diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9781 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 1626,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 1.2646,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 1.1865,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 1.209,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 1.1436,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.0659,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.0713,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.8657,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 0.5254,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 0.4575,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 0.386,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 0.2015,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 0.2046,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 0.1884,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.1812,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 0.2687,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 0.2764,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 0.2101,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 0.1816,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 0.1924,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 0.1964,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.1799,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 0.1705,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 0.1851,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 0.1763,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0204081632653063e-05,
+      "loss": 0.2079,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0612244897959186e-05,
+      "loss": 0.1682,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 0.1788,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.1274,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 0.1627,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 0.1804,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2653061224489798e-05,
+      "loss": 0.1431,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 0.1346,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 0.1613,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 0.1469,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.1307,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 0.1355,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 0.1324,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 0.1292,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 0.1107,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 0.1312,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 0.1328,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.1159,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 0.1141,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.795918367346939e-05,
+      "loss": 0.132,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 0.1318,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 0.105,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 0.1204,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 0.1407,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999980157050852e-05,
+      "loss": 0.1264,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999992062828215e-05,
+      "loss": 0.1395,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999821413930146e-05,
+      "loss": 0.1053,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999968251438858e-05,
+      "loss": 0.1467,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999950393020868e-05,
+      "loss": 0.1345,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999285662099182e-05,
+      "loss": 0.1335,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99990277109263e-05,
+      "loss": 0.1137,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998730077713732e-05,
+      "loss": 0.1348,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999839276364267e-05,
+      "loss": 0.1172,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998015770051766e-05,
+      "loss": 0.1409,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997599098437162e-05,
+      "loss": 0.1337,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999714275045245e-05,
+      "loss": 0.1395,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999664672790869e-05,
+      "loss": 0.1086,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996111032774398e-05,
+      "loss": 0.1068,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995535667175517e-05,
+      "loss": 0.1271,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994920633395445e-05,
+      "loss": 0.1198,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994265933875e-05,
+      "loss": 0.1198,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993571571212408e-05,
+      "loss": 0.1202,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992837548163315e-05,
+      "loss": 0.0931,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992063867640757e-05,
+      "loss": 0.1003,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991250532715158e-05,
+      "loss": 0.1183,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99903975466143e-05,
+      "loss": 0.1287,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989504912723347e-05,
+      "loss": 0.1229,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988572634584792e-05,
+      "loss": 0.1257,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987600715898462e-05,
+      "loss": 0.0889,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986589160521508e-05,
+      "loss": 0.0991,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985537972468374e-05,
+      "loss": 0.1008,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984447155910796e-05,
+      "loss": 0.0897,
+      "step": 77
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983316715177784e-05,
+      "loss": 0.1328,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998214665475558e-05,
+      "loss": 0.12,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980936979287685e-05,
+      "loss": 0.098,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99796876935748e-05,
+      "loss": 0.1141,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997839880257483e-05,
+      "loss": 0.1628,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977070311402854e-05,
+      "loss": 0.0978,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975702225331108e-05,
+      "loss": 0.1315,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974294549788964e-05,
+      "loss": 0.1412,
+      "step": 85
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972847290362906e-05,
+      "loss": 0.1265,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971360452796523e-05,
+      "loss": 0.1239,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9969834042990452e-05,
+      "loss": 0.1092,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968268067002396e-05,
+      "loss": 0.132,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966662531047065e-05,
+      "loss": 0.1251,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965017441496174e-05,
+      "loss": 0.1124,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996333280487841e-05,
+      "loss": 0.1107,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996160862787941e-05,
+      "loss": 0.1136,
+      "step": 93
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995984491734172e-05,
+      "loss": 0.1216,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958041680264777e-05,
+      "loss": 0.0949,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956198923804898e-05,
+      "loss": 0.1031,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995431665527523e-05,
+      "loss": 0.0992,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952394882145718e-05,
+      "loss": 0.112,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950433612043092e-05,
+      "loss": 0.111,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994843285275083e-05,
+      "loss": 0.0984,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994639261220913e-05,
+      "loss": 0.0939,
+      "step": 101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944312898514862e-05,
+      "loss": 0.0968,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9942193719921557e-05,
+      "loss": 0.1065,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994003508483937e-05,
+      "loss": 0.0937,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937837001835038e-05,
+      "loss": 0.1018,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993559947963185e-05,
+      "loss": 0.1007,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933322527109614e-05,
+      "loss": 0.1023,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931006153304618e-05,
+      "loss": 0.1031,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928650367409602e-05,
+      "loss": 0.1252,
+      "step": 109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9926255178773713e-05,
+      "loss": 0.0925,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923820596902473e-05,
+      "loss": 0.13,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921346631457737e-05,
+      "loss": 0.0961,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918833292257663e-05,
+      "loss": 0.0994,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991628058927666e-05,
+      "loss": 0.0879,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991368853264536e-05,
+      "loss": 0.0966,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991105713265057e-05,
+      "loss": 0.105,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908386399735243e-05,
+      "loss": 0.1266,
