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.820925282753201e-06,
+      "loss": 0.6559,
+      "step": 3244
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8119604871835439e-06,
+      "loss": 0.6275,
+      "step": 3246
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8030156148854493e-06,
+      "loss": 0.6545,
+      "step": 3248
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7940906876237285e-06,
+      "loss": 0.6587,
+      "step": 3250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.78518572711466e-06,
+      "loss": 0.5917,
+      "step": 3252
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7763007550259392e-06,
+      "loss": 0.6909,
+      "step": 3254
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.767435792976626e-06,
+      "loss": 0.6103,
+      "step": 3256
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7585908625370908e-06,
+      "loss": 0.5358,
+      "step": 3258
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7497659852289629e-06,
+      "loss": 0.6556,
+      "step": 3260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7409611825250771e-06,
+      "loss": 0.6495,
+      "step": 3262
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7321764758494252e-06,
+      "loss": 0.6276,
+      "step": 3264
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7234118865770988e-06,
+      "loss": 0.6456,
+      "step": 3266
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7146674360342374e-06,
+      "loss": 0.6235,
+      "step": 3268
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7059431454979825e-06,
+      "loss": 0.6309,
+      "step": 3270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6972390361964197e-06,
+      "loss": 0.6477,
+      "step": 3272
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.688555129308531e-06,
+      "loss": 0.5945,
+      "step": 3274
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6798914459641435e-06,
+      "loss": 0.7061,
+      "step": 3276
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6712480072438665e-06,
+      "loss": 0.6144,
+      "step": 3278
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6626248341790596e-06,
+      "loss": 0.6135,
+      "step": 3280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6540219477517683e-06,
+      "loss": 0.6335,
+      "step": 3282
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6454393688946769e-06,
+      "loss": 0.6202,
+      "step": 3284
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6368771184910559e-06,
+      "loss": 0.5869,
+      "step": 3286
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6283352173747148e-06,
+      "loss": 0.6373,
+      "step": 3288
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6198136863299463e-06,
+      "loss": 0.6223,
+      "step": 3290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6113125460914759e-06,
+      "loss": 0.6037,
+      "step": 3292
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6028318173444202e-06,
+      "loss": 0.6253,
+      "step": 3294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.594371520724226e-06,
+      "loss": 0.5721,
+      "step": 3296
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5859316768166243e-06,
+      "loss": 0.6296,
+      "step": 3298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5775123061575836e-06,
+      "loss": 0.684,
+      "step": 3300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5691134292332522e-06,
+      "loss": 0.6273,
+      "step": 3302
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5607350664799158e-06,
+      "loss": 0.6815,
+      "step": 3304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.552377238283943e-06,
+      "loss": 0.6375,
+      "step": 3306
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5440399649817384e-06,
+      "loss": 0.619,
+      "step": 3308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5357232668596932e-06,
+      "loss": 0.6403,
+      "step": 3310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5274271641541293e-06,
+      "loss": 0.6767,
+      "step": 3312
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.519151677051265e-06,
+      "loss": 0.6177,
+      "step": 3314
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5108968256871437e-06,
+      "loss": 0.6535,
+      "step": 3316
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5026626301476088e-06,
+      "loss": 0.6536,
+      "step": 3318
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.494449110468238e-06,
+      "loss": 0.6237,
+      "step": 3320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4862562866343033e-06,
+      "loss": 0.6474,
+      "step": 3322
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4780841785807166e-06,
+      "loss": 0.6878,
+      "step": 3324
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4699328061919848e-06,
+      "loss": 0.5944,
+      "step": 3326
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4618021893021605e-06,
+      "loss": 0.5424,
+      "step": 3328
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4536923476947938e-06,
+      "loss": 0.6762,
+      "step": 3330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4456033011028836e-06,
+      "loss": 0.6121,
+      "step": 3332
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.437535069208833e-06,
+      "loss": 0.682,
+      "step": 3334
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4294876716443906e-06,
+      "loss": 0.6618,
+      "step": 3336
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4214611279906188e-06,