+      "step": 117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990567634449842e-05,
+      "loss": 0.1223,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902926977695196e-05,
+      "loss": 0.1089,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9900138310236683e-05,
+      "loss": 0.1144,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897310353189958e-05,
+      "loss": 0.0997,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9894443117778022e-05,
+      "loss": 0.0906,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9891536615379756e-05,
+      "loss": 0.0904,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888590857529878e-05,
+      "loss": 0.1114,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885605855918887e-05,
+      "loss": 0.0946,
+      "step": 125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9882581622393036e-05,
+      "loss": 0.0934,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9879518168954265e-05,
+      "loss": 0.088,
+      "step": 127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9876415507760164e-05,
+      "loss": 0.096,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9873273651123925e-05,
+      "loss": 0.0897,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870092611514287e-05,
+      "loss": 0.0795,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986687240155549e-05,
+      "loss": 0.0791,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 0.0919,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986031452186459e-05,
+      "loss": 0.0885,
+      "step": 133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9856976878158024e-05,
+      "loss": 0.0889,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985360011615326e-05,
+      "loss": 0.0799,
+      "step": 135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9850184249251298e-05,
+      "loss": 0.1195,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9846729291008293e-05,
+      "loss": 0.1328,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843235255135572e-05,
+      "loss": 0.0981,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983970215549952e-05,
+      "loss": 0.0985,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9836130006121565e-05,
+      "loss": 0.1119,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9832518821178103e-05,
+      "loss": 0.1516,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9828868615000447e-05,
+      "loss": 0.0893,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982517940207476e-05,
+      "loss": 0.0882,
+      "step": 143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9821451197042028e-05,
+      "loss": 0.1068,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981768401469796e-05,
+      "loss": 0.0942,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9813877869992955e-05,
+      "loss": 0.0942,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9810032778032043e-05,
+      "loss": 0.0958,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980614875407482e-05,
+      "loss": 0.0868,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9802225813535383e-05,
+      "loss": 0.1014,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979826397198227e-05,
+      "loss": 0.0912,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9794263245138406e-05,
+      "loss": 0.1119,
+      "step": 151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9790223648881036e-05,
+      "loss": 0.086,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978614519924166e-05,
+      "loss": 0.1036,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9782027912405973e-05,
+      "loss": 0.0825,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97778718047138e-05,
+      "loss": 0.1098,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9773676892659025e-05,
+      "loss": 0.0965,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976944319288953e-05,
+      "loss": 0.0931,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9765170722207135e-05,
+      "loss": 0.0986,
+      "step": 158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.976085949756753e-05,
+      "loss": 0.1097,
+      "step": 159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9756509536080187e-05,
+      "loss": 0.1147,
+      "step": 160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9752120855008326e-05,
+      "loss": 0.0898,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974769347176882e-05,
+      "loss": 0.1077,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9743227403932135e-05,
+      "loss": 0.092,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9738722669222268e-05,
+      "loss": 0.1033,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9734179285516657e-05,
+      "loss": 0.1052,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9729597270846134e-05,
+      "loss": 0.1092,
+      "step": 166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972497664339483e-05,
+      "loss": 0.1016,
+      "step": 167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9720317421500123e-05,
+      "loss": 0.1117,
+      "step": 168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9715619623652554e-05,
+      "loss": 0.0913,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971088326849576e-05,
+      "loss": 0.0987,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9706108374826383e-05,
+      "loss": 0.1089,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9701294961594022e-05,
+      "loss": 0.0965,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969644304790114e-05,
+      "loss": 0.1017,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9691552653002993e-05,
+      "loss": 0.1177,
+      "step": 174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968662379630755e-05,
+      "loss": 0.1043,
+      "step": 175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9681656497375427e-05,
+      "loss": 0.1058,
+      "step": 176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967665077591979e-05,
+      "loss": 0.108,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9671606651806293e-05,
+      "loss": 0.0977,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9666524145053004e-05,
+      "loss": 0.1074,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96614032758303e-05,
+      "loss": 0.11,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965624406446081e-05,
+      "loss": 0.0941,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9651046531419335e-05,
+      "loss": 0.0945,
+      "step": 182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9645810697332747e-05,
+      "loss": 0.0912,
+      "step": 183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9640536582979924e-05,
+      "loss": 0.104,
+      "step": 184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963522420929166e-05,
+      "loss": 0.1095,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.96298735973506e-05,
+      "loss": 0.1027,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9624484768391106e-05,
+      "loss": 0.0943,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961905774379925e-05,
+      "loss": 0.1064,
+      "step": 188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9613592545112657e-05,
+      "loss": 0.0957,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960808919402046e-05,
+      "loss": 0.0861,
+      "step": 190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9602547712363205e-05,
+      "loss": 0.0869,
+      "step": 191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9596968122132757e-05,
+      "loss": 0.118,
+      "step": 192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9591350445472218e-05,
+      "loss": 0.0942,
+      "step": 193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958569470467585e-05,
+      "loss": 0.0959,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9580000922188967e-05,
+      "loss": 0.0911,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957426912060785e-05,
+      "loss": 0.0946,
+      "step": 196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9568499322679676e-05,
+      "loss": 0.0888,
+      "step": 197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95626915513024e-05,
+      "loss": 0.1021,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9556845829524683e-05,
+      "loss": 0.1072,
+      "step": 199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9550962180545807e-05,
+      "loss": 0.1157,
+      "step": 200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9545040627715554e-05,
+      "loss": 0.0898,
+      "step": 201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953908119453414e-05,
+      "loss": 0.0882,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953308390465211e-05,
+      "loss": 0.0879,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527048781870248e-05,
+      "loss": 0.0979,
+      "step": 204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520975850139485e-05,
+      "loss": 0.1176,
+      "step": 205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9514865133560788e-05,
+      "loss": 0.0724,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950871665638509e-05,
+      "loss": 0.0881,
+      "step": 207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950253044301318e-05,
+      "loss": 0.0869,
+      "step": 208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949630651799559e-05,
+      "loss": 0.096,
+      "step": 209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949004490603253e-05,
+      "loss": 0.1387,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9483745631973777e-05,
+      "loss": 0.111,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9477408720818553e-05,
+      "loss": 0.0874,