+      "loss": 0.6454,
+      "step": 3338
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4134554577778337e-06,
+      "loss": 0.6992,
+      "step": 3340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4054706804855634e-06,
+      "loss": 0.732,
+      "step": 3342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3975068155424975e-06,
+      "loss": 0.6452,
+      "step": 3344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3895638823264447e-06,
+      "loss": 0.7029,
+      "step": 3346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3816419001642778e-06,
+      "loss": 0.6145,
+      "step": 3348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3737408883318948e-06,
+      "loss": 0.569,
+      "step": 3350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.365860866054165e-06,
+      "loss": 0.7124,
+      "step": 3352
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.358001852504891e-06,
+      "loss": 0.6782,
+      "step": 3354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3501638668067485e-06,
+      "loss": 0.6796,
+      "step": 3356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3423469280312562e-06,
+      "loss": 0.5775,
+      "step": 3358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3345510551987128e-06,
+      "loss": 0.6854,
+      "step": 3360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.326776267278167e-06,
+      "loss": 0.6888,
+      "step": 3362
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3190225831873582e-06,
+      "loss": 0.618,
+      "step": 3364
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3112900217926784e-06,
+      "loss": 0.631,
+      "step": 3366
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3035786019091222e-06,
+      "loss": 0.6385,
+      "step": 3368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2958883423002423e-06,
+      "loss": 0.6888,
+      "step": 3370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2882192616781031e-06,
+      "loss": 0.617,
+      "step": 3372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.280571378703238e-06,
+      "loss": 0.5717,
+      "step": 3374
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2729447119846017e-06,
+      "loss": 0.7095,
+      "step": 3376
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.265339280079525e-06,
+      "loss": 0.654,
+      "step": 3378
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2577551014936651e-06,
+      "loss": 0.6813,
+      "step": 3380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2501921946809715e-06,
+      "loss": 0.6353,
+      "step": 3382
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2426505780436326e-06,
+      "loss": 0.6768,
+      "step": 3384
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2351302699320334e-06,
+      "loss": 0.6399,
+      "step": 3386
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2276312886447106e-06,
+      "loss": 0.6833,
+      "step": 3388
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2201536524283075e-06,
+      "loss": 0.6229,
+      "step": 3390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2126973794775343e-06,
+      "loss": 0.6262,
+      "step": 3392
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.6546,
+      "step": 3394
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1978489958917382e-06,
+      "loss": 0.6821,
+      "step": 3396
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1904569213860473e-06,
+      "loss": 0.6165,
+      "step": 3398
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1830862824045552e-06,
+      "loss": 0.6649,
+      "step": 3400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1757370968816217e-06,
+      "loss": 0.6165,
+      "step": 3402
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1684093826994026e-06,
+      "loss": 0.5745,
+      "step": 3404
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1611031576878118e-06,
+      "loss": 0.5688,
+      "step": 3406
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1538184396244777e-06,
+      "loss": 0.6317,
+      "step": 3408
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.146555246234694e-06,
+      "loss": 0.6626,
+      "step": 3410
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1393135951913826e-06,
+      "loss": 0.6329,
+      "step": 3412
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.132093504115046e-06,
+      "loss": 0.6357,
+      "step": 3414
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1248949905737283e-06,
+      "loss": 0.6817,
+      "step": 3416
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1177180720829694e-06,
+      "loss": 0.6552,
+      "step": 3418
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1105627661057671e-06,
+      "loss": 0.6066,
+      "step": 3420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1034290900525279e-06,
+      "loss": 0.6343,
+      "step": 3422
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.096317061281027e-06,
+      "loss": 0.6234,
+      "step": 3424
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.0892266970963706e-06,
+      "loss": 0.6055,
+      "step": 3426
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.082158014750948e-06,
+      "loss": 0.6729,
+      "step": 3428
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.0751110314443958e-06,
+      "loss": 0.5982,
+      "step": 3430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.068085764323543e-06,
+      "loss": 0.6599,
+      "step": 3432