+      "step": 212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947103419771547e-05,
+      "loss": 0.0999,
+      "step": 213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946462208796239e-05,
+      "loss": 0.09,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 0.1143,
+      "step": 215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9451685210443443e-05,
+      "loss": 0.0936,
+      "step": 216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9445160494018734e-05,
+      "loss": 0.0827,
+      "step": 217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9438598293626148e-05,
+      "loss": 0.0919,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9431998635308372e-05,
+      "loss": 0.1132,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942536154525673e-05,
+      "loss": 0.1217,
+      "step": 220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9418687049811116e-05,
+      "loss": 0.0837,
+      "step": 221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9411975175459864e-05,
+      "loss": 0.0896,
+      "step": 222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9405225948839657e-05,
+      "loss": 0.1046,
+      "step": 223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9398439396735396e-05,
+      "loss": 0.0971,
+      "step": 224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9391615546080132e-05,
+      "loss": 0.1008,
+      "step": 225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9384754423954927e-05,
+      "loss": 0.0933,
+      "step": 226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9377856057588756e-05,
+      "loss": 0.0978,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937092047435841e-05,
+      "loss": 0.0978,
+      "step": 228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9363947701788374e-05,
+      "loss": 0.0815,
+      "step": 229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9356937767550715e-05,
+      "loss": 0.1035,
+      "step": 230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9349890699464997e-05,
+      "loss": 0.1002,
+      "step": 231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934280652549814e-05,
+      "loss": 0.1176,
+      "step": 232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9335685273764322e-05,
+      "loss": 0.1075,
+      "step": 233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932852697252487e-05,
+      "loss": 0.0876,
+      "step": 234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932133165018815e-05,
+      "loss": 0.0895,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931409933530944e-05,
+      "loss": 0.0905,
+      "step": 236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9306830056590832e-05,
+      "loss": 0.0938,
+      "step": 237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9299523842881117e-05,
+      "loss": 0.0899,
+      "step": 238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9292180723175656e-05,
+      "loss": 0.0771,
+      "step": 239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9284800726616276e-05,
+      "loss": 0.1033,
+      "step": 240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9277383882491162e-05,
+      "loss": 0.0999,
+      "step": 241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926993022023472e-05,
+      "loss": 0.0803,
+      "step": 242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9262439769427488e-05,
+      "loss": 0.1154,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9254912559795984e-05,
+      "loss": 0.0823,
+      "step": 244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924734862121262e-05,
+      "loss": 0.0935,
+      "step": 245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9239747983695564e-05,
+      "loss": 0.0869,
+      "step": 246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9232110677408625e-05,
+      "loss": 0.097,
+      "step": 247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9224436732661148e-05,
+      "loss": 0.09,
+      "step": 248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9216726179907866e-05,
+      "loss": 0.0753,
+      "step": 249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9208979049748808e-05,
+      "loss": 0.0889,
+      "step": 250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920119537292914e-05,
+      "loss": 0.0893,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9193375180339094e-05,
+      "loss": 0.0935,
+      "step": 252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918551850301381e-05,
+      "loss": 0.0858,
+      "step": 253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9177625372133208e-05,
+      "loss": 0.0898,
+      "step": 254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9169695819021893e-05,
+      "loss": 0.0827,
+      "step": 255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9161729875149006e-05,
+      "loss": 0.0902,
+      "step": 256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9153727572128113e-05,
+      "loss": 0.0881,
+      "step": 257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9145688941717074e-05,
+      "loss": 0.1011,
+      "step": 258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9137614015817913e-05,
+      "loss": 0.0789,
+      "step": 259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9129502826476697e-05,
+      "loss": 0.0945,
+      "step": 260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9121355405883416e-05,
+      "loss": 0.0928,
+      "step": 261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911317178637183e-05,
+      "loss": 0.0818,
+      "step": 262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910495200041938e-05,
+      "loss": 0.0779,
+      "step": 263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.909669608064702e-05,
+      "loss": 0.0792,
+      "step": 264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9088404059819106e-05,
+      "loss": 0.1046,
+      "step": 265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908007597084327e-05,
+      "loss": 0.0768,
+      "step": 266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907171184677028e-05,
+      "loss": 0.0875,
+      "step": 267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906331172079392e-05,
+      "loss": 0.0781,
+      "step": 268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9054875626250836e-05,
+      "loss": 0.1063,
+      "step": 269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9046403596620432e-05,
+      "loss": 0.0774,
+      "step": 270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9037895665524716e-05,
+      "loss": 0.0821,
+      "step": 271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.902935186672818e-05,
+      "loss": 0.0793,
+      "step": 272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9020772234137656e-05,
+      "loss": 0.0998,
+      "step": 273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9012156801802184e-05,
+      "loss": 0.1079,
+      "step": 274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9003505603912884e-05,
+      "loss": 0.1056,
+      "step": 275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899481867480281e-05,
+      "loss": 0.103,
+      "step": 276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8986096048946826e-05,
+      "loss": 0.0781,
+      "step": 277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8977337760961444e-05,
+      "loss": 0.0867,
+      "step": 278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896854384560473e-05,
+      "loss": 0.1021,
+      "step": 279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.895971433777612e-05,
+      "loss": 0.0839,
+      "step": 280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.895084927251631e-05,
+      "loss": 0.105,
+      "step": 281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894194868500711e-05,
+      "loss": 0.1093,
+      "step": 282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8933012610571295e-05,
+      "loss": 0.11,
+      "step": 283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8924041084672486e-05,
+      "loss": 0.0906,
+      "step": 284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8915034142914988e-05,
+      "loss": 0.0861,
+      "step": 285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8905991821043656e-05,
+      "loss": 0.0892,
+      "step": 286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8896914154943756e-05,
+      "loss": 0.1085,
+      "step": 287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8887801180640826e-05,
+      "loss": 0.0923,
+      "step": 288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.887865293430052e-05,
+      "loss": 0.0963,
+      "step": 289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8869469452228476e-05,
+      "loss": 0.1025,
+      "step": 290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8860250770870167e-05,
+      "loss": 0.098,
+      "step": 291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885099692681076e-05,
+      "loss": 0.0775,
+      "step": 292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8841707956774965e-05,
+      "loss": 0.0848,
+      "step": 293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8832383897626892e-05,
+      "loss": 0.0968,
+      "step": 294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882302478636991e-05,
+      "loss": 0.1071,
+      "step": 295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881363066014649e-05,
+      "loss": 0.1039,
+      "step": 296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8804201556238068e-05,
+      "loss": 0.086,
+      "step": 297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 0.093,
+      "step": 298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878523856518587e-05,
+      "loss": 0.1065,
+      "step": 299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8775704753298423e-05,
+      "loss": 0.0929,
+      "step": 300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876613611423834e-05,
+      "loss": 0.099,
+      "step": 301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875653268597963e-05,
+      "loss": 0.0836,
+      "step": 302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8746894506634355e-05,
+      "loss": 0.0891,
+      "step": 303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8737221614452497e-05,
+      "loss": 0.0778,
+      "step": 304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.87275140478218e-05,
+      "loss": 0.1147,
+      "step": 305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8717771845267615e-05,
+      "loss": 0.104,
+      "step": 306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8707995045452744e-05,
+      "loss": 0.0832,
+      "step": 307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8698183687177295e-05,
+      "loss": 0.0939,
+      "step": 308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868833780937853e-05,
+      "loss": 0.0778,
+      "step": 309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8678457451130693e-05,
+      "loss": 0.0812,
+      "step": 310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866854265164488e-05,
+      "loss": 0.0859,
+      "step": 311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8658593450268853e-05,
+      "loss": 0.0765,
+      "step": 312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8648609886486923e-05,
+      "loss": 0.0912,
+      "step": 313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8638591999919755e-05,
+      "loss": 0.0885,
+      "step": 314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862853983032423e-05,
+      "loss": 0.0799,
+      "step": 315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8618453417593287e-05,
+      "loss": 0.0827,
+      "step": 316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8608332801755764e-05,
+      "loss": 0.1009,
+      "step": 317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859817802297623e-05,
+      "loss": 0.1001,
+      "step": 318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8587989121554837e-05,
+      "loss": 0.1012,
+      "step": 319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8577766137927163e-05,
+      "loss": 0.0791,
+      "step": 320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8567509112664024e-05,
+      "loss": 0.0901,
+      "step": 321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8557218086471356e-05,
+      "loss": 0.0767,
+      "step": 322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854689310019002e-05,
+      "loss": 0.0936,
+      "step": 323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8536534194795647e-05,
+      "loss": 0.0967,
+      "step": 324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8526141411398482e-05,
+      "loss": 0.0651,
+      "step": 325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851571479124323e-05,
+      "loss": 0.1057,
+      "step": 326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850525437570886e-05,
+      "loss": 0.1047,
+      "step": 327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8494760206308476e-05,
+      "loss": 0.0876,
+      "step": 328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8484232324689127e-05,
+      "loss": 0.0858,
+      "step": 329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847367077263166e-05,
+      "loss": 0.0903,
+      "step": 330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846307559205055e-05,
+      "loss": 0.1116,
+      "step": 331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8452446824993704e-05,
+      "loss": 0.0996,
+      "step": 332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.844178451364236e-05,
+      "loss": 0.0898,
+      "step": 333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8431088700310846e-05,
+      "loss": 0.0746,
+      "step": 334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842035942744646e-05,
+      "loss": 0.0701,
+      "step": 335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840959673762929e-05,
+      "loss": 0.1063,
+      "step": 336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8398800673572032e-05,
+      "loss": 0.0771,
+      "step": 337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8387971278119834e-05,
+      "loss": 0.0981,
+      "step": 338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837710859425013e-05,
+      "loss": 0.0967,
+      "step": 339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8366212665072455e-05,
+      "loss": 0.0949,
+      "step": 340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8355283533828285e-05,
+      "loss": 0.0977,
+      "step": 341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8344321243890856e-05,
+      "loss": 0.0936,
+      "step": 342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8333325838765e-05,
+      "loss": 0.0785,
+      "step": 343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8322297362086972e-05,
+      "loss": 0.0927,
+      "step": 344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831123585762427e-05,
+      "loss": 0.0868,
+      "step": 345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8300141369275472e-05,
+      "loss": 0.0973,
+      "step": 346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8289013941070046e-05,
+      "loss": 0.0775,
+      "step": 347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8277853617168197e-05,
+      "loss": 0.0939,
+      "step": 348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8266660441860666e-05,
+      "loss": 0.0839,
+      "step": 349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.825543445956858e-05,
+      "loss": 0.0847,
+      "step": 350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8244175714843256e-05,
+      "loss": 0.0966,
+      "step": 351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8232884252366036e-05,
+      "loss": 0.1016,
+      "step": 352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8221560116948103e-05,
+      "loss": 0.0744,
+      "step": 353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.82102033535303e-05,
+      "loss": 0.0894,
+      "step": 354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819881400718297e-05,
+      "loss": 0.0811,
+      "step": 355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8187392123105752e-05,
+      "loss": 0.0927,
+      "step": 356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.817593774662742e-05,
+      "loss": 0.0945,
+      "step": 357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8164450923205697e-05,
+      "loss": 0.0894,
+      "step": 358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8152931698427076e-05,
+      "loss": 0.0939,
+      "step": 359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8141380118006632e-05,
+      "loss": 0.0999,
+      "step": 360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.812979622778785e-05,
+      "loss": 0.089,
+      "step": 361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8118180073742442e-05,
+      "loss": 0.0957,
+      "step": 362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.810653170197015e-05,
+      "loss": 0.0742,
+      "step": 363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8094851158698597e-05,
+      "loss": 0.1087,
+      "step": 364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8083138490283057e-05,
+      "loss": 0.0948,
+      "step": 365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807139374320631e-05,
+      "loss": 0.0744,
+      "step": 366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8059616964078443e-05,
+      "loss": 0.0789,
+      "step": 367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.804780819963666e-05,
+      "loss": 0.0861,
+      "step": 368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80359674967451e-05,
+      "loss": 0.1033,
+      "step": 369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.802409490239466e-05,
+      "loss": 0.0928,
+      "step": 370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.80121904637028e-05,
+      "loss": 0.0718,
+      "step": 371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8000254227913346e-05,
+      "loss": 0.0928,
+      "step": 372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.798828624239633e-05,
+      "loss": 0.0926,
+      "step": 373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7976286554647773e-05,
+      "loss": 0.0819,
+      "step": 374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7964255212289513e-05,
+      "loss": 0.0741,
+      "step": 375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7952192263069018e-05,
+      "loss": 0.0729,
+      "step": 376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7940097754859177e-05,
+      "loss": 0.0887,
+      "step": 377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7927971735658143e-05,
+      "loss": 0.0862,
+      "step": 378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791581425358911e-05,
+      "loss": 0.0792,
+      "step": 379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790362535690013e-05,
+      "loss": 0.0837,
+      "step": 380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 0.0854,
+      "step": 381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787915351327775e-05,
+      "loss": 0.1124,
+      "step": 382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7866870663463057e-05,
+      "loss": 0.102,
+      "step": 383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785455659326546e-05,