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.0610822304823887e-06,
+      "loss": 0.5547,
+      "step": 3434
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0541004469620453e-06,
+      "loss": 0.6552,
+      "step": 3436
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0471404307507016e-06,
+      "loss": 0.6197,
+      "step": 3438
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0402021987835831e-06,
+      "loss": 0.5363,
+      "step": 3440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0332857679429097e-06,
+      "loss": 0.6879,
+      "step": 3442
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0263911550578532e-06,
+      "loss": 0.6494,
+      "step": 3444
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0195183769045015e-06,
+      "loss": 0.6298,
+      "step": 3446
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0126674502058054e-06,
+      "loss": 0.7105,
+      "step": 3448
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.005838391631555e-06,
+      "loss": 0.6143,
+      "step": 3450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.990312177983264e-07,
+      "loss": 0.6185,
+      "step": 3452
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.922459452694466e-07,
+      "loss": 0.6458,
+      "step": 3454
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.854825905549502e-07,
+      "loss": 0.6494,
+      "step": 3456
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.787411701115457e-07,
+      "loss": 0.6772,
+      "step": 3458
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.720217003425648e-07,
+      "loss": 0.5951,
+      "step": 3460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.65324197597931e-07,
+      "loss": 0.6596,
+      "step": 3462
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.586486781741212e-07,
+      "loss": 0.5474,
+      "step": 3464
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.519951583141129e-07,
+      "loss": 0.5937,
+      "step": 3466
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.453636542073629e-07,
+      "loss": 0.6288,
+      "step": 3468
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.38754181989755e-07,
+      "loss": 0.6382,
+      "step": 3470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.321667577435634e-07,
+      "loss": 0.6346,
+      "step": 3472
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.256013974974176e-07,
+      "loss": 0.6447,
+      "step": 3474
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.190581172262581e-07,
+      "loss": 0.6427,
+      "step": 3476
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.125369328513035e-07,
+      "loss": 0.578,
+      "step": 3478
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.060378602400055e-07,
+      "loss": 0.6327,
+      "step": 3480
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.995609152060136e-07,
+      "loss": 0.6683,
+      "step": 3482
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.931061135091356e-07,
+      "loss": 0.6191,
+      "step": 3484
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.866734708553015e-07,
+      "loss": 0.6603,
+      "step": 3486
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.802630028965242e-07,
+      "loss": 0.6266,
+      "step": 3488
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.738747252308555e-07,
+      "loss": 0.6027,
+      "step": 3490
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.675086534023591e-07,
+      "loss": 0.5987,
+      "step": 3492
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.611648029010645e-07,
+      "loss": 0.6035,
+      "step": 3494
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.548431891629317e-07,
+      "loss": 0.6564,
+      "step": 3496
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.485438275698154e-07,
+      "loss": 0.7294,
+      "step": 3498
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.42266733449425e-07,
+      "loss": 0.6077,
+      "step": 3500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.360119220752893e-07,
+      "loss": 0.6056,
+      "step": 3502
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.297794086667166e-07,
+      "loss": 0.6647,
+      "step": 3504
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.235692083887614e-07,
+      "loss": 0.6177,
+      "step": 3506
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.173813363521843e-07,
+      "loss": 0.5869,
+      "step": 3508
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.112158076134158e-07,
+      "loss": 0.6142,
+      "step": 3510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.050726371745221e-07,
+      "loss": 0.5782,
+      "step": 3512
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.989518399831642e-07,
+      "loss": 0.6378,
+      "step": 3514
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.928534309325675e-07,
+      "loss": 0.7053,
+      "step": 3516
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.867774248614801e-07,
+      "loss": 0.6973,
+      "step": 3518
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807238365541392e-07,
+      "loss": 0.6,
+      "step": 3520
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.746926807402344e-07,
+      "loss": 0.6411,
+      "step": 3522
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.686839720948735e-07,
+      "loss": 0.6457,
+      "step": 3524
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.626977252385437e-07,
+      "loss": 0.657,
+      "step": 3526
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.567339547370789e-07,
+      "loss": 0.6444,
+      "step": 3528
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.507926751016248e-07,
+      "loss": 0.6061,
+      "step": 3530
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.448739007885985e-07,
+      "loss": 0.5811,
+      "step": 3532
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.389776461996578e-07,
+      "loss": 0.6958,
+      "step": 3534
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.331039256816664e-07,
+      "loss": 0.6425,
+      "step": 3536
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.27252753526656e-07,
+      "loss": 0.6796,
+      "step": 3538
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.214241439717961e-07,
+      "loss": 0.6496,
+      "step": 3540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.156181111993543e-07,
+      "loss": 0.7017,
+      "step": 3542
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.098346693366642e-07,
+      "loss": 0.5806,
+      "step": 3544
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.040738324560914e-07,
+      "loss": 0.6105,
+      "step": 3546
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.983356145749976e-07,
+      "loss": 0.6146,
+      "step": 3548
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.926200296557084e-07,
+      "loss": 0.655,
+      "step": 3550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.869270916054782e-07,
+      "loss": 0.6617,
+      "step": 3552
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.812568142764576e-07,
+      "loss": 0.6233,
+      "step": 3554
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.756092114656587e-07,
+      "loss": 0.6708,
+      "step": 3556
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.699842969149195e-07,
+      "loss": 0.6631,
+      "step": 3558
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.64382084310875e-07,
+      "loss": 0.6319,
+      "step": 3560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.588025872849191e-07,
+      "loss": 0.608,
+      "step": 3562
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.532458194131763e-07,
+      "loss": 0.5455,
+      "step": 3564
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.477117942164657e-07,
+      "loss": 0.6194,
+      "step": 3566
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.422005251602659e-07,
+      "loss": 0.6055,
+      "step": 3568
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.367120256546888e-07,
+      "loss": 0.6543,
+      "step": 3570
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.312463090544396e-07,
+      "loss": 0.5635,
+      "step": 3572
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.258033886587911e-07,
+      "loss": 0.6396,
+      "step": 3574
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.203832777115449e-07,
+      "loss": 0.6902,
+      "step": 3576
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.149859894010035e-07,
+      "loss": 0.6541,
+      "step": 3578
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.096115368599364e-07,
+      "loss": 0.6941,
+      "step": 3580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.042599331655496e-07,
+      "loss": 0.6815,
+      "step": 3582
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.989311913394547e-07,
+      "loss": 0.6164,
+      "step": 3584
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.936253243476319e-07,
+      "loss": 0.592,
+      "step": 3586
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.883423451004033e-07,
+      "loss": 0.5829,
+      "step": 3588
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.830822664523994e-07,
+      "loss": 0.6957,
+      "step": 3590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.77845101202531e-07,
+      "loss": 0.5859,
+      "step": 3592
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.726308620939536e-07,
+      "loss": 0.6012,
+      "step": 3594
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.674395618140393e-07,
+      "loss": 0.6696,
+      "step": 3596
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.622712129943453e-07,
+      "loss": 0.571,
+      "step": 3598
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.571258282105829e-07,
+      "loss": 0.6771,
+      "step": 3600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.520034199825841e-07,
+      "loss": 0.6563,
+      "step": 3602
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.469040007742776e-07,
+      "loss": 0.6631,
+      "step": 3604
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.6118,
+      "step": 3606
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.36774178992735e-07,
+      "loss": 0.6129,
+      "step": 3608
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.317438010675468e-07,
+      "loss": 0.6661,
+      "step": 3610
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.267364614580861e-07,
+      "loss": 0.6081,
+      "step": 3612
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.217521723482943e-07,
+      "loss": 0.665,
+      "step": 3614
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.167909458660258e-07,
+      "loss": 0.644,
+      "step": 3616
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.118527940830164e-07,
+      "loss": 0.6077,
+      "step": 3618
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.069377290148603e-07,
+      "loss": 0.6896,
+      "step": 3620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.020457626209708e-07,
+      "loss": 0.6148,
+      "step": 3622