+      "loss": 0.081,
+      "step": 384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.784221135155445e-05,
+      "loss": 0.0797,
+      "step": 385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.782983498732322e-05,
+      "loss": 0.0941,
+      "step": 386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7817427549688493e-05,
+      "loss": 0.0719,
+      "step": 387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.78049890878903e-05,
+      "loss": 0.0892,
+      "step": 388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7792519651291783e-05,
+      "loss": 0.0747,
+      "step": 389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7780019289379033e-05,
+      "loss": 0.0829,
+      "step": 390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7767488051760858e-05,
+      "loss": 0.0736,
+      "step": 391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7754925988168592e-05,
+      "loss": 0.0901,
+      "step": 392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7742333148455922e-05,
+      "loss": 0.0973,
+      "step": 393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7729709582598655e-05,
+      "loss": 0.082,
+      "step": 394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7717055340694555e-05,
+      "loss": 0.1004,
+      "step": 395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7704370472963107e-05,
+      "loss": 0.0738,
+      "step": 396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769165502974536e-05,
+      "loss": 0.0884,
+      "step": 397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7678909061503685e-05,
+      "loss": 0.0656,
+      "step": 398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7666132618821605e-05,
+      "loss": 0.0684,
+      "step": 399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7653325752403576e-05,
+      "loss": 0.0775,
+      "step": 400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7640488513074804e-05,
+      "loss": 0.0779,
+      "step": 401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7627620951781024e-05,
+      "loss": 0.0899,
+      "step": 402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7614723119588305e-05,
+      "loss": 0.063,
+      "step": 403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.760179506768286e-05,
+      "loss": 0.0573,
+      "step": 404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7588836847370817e-05,
+      "loss": 0.0805,
+      "step": 405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7575848510078047e-05,
+      "loss": 0.0758,
+      "step": 406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7562830107349923e-05,
+      "loss": 0.1009,
+      "step": 407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7549781690851148e-05,
+      "loss": 0.0785,
+      "step": 408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.753670331236554e-05,
+      "loss": 0.1014,
+      "step": 409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7523595023795814e-05,
+      "loss": 0.0843,
+      "step": 410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7510456877163394e-05,
+      "loss": 0.0962,
+      "step": 411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7497288924608192e-05,
+      "loss": 0.0856,
+      "step": 412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7484091218388414e-05,
+      "loss": 0.0955,
+      "step": 413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7470863810880335e-05,
+      "loss": 0.09,
+      "step": 414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7457606754578122e-05,
+      "loss": 0.1079,
+      "step": 415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7444320102093586e-05,
+      "loss": 0.0975,
+      "step": 416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7431003906156e-05,
+      "loss": 0.085,
+      "step": 417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7417658219611892e-05,
+      "loss": 0.082,
+      "step": 418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7404283095424803e-05,
+      "loss": 0.0768,
+      "step": 419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7390878586675127e-05,
+      "loss": 0.0977,
+      "step": 420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7377444746559865e-05,
+      "loss": 0.0636,
+      "step": 421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7363981628392405e-05,
+      "loss": 0.0862,
+      "step": 422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7350489285602345e-05,
+      "loss": 0.0921,
+      "step": 423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7336967771735266e-05,
+      "loss": 0.0857,
+      "step": 424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7323417140452507e-05,
+      "loss": 0.0764,
+      "step": 425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7309837445530958e-05,
+      "loss": 0.1004,
+      "step": 426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7296228740862872e-05,
+      "loss": 0.0869,
+      "step": 427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7282591080455602e-05,
+      "loss": 0.085,
+      "step": 428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7268924518431437e-05,
+      "loss": 0.0819,
+      "step": 429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7255229109027357e-05,
+      "loss": 0.0714,
+      "step": 430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.724150490659482e-05,
+      "loss": 0.0966,
+      "step": 431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7227751965599556e-05,
+      "loss": 0.0691,
+      "step": 432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7213970340621346e-05,
+      "loss": 0.0778,
+      "step": 433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7200160086353815e-05,
+      "loss": 0.0892,
+      "step": 434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7186321257604186e-05,
+      "loss": 0.0899,
+      "step": 435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7172453909293105e-05,
+      "loss": 0.0754,
+      "step": 436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.715855809645438e-05,
+      "loss": 0.1022,
+      "step": 437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71446338742348e-05,
+      "loss": 0.0779,
+      "step": 438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7130681297893884e-05,
+      "loss": 0.1035,
+      "step": 439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7116700422803694e-05,
+      "loss": 0.0765,
+      "step": 440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.710269130444858e-05,
+      "loss": 0.0885,
+      "step": 441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7088653998424995e-05,
+      "loss": 0.1034,
+      "step": 442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.707458856044124e-05,
+      "loss": 0.0789,
+      "step": 443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7060495046317273e-05,
+      "loss": 0.1259,
+      "step": 444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7046373511984477e-05,
+      "loss": 0.1019,
+      "step": 445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7032224013485417e-05,
+      "loss": 0.0759,
+      "step": 446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701804660697366e-05,
+      "loss": 0.0869,
+      "step": 447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.700384134871351e-05,
+      "loss": 0.0926,
+      "step": 448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6989608295079817e-05,
+      "loss": 0.1235,
+      "step": 449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6975347502557724e-05,
+      "loss": 0.0898,
+      "step": 450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6961059027742474e-05,
+      "loss": 0.1056,
+      "step": 451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6946742927339165e-05,
+      "loss": 0.108,
+      "step": 452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6932399258162518e-05,
+      "loss": 0.0877,
+      "step": 453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691802807713668e-05,
+      "loss": 0.0891,
+      "step": 454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6903629441294978e-05,
+      "loss": 0.081,
+      "step": 455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6889203407779678e-05,
+      "loss": 0.0975,
+      "step": 456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6874750033841802e-05,
+      "loss": 0.0823,
+      "step": 457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6860269376840856e-05,
+      "loss": 0.0913,
+      "step": 458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6845761494244633e-05,
+      "loss": 0.0668,
+      "step": 459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6831226443628966e-05,
+      "loss": 0.1065,
+      "step": 460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.681666428267751e-05,
+      "loss": 0.098,
+      "step": 461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.680207506918151e-05,
+      "loss": 0.0768,
+      "step": 462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6787458861039562e-05,
+      "loss": 0.0893,
+      "step": 463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 0.0721,
+      "step": 464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.675814569294769e-05,
+      "loss": 0.0936,
+      "step": 465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6743448849329702e-05,
+      "loss": 0.0724,
+      "step": 466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.672872524372919e-05,
+      "loss": 0.075,
+      "step": 467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.671397493457811e-05,
+      "loss": 0.0751,
+      "step": 468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6699197980414385e-05,