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.971769068045629e-07,
+      "loss": 0.61,
+      "step": 3624
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.923311734126135e-07,
+      "loss": 0.642,
+      "step": 3626
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.875085742358432e-07,
+      "loss": 0.5783,
+      "step": 3628
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.827091210086776e-07,
+      "loss": 0.5715,
+      "step": 3630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.779328254092253e-07,
+      "loss": 0.602,
+      "step": 3632
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.731796990592452e-07,
+      "loss": 0.6423,
+      "step": 3634
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.6844975352412503e-07,
+      "loss": 0.5525,
+      "step": 3636
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.637430003128429e-07,
+      "loss": 0.6141,
+      "step": 3638
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5905945087795e-07,
+      "loss": 0.551,
+      "step": 3640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.543991166155337e-07,
+      "loss": 0.6439,
+      "step": 3642
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.497620088651966e-07,
+      "loss": 0.6968,
+      "step": 3644
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.451481389100232e-07,
+      "loss": 0.6661,
+      "step": 3646
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4055751797655865e-07,
+      "loss": 0.6425,
+      "step": 3648
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.359901572347758e-07,
+      "loss": 0.6854,
+      "step": 3650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3144606779805363e-07,
+      "loss": 0.5804,
+      "step": 3652
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.2692526072314224e-07,
+      "loss": 0.6967,
+      "step": 3654
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.2242774701014453e-07,
+      "loss": 0.7088,
+      "step": 3656
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1795353760248567e-07,
+      "loss": 0.5739,
+      "step": 3658
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.135026433868827e-07,
+      "loss": 0.6011,
+      "step": 3660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0907507519332477e-07,
+      "loss": 0.6008,
+      "step": 3662
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.046708437950464e-07,
+      "loss": 0.612,
+      "step": 3664
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0028995990849086e-07,
+      "loss": 0.64,
+      "step": 3666
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9593243419329907e-07,
+      "loss": 0.6527,
+      "step": 3668
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9159827725227196e-07,
+      "loss": 0.6117,
+      "step": 3670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8728749963135136e-07,
+      "loss": 0.6583,
+      "step": 3672
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.8300011181959364e-07,
+      "loss": 0.6293,
+      "step": 3674
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.787361242491394e-07,
+      "loss": 0.6873,
+      "step": 3676
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.744955472951928e-07,
+      "loss": 0.5988,
+      "step": 3678
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.702783912759955e-07,
+      "loss": 0.5805,
+      "step": 3680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.660846664528006e-07,
+      "loss": 0.6005,
+      "step": 3682
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.619143830298477e-07,
+      "loss": 0.6282,
+      "step": 3684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.577675511543388e-07,
+      "loss": 0.6577,
+      "step": 3686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5364418091641374e-07,
+      "loss": 0.5849,
+      "step": 3688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.4954428234912243e-07,
+      "loss": 0.6813,
+      "step": 3690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.4546786542840604e-07,
+      "loss": 0.6399,
+      "step": 3692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.414149400730682e-07,
+      "loss": 0.5792,
+      "step": 3694
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3738551614475477e-07,
+      "loss": 0.6974,
+      "step": 3696
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.333796034479242e-07,
+      "loss": 0.6802,
+      "step": 3698
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.293972117298294e-07,
+      "loss": 0.6719,
+      "step": 3700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.254383506804926e-07,
+      "loss": 0.6227,
+      "step": 3702
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.21503029932676e-07,
+      "loss": 0.6234,
+      "step": 3704
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1759125906186796e-07,
+      "loss": 0.6487,
+      "step": 3706
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1370304758625347e-07,
+      "loss": 0.6597,
+      "step": 3708
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.098384049666925e-07,
+      "loss": 0.6704,
+      "step": 3710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.059973406066963e-07,
+      "loss": 0.6164,
+      "step": 3712
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0217986385240536e-07,
+      "loss": 0.58,
+      "step": 3714
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9838598399256623e-07,
+      "loss": 0.7162,
+      "step": 3716
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.94615710258509e-07,
+      "loss": 0.6323,
+      "step": 3718
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.908690518241275e-07,
+      "loss": 0.559,
+      "step": 3720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.871460178058494e-07,
+      "loss": 0.6824,
+      "step": 3722
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.834466172626238e-07,
+      "loss": 0.6303,
+      "step": 3724
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7977085919589253e-07,
+      "loss": 0.5992,
+      "step": 3726
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.76118752549569e-07,
+      "loss": 0.6421,
+      "step": 3728
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7249030621001925e-07,
+      "loss": 0.6042,
+      "step": 3730
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6888552900603993e-07,
+      "loss": 0.6279,
+      "step": 3732
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.653044297088314e-07,
+      "loss": 0.5844,
+      "step": 3734
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.617470170319847e-07,
+      "loss": 0.6478,
+      "step": 3736
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.5821329963145346e-07,
+      "loss": 0.6808,
+      "step": 3738
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5470328610553764e-07,
+      "loss": 0.6273,
+      "step": 3740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.512169849948576e-07,
+      "loss": 0.5835,
+      "step": 3742
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.477544047823399e-07,
+      "loss": 0.6861,
+      "step": 3744
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.443155538931907e-07,
+      "loss": 0.6322,
+      "step": 3746
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.409004406948778e-07,
+      "loss": 0.6361,
+      "step": 3748
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3750907349711084e-07,
+      "loss": 0.6046,
+      "step": 3750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.341414605518211e-07,
+      "loss": 0.6514,
+      "step": 3752
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.307976100531384e-07,
+      "loss": 0.5473,
+      "step": 3754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.2747753013737438e-07,
+      "loss": 0.6101,
+      "step": 3756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.2418122888300032e-07,
+      "loss": 0.6209,
+      "step": 3758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2090871431063256e-07,
+      "loss": 0.6046,
+      "step": 3760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.176599943830071e-07,
+      "loss": 0.5897,
+      "step": 3762
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1443507700495968e-07,
+      "loss": 0.6813,
+      "step": 3764
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.112339700234156e-07,
+      "loss": 0.5967,
+      "step": 3766
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0805668122735767e-07,
+      "loss": 0.5652,
+      "step": 3768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0490321834781835e-07,
+      "loss": 0.6635,
+      "step": 3770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0177358905785538e-07,
+      "loss": 0.5666,
+      "step": 3772
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.986678009725329e-07,
+      "loss": 0.5857,
+      "step": 3774
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9558586164890592e-07,
+      "loss": 0.6268,
+      "step": 3776
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9252777858599913e-07,
+      "loss": 0.6103,
+      "step": 3778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.894935592247915e-07,
+      "loss": 0.5389,
+      "step": 3780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8648321094819288e-07,
+      "loss": 0.665,
+      "step": 3782
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8349674108103287e-07,
+      "loss": 0.6127,
+      "step": 3784
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8053415689003872e-07,
+      "loss": 0.6544,
+      "step": 3786
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7759546558381967e-07,
+      "loss": 0.5639,
+      "step": 3788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7468067431284708e-07,
+      "loss": 0.6099,
+      "step": 3790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7178979016943765e-07,
+      "loss": 0.55,
+      "step": 3792
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.689228201877391e-07,
+      "loss": 0.6034,
+      "step": 3794
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.660797713437079e-07,
+      "loss": 0.6337,
+      "step": 3796
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6326065055510042e-07,
+      "loss": 0.6308,
+      "step": 3798
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6046546468144408e-07,
+      "loss": 0.6309,
+      "step": 3800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5769422052403172e-07,
+      "loss": 0.6653,
+      "step": 3802
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5494692482590057e-07,
+      "loss": 0.6768,
+      "step": 3804
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.522235842718156e-07,
+      "loss": 0.7077,
+      "step": 3806
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4952420548825286e-07,
+      "loss": 0.6111,
+      "step": 3808
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.468487950433839e-07,
+      "loss": 0.5719,
+      "step": 3810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.441973594470636e-07,
+      "loss": 0.5759,
+      "step": 3812