+      "loss": 0.0825,
+      "step": 469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6684394439881688e-05,
+      "loss": 0.0908,
+      "step": 470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6669564371729196e-05,
+      "loss": 0.0822,
+      "step": 471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665470783481137e-05,
+      "loss": 0.0824,
+      "step": 472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663982488808771e-05,
+      "loss": 0.0953,
+      "step": 473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6624915590622527e-05,
+      "loss": 0.0791,
+      "step": 474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6609980001584708e-05,
+      "loss": 0.0733,
+      "step": 475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6595018180247478e-05,
+      "loss": 0.1086,
+      "step": 476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6580030185988167e-05,
+      "loss": 0.0836,
+      "step": 477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6565016078287984e-05,
+      "loss": 0.0847,
+      "step": 478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654997591673176e-05,
+      "loss": 0.0902,
+      "step": 479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6534909761007723e-05,
+      "loss": 0.0702,
+      "step": 480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.651981767090727e-05,
+      "loss": 0.0858,
+      "step": 481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6504699706324717e-05,
+      "loss": 0.0799,
+      "step": 482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6489555927257063e-05,
+      "loss": 0.0693,
+      "step": 483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647438639380375e-05,
+      "loss": 0.1123,
+      "step": 484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.645919116616645e-05,
+      "loss": 0.0936,
+      "step": 485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644397030464877e-05,
+      "loss": 0.1063,
+      "step": 486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642872386965608e-05,
+      "loss": 0.0695,
+      "step": 487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.641345192169522e-05,
+      "loss": 0.0672,
+      "step": 488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.639815452137429e-05,
+      "loss": 0.0824,
+      "step": 489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6382831729402395e-05,
+      "loss": 0.0959,
+      "step": 490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6367483606589413e-05,
+      "loss": 0.0889,
+      "step": 491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6352110213845746e-05,
+      "loss": 0.0782,
+      "step": 492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.633671161218209e-05,
+      "loss": 0.0632,
+      "step": 493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6321287862709173e-05,
+      "loss": 0.0774,
+      "step": 494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6305839026637532e-05,
+      "loss": 0.1093,
+      "step": 495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6290365165277263e-05,
+      "loss": 0.0845,
+      "step": 496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627486634003777e-05,
+      "loss": 0.0668,
+      "step": 497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625934261242754e-05,
+      "loss": 0.0862,
+      "step": 498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624379404405387e-05,
+      "loss": 0.0817,
+      "step": 499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.622822069662266e-05,
+      "loss": 0.0754,
+      "step": 500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6212622631938137e-05,
+      "loss": 0.0861,
+      "step": 501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.619699991190262e-05,
+      "loss": 0.0885,
+      "step": 502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6181352598516275e-05,
+      "loss": 0.0726,
+      "step": 503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6165680753876872e-05,
+      "loss": 0.0966,
+      "step": 504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.614998444017954e-05,
+      "loss": 0.0753,
+      "step": 505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61342637197165e-05,
+      "loss": 0.0747,
+      "step": 506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.611851865487685e-05,
+      "loss": 0.09,
+      "step": 507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6102749308146285e-05,
+      "loss": 0.0911,
+      "step": 508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608695574210689e-05,
+      "loss": 0.084,
+      "step": 509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607113801943684e-05,
+      "loss": 0.0721,
+      "step": 510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.605529620291019e-05,
+      "loss": 0.0927,
+      "step": 511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.603943035539661e-05,
+      "loss": 0.0732,
+      "step": 512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6023540539861144e-05,
+      "loss": 0.0688,
+      "step": 513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6007626819363955e-05,
+      "loss": 0.0683,
+      "step": 514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5991689257060067e-05,
+      "loss": 0.0739,
+      "step": 515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5975727916199128e-05,
+      "loss": 0.0953,
+      "step": 516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5959742860125153e-05,
+      "loss": 0.0833,
+      "step": 517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5943734152276277e-05,
+      "loss": 0.0616,
+      "step": 518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5927701856184494e-05,
+      "loss": 0.0848,
+      "step": 519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591164603547541e-05,
+      "loss": 0.0801,
+      "step": 520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.589556675386799e-05,
+      "loss": 0.0747,
+      "step": 521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587946407517431e-05,
+      "loss": 0.0627,
+      "step": 522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5863338063299296e-05,
+      "loss": 0.0657,
+      "step": 523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5847188782240473e-05,
+      "loss": 0.0676,
+      "step": 524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5831016296087714e-05,
+      "loss": 0.0775,
+      "step": 525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5814820669022988e-05,
+      "loss": 0.0668,
+      "step": 526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5798601965320096e-05,
+      "loss": 0.0826,
+      "step": 527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.578236024934441e-05,
+      "loss": 0.0708,
+      "step": 528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5766095585552648e-05,
+      "loss": 0.0992,
+      "step": 529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5749808038492587e-05,
+      "loss": 0.0591,
+      "step": 530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.573349767280282e-05,
+      "loss": 0.067,
+      "step": 531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.57171645532125e-05,
+      "loss": 0.074,
+      "step": 532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.570080874454108e-05,
+      "loss": 0.0922,
+      "step": 533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.568443031169805e-05,
+      "loss": 0.0903,
+      "step": 534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5668029319682698e-05,
+      "loss": 0.0968,
+      "step": 535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5651605833583834e-05,
+      "loss": 0.0839,
+      "step": 536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5635159918579537e-05,
+      "loss": 0.0827,
+      "step": 537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5618691639936897e-05,
+      "loss": 0.0935,
+      "step": 538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5602201063011754e-05,
+      "loss": 0.0938,
+      "step": 539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.558568825324845e-05,
+      "loss": 0.0834,
+      "step": 540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5569153276179546e-05,
+      "loss": 0.0829,
+      "step": 541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5552596197425596e-05,
+      "loss": 0.0757,
+      "step": 542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5536017082694846e-05,
+      "loss": 0.0796,
+      "step": 543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5519415997783003e-05,
+      "loss": 0.0959,
+      "step": 544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5502793008572964e-05,
+      "loss": 0.0686,
+      "step": 545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5486148181034555e-05,
+      "loss": 0.0927,
+      "step": 546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.0837,
+      "step": 547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5452793275285006e-05,
+      "loss": 0.0677,
+      "step": 548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5436083329445807e-05,
+      "loss": 0.0806,
+      "step": 549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.541935181002159e-05,
+      "loss": 0.0696,
+      "step": 550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5402598783412897e-05,
+      "loss": 0.0777,
+      "step": 551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5385824316105615e-05,
+      "loss": 0.0859,
+      "step": 552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.536902847467073e-05,
+      "loss": 0.0651,
+      "step": 553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5352211325764045e-05,
+      "loss": 0.0799,
+      "step": 554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.533537293612592e-05,
+      "loss": 0.0856,
+      "step": 555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5318513372581026e-05,
+      "loss": 0.0809,
+      "step": 556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5301632702038047e-05,
+      "loss": 0.092,
+      "step": 557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5284730991489448e-05,
+      "loss": 0.0685,
+      "step": 558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5267808308011183e-05,
+      "loss": 0.0849,
+      "step": 559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.525086471876244e-05,
+      "loss": 0.0819,
+      "step": 560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5233900290985375e-05,
+      "loss": 0.0855,
+      "step": 561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5216915092004847e-05,
+      "loss": 0.0922,
+      "step": 562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5199909189228139e-05,
+      "loss": 0.0982,
+      "step": 563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.518288265014471e-05,
+      "loss": 0.0703,
+      "step": 564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5165835542325901e-05,
+      "loss": 0.0814,
+      "step": 565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5148767933424697e-05,
+      "loss": 0.075,
+      "step": 566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.513167989117544e-05,
+      "loss": 0.06,
+      "step": 567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5114571483393554e-05,
+      "loss": 0.0829,
+      "step": 568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5097442777975295e-05,
+      "loss": 0.0938,
+      "step": 569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.508029384289747e-05,
+      "loss": 0.0826,
+      "step": 570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5063124746217165e-05,
+      "loss": 0.0596,
+      "step": 571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5045935556071487e-05,
+      "loss": 0.0837,
+      "step": 572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5028726340677278e-05,
+      "loss": 0.069,
+      "step": 573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5011497168330853e-05,
+      "loss": 0.076,
+      "step": 574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4994248107407736e-05,
+      "loss": 0.1051,
+      "step": 575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4976979226362372e-05,
+      "loss": 0.0864,
+      "step": 576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4959690593727867e-05,
+      "loss": 0.0729,
+      "step": 577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4942382278115713e-05,
+      "loss": 0.0778,
+      "step": 578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4925054348215514e-05,
+      "loss": 0.0801,
+      "step": 579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4907706872794717e-05,
+      "loss": 0.0569,
+      "step": 580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4890339920698334e-05,
+      "loss": 0.0748,
+      "step": 581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4872953560848678e-05,
+      "loss": 0.0768,
+      "step": 582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4855547862245082e-05,
+      "loss": 0.0943,
+      "step": 583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4838122893963619e-05,
+      "loss": 0.0739,
+      "step": 584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4820678725156844e-05,
+      "loss": 0.0809,
+      "step": 585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4803215425053505e-05,
+      "loss": 0.086,
+      "step": 586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4785733062958284e-05,
+      "loss": 0.0972,
+      "step": 587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4768231708251497e-05,
+      "loss": 0.0846,
+      "step": 588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4750711430388846e-05,
+      "loss": 0.0782,
+      "step": 589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.473317229890113e-05,
+      "loss": 0.0685,
+      "step": 590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4715614383393963e-05,
+      "loss": 0.0742,
+      "step": 591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4698037753547513e-05,
+      "loss": 0.077,
+      "step": 592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4680442479116215e-05,
+      "loss": 0.0718,
+      "step": 593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4662828629928496e-05,
+      "loss": 0.0983,
+      "step": 594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4645196275886497e-05,
+      "loss": 0.0732,
+      "step": 595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4627545486965799e-05,
+      "loss": 0.0865,
+      "step": 596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4609876333215143e-05,
+      "loss": 0.0994,
+      "step": 597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4592188884756155e-05,
+      "loss": 0.0811,
+      "step": 598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4574483211783061e-05,
+      "loss": 0.0928,
+      "step": 599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4556759384562418e-05,
+      "loss": 0.1023,
+      "step": 600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4539017473432822e-05,
+      "loss": 0.0704,
+      "step": 601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4521257548804644e-05,
+      "loss": 0.0734,
+      "step": 602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.450347968115974e-05,
+      "loss": 0.0878,
+      "step": 603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4485683941051174e-05,
+      "loss": 0.0966,
+      "step": 604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.446787039910294e-05,
+      "loss": 0.0792,
+      "step": 605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4450039126009679e-05,
+      "loss": 0.086,
+      "step": 606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4432190192536398e-05,
+      "loss": 0.0756,
+      "step": 607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4414323669518194e-05,
+      "loss": 0.072,
+      "step": 608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.439643962785997e-05,
+      "loss": 0.0839,
+      "step": 609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4378538138536154e-05,
+      "loss": 0.0833,
+      "step": 610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4360619272590413e-05,
+      "loss": 0.077,
+      "step": 611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.434268310113537e-05,
+      "loss": 0.082,
+      "step": 612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4324729695352338e-05,
+      "loss": 0.0713,
+      "step": 613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4306759126491021e-05,
+      "loss": 0.0855,
+      "step": 614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4288771465869236e-05,
+      "loss": 0.0626,
+      "step": 615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4270766784872627e-05,
+      "loss": 0.0873,
+      "step": 616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4252745154954393e-05,
+      "loss": 0.0771,
+      "step": 617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4234706647634982e-05,
+      "loss": 0.0602,
+      "step": 618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.421665133450184e-05,
+      "loss": 0.0777,
+      "step": 619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4198579287209098e-05,
+      "loss": 0.0768,
+      "step": 620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4180490577477295e-05,
+      "loss": 0.0785,
+      "step": 621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4162385277093102e-05,
+      "loss": 0.077,
+      "step": 622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4144263457909028e-05,
+      "loss": 0.0874,
+      "step": 623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4126125191843146e-05,
+      "loss": 0.0868,
+      "step": 624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.410797055087879e-05,
+      "loss": 0.1035,
+      "step": 625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4089799607064282e-05,
+      "loss": 0.091,
+      "step": 626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4071612432512652e-05,
+      "loss": 0.0817,
+      "step": 627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4053409099401325e-05,
+      "loss": 0.0782,
+      "step": 628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4035189679971874e-05,
+      "loss": 0.0683,
+      "step": 629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 0.0681,
+      "step": 630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.399870287144375e-05,
+      "loss": 0.0854,
+      "step": 631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3980435627146252e-05,
+      "loss": 0.0875,
+      "step": 632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3962152586132409e-05,
+      "loss": 0.0905,
+      "step": 633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3943853820960104e-05,
+      "loss": 0.0985,
+      "step": 634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3925539404249638e-05,
+      "loss": 0.085,
+      "step": 635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3907209408683415e-05,
+      "loss": 0.0953,
+      "step": 636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3888863907005669e-05,
+      "loss": 0.0796,
+      "step": 637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3870502972022175e-05,
+      "loss": 0.075,