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4156990515080682e-07,
+      "loss": 0.6631,
+      "step": 3814
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.3896643854777846e-07,
+      "loss": 0.616,
+      "step": 3816
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.623,
+      "step": 3818
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.338314937022145e-07,
+      "loss": 0.6053,
+      "step": 3820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.313000279541121e-07,
+      "loss": 0.6248,
+      "step": 3822
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.287925748880703e-07,
+      "loss": 0.6376,
+      "step": 3824
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2630914060526523e-07,
+      "loss": 0.5939,
+      "step": 3826
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2384973114843103e-07,
+      "loss": 0.631,
+      "step": 3828
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2141435250184186e-07,
+      "loss": 0.6548,
+      "step": 3830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1900301059130092e-07,
+      "loss": 0.6226,
+      "step": 3832
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1661571128412597e-07,
+      "loss": 0.6512,
+      "step": 3834
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1425246038913151e-07,
+      "loss": 0.639,
+      "step": 3836
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1191326365661891e-07,
+      "loss": 0.6667,
+      "step": 3838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0959812677835968e-07,
+      "loss": 0.6857,
+      "step": 3840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0730705538758324e-07,
+      "loss": 0.6044,
+      "step": 3842
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0504005505896141e-07,
+      "loss": 0.6689,
+      "step": 3844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0279713130859515e-07,
+      "loss": 0.7265,
+      "step": 3846
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0057828959400551e-07,
+      "loss": 0.6334,
+      "step": 3848
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.838353531411272e-08,
+      "loss": 0.6233,
+      "step": 3850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.62128738092294e-08,
+      "loss": 0.6059,
+      "step": 3852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.406631036104507e-08,
+      "loss": 0.5761,
+      "step": 3854
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.194385019261287e-08,
+      "loss": 0.7302,
+      "step": 3856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.984549846833612e-08,
+      "loss": 0.6299,
+      "step": 3858
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.777126029396066e-08,
+      "loss": 0.6111,
+      "step": 3860
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.57211407165548e-08,
+      "loss": 0.559,
+      "step": 3862
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.36951447245038e-08,
+      "loss": 0.5878,
+      "step": 3864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.169327724749543e-08,
+      "loss": 0.5786,
+      "step": 3866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.971554315650443e-08,
+      "loss": 0.655,
+      "step": 3868
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.776194726378583e-08,
+      "loss": 0.6681,
+      "step": 3870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.583249432286277e-08,
+      "loss": 0.5854,
+      "step": 3872
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.392718902850981e-08,
+      "loss": 0.7254,
+      "step": 3874
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.204603601674853e-08,
+      "loss": 0.6167,
+      "step": 3876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.018903986483083e-08,
+      "loss": 0.5869,
+      "step": 3878
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.835620509122898e-08,
+      "loss": 0.6055,
+      "step": 3880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.65475361556267e-08,
+      "loss": 0.75,
+      "step": 3882
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.47630374589081e-08,
+      "loss": 0.6447,
+      "step": 3884
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.300271334314434e-08,
+      "loss": 0.6669,
+      "step": 3886
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.12665680915836e-08,
+      "loss": 0.6791,
+      "step": 3888
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955460592864337e-08,
+      "loss": 0.6087,
+      "step": 3890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.786683101989821e-08,
+      "loss": 0.6343,
+      "step": 3892
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.6203247472070844e-08,
+      "loss": 0.6591,
+      "step": 3894
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4563859333017776e-08,
+      "loss": 0.6753,
+      "step": 3896
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.294867059172593e-08,
+      "loss": 0.6953,
+      "step": 3898
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.13576851782982e-08,
+      "loss": 0.5878,
+      "step": 3900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.979090696394795e-08,
+      "loss": 0.6771,
+      "step": 3902
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.824833976098453e-08,
+      "loss": 0.6432,
+      "step": 3904
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.672998732280776e-08,
+      "loss": 0.6002,
+      "step": 3906
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.52358533438968e-08,
+      "loss": 0.5957,
+      "step": 3908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3765941459804616e-08,
+      "loss": 0.6148,
+      "step": 3910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.232025524714356e-08,
+      "loss": 0.5838,
+      "step": 3912
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0898798223582006e-08,
+      "loss": 0.6766,
+      "step": 3914
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.950157384783104e-08,
+      "loss": 0.6504,
+      "step": 3916
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.812858551964005e-08,
+      "loss": 0.6663,
+      "step": 3918
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.677983657978779e-08,
+      "loss": 0.6125,
+      "step": 3920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5455330310071314e-08,
+      "loss": 0.6286,
+      "step": 3922
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4155069933301535e-08,
+      "loss": 0.6218,
+      "step": 3924
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2879058613292106e-08,
+      "loss": 0.6801,
+      "step": 3926
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.16272994548561e-08,
+      "loss": 0.6241,
+      "step": 3928
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0399795503793795e-08,
+      "loss": 0.6436,
+      "step": 3930
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9196549746888237e-08,
+      "loss": 0.6098,
+      "step": 3932
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.801756511189524e-08,
+      "loss": 0.6534,
+      "step": 3934
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6862844467540062e-08,
+      "loss": 0.6356,
+      "step": 3936
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5732390623509628e-08,
+      "loss": 0.6533,
+      "step": 3938
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.462620633044033e-08,
+      "loss": 0.6534,
+      "step": 3940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3544294279918002e-08,
+      "loss": 0.6567,
+      "step": 3942
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2486657104471288e-08,
+      "loss": 0.606,
+      "step": 3944
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1453297377557193e-08,
+      "loss": 0.694,
+      "step": 3946
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.044421761356552e-08,
+      "loss": 0.6462,
+      "step": 3948
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9459420267804452e-08,
+      "loss": 0.6106,
+      "step": 3950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8498907736499428e-08,
+      "loss": 0.6251,
+      "step": 3952
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7562682356786488e-08,
+      "loss": 0.6783,
+      "step": 3954
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6650746406702277e-08,
+      "loss": 0.6726,
+      "step": 3956
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.576310210518517e-08,
+      "loss": 0.5934,
+      "step": 3958
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.489975161206636e-08,
+      "loss": 0.636,
+      "step": 3960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4060697028063231e-08,
+      "loss": 0.5639,
+      "step": 3962
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3245940394778223e-08,
+      "loss": 0.6345,
+      "step": 3964
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2455483694689962e-08,
+      "loss": 0.561,
+      "step": 3966
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1689328851151038e-08,
+      "loss": 0.6328,
+      "step": 3968
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.094747772838134e-08,
+      "loss": 0.57,
+      "step": 3970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0229932131465836e-08,
+      "loss": 0.643,
+      "step": 3972
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.53669380634792e-09,
+      "loss": 0.6187,
+      "step": 3974
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.867764439826066e-09,
+      "loss": 0.6811,
+      "step": 3976
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.223145659550513e-09,
+      "loss": 0.5876,
+      "step": 3978
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.602839034017705e-09,
+      "loss": 0.6392,
+      "step": 3980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0068460725680746e-09,
+      "loss": 0.6035,
+      "step": 3982
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.435168225381594e-09,
+      "loss": 0.6092,
+      "step": 3984
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.887806883474456e-09,
+      "loss": 0.6194,
+      "step": 3986
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.364763378694626e-09,
+      "loss": 0.6296,
+      "step": 3988
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.8660389837207336e-09,
+      "loss": 0.7165,
+      "step": 3990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.39163491205652e-09,
+      "loss": 0.6239,
+      "step": 3992
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.9415523180297285e-09,
+      "loss": 0.6724,
+      "step": 3994
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.5157922967898885e-09,
+      "loss": 0.6761,
+      "step": 3996
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.114355884301645e-09,
+      "loss": 0.6321,
+      "step": 3998
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7372440573469884e-09,
+      "loss": 0.6319,
+      "step": 4000
+    }
+  ],
+  "max_steps": 4030,
+  "num_train_epochs": 2,
+  "total_flos": 835628442419200.0,
+  "trial_name": null,
+  "trial_params": null
+}