+      "step": 638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3852126676599944e-05,
+      "loss": 0.0712,
+      "step": 639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3833735093666962e-05,
+      "loss": 0.0847,
+      "step": 640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3815328296211878e-05,
+      "loss": 0.0783,
+      "step": 641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3796906357283723e-05,
+      "loss": 0.0901,
+      "step": 642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3778469349991612e-05,
+      "loss": 0.097,
+      "step": 643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3760017347504462e-05,
+      "loss": 0.0814,
+      "step": 644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3741550423050711e-05,
+      "loss": 0.072,
+      "step": 645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3723068649918e-05,
+      "loss": 0.0859,
+      "step": 646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3704572101452911e-05,
+      "loss": 0.0704,
+      "step": 647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3686060851060657e-05,
+      "loss": 0.0822,
+      "step": 648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3667534972204795e-05,
+      "loss": 0.086,
+      "step": 649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.364899453840694e-05,
+      "loss": 0.0628,
+      "step": 650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3630439623246474e-05,
+      "loss": 0.0668,
+      "step": 651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.361187030036024e-05,
+      "loss": 0.0688,
+      "step": 652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3593286643442265e-05,
+      "loss": 0.0773,
+      "step": 653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3574688726243462e-05,
+      "loss": 0.0883,
+      "step": 654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3556076622571332e-05,
+      "loss": 0.0902,
+      "step": 655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3537450406289686e-05,
+      "loss": 0.0685,
+      "step": 656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.351881015131833e-05,
+      "loss": 0.0714,
+      "step": 657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3500155931632799e-05,
+      "loss": 0.1048,
+      "step": 658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3481487821264034e-05,
+      "loss": 0.0834,
+      "step": 659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3462805894298105e-05,
+      "loss": 0.0802,
+      "step": 660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3444110224875926e-05,
+      "loss": 0.1218,
+      "step": 661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3425400887192934e-05,
+      "loss": 0.0936,
+      "step": 662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3406677955498817e-05,
+      "loss": 0.0942,
+      "step": 663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3387941504097213e-05,
+      "loss": 0.0864,
+      "step": 664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.336919160734541e-05,
+      "loss": 0.0802,
+      "step": 665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3350428339654057e-05,
+      "loss": 0.0757,
+      "step": 666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3331651775486874e-05,
+      "loss": 0.0853,
+      "step": 667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3312861989360336e-05,
+      "loss": 0.072,
+      "step": 668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.32940590558434e-05,
+      "loss": 0.0948,
+      "step": 669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3275243049557195e-05,
+      "loss": 0.0751,
+      "step": 670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3256414045174734e-05,
+      "loss": 0.0796,
+      "step": 671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3237572117420612e-05,
+      "loss": 0.0924,
+      "step": 672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3218717341070709e-05,
+      "loss": 0.0884,
+      "step": 673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3199849790951901e-05,
+      "loss": 0.0778,
+      "step": 674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.318096954194176e-05,
+      "loss": 0.1088,
+      "step": 675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.316207666896824e-05,
+      "loss": 0.0738,
+      "step": 676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3143171247009414e-05,
+      "loss": 0.0818,
+      "step": 677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3124253351093141e-05,
+      "loss": 0.0872,
+      "step": 678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3105323056296797e-05,
+      "loss": 0.0835,
+      "step": 679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3086380437746948e-05,
+      "loss": 0.0842,
+      "step": 680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3067425570619083e-05,
+      "loss": 0.0822,
+      "step": 681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3048458530137299e-05,
+      "loss": 0.1031,
+      "step": 682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3029479391573991e-05,
+      "loss": 0.0943,
+      "step": 683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3010488230249583e-05,
+      "loss": 0.0728,
+      "step": 684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.29914851215322e-05,
+      "loss": 0.0755,
+      "step": 685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2972470140837386e-05,
+      "loss": 0.0753,
+      "step": 686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2953443363627805e-05,
+      "loss": 0.0835,
+      "step": 687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2934404865412924e-05,
+      "loss": 0.067,
+      "step": 688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2915354721748739e-05,
+      "loss": 0.0862,
+      "step": 689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2896293008237454e-05,
+      "loss": 0.0893,
+      "step": 690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2877219800527194e-05,
+      "loss": 0.075,
+      "step": 691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2858135174311692e-05,
+      "loss": 0.0863,
+      "step": 692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.283903920533001e-05,
+      "loss": 0.0872,
+      "step": 693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2819931969366208e-05,
+      "loss": 0.0775,
+      "step": 694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2800813542249073e-05,
+      "loss": 0.079,
+      "step": 695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2781683999851794e-05,
+      "loss": 0.0773,
+      "step": 696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2762543418091688e-05,
+      "loss": 0.0782,
+      "step": 697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2743391872929864e-05,
+      "loss": 0.1129,
+      "step": 698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2724229440370959e-05,
+      "loss": 0.0717,
+      "step": 699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2705056196462801e-05,
+      "loss": 0.0704,
+      "step": 700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2685872217296131e-05,
+      "loss": 0.0724,
+      "step": 701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2666677579004296e-05,
+      "loss": 0.0721,
+      "step": 702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.264747235776294e-05,
+      "loss": 0.1017,
+      "step": 703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2628256629789712e-05,
+      "loss": 0.0685,
+      "step": 704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2609030471343953e-05,
+      "loss": 0.0841,
+      "step": 705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2589793958726398e-05,
+      "loss": 0.0722,
+      "step": 706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2570547168278875e-05,
+      "loss": 0.0823,
+      "step": 707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2551290176384005e-05,
+      "loss": 0.0836,
+      "step": 708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.253202305946489e-05,
+      "loss": 0.0731,
+      "step": 709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.251274589398481e-05,
+      "loss": 0.0535,
+      "step": 710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.249345875644693e-05,
+      "loss": 0.0787,
+      "step": 711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2474161723393987e-05,
+      "loss": 0.0844,
+      "step": 712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.077,
+      "step": 713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2435538277109919e-05,
+      "loss": 0.0858,
+      "step": 714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2416212017159413e-05,
+      "loss": 0.0847,
+      "step": 715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2396876168254468e-05,
+      "loss": 0.0807,
+      "step": 716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2377530807131138e-05,
+      "loss": 0.0827,
+      "step": 717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2358176010563223e-05,
+      "loss": 0.071,
+      "step": 718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2338811855361978e-05,
+      "loss": 0.0819,
+      "step": 719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2319438418375792e-05,
+      "loss": 0.0788,
+      "step": 720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2300055776489883e-05,
+      "loss": 0.0839,
+      "step": 721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2280664006626012e-05,
+      "loss": 0.0717,
+      "step": 722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.226126318574216e-05,
+      "loss": 0.0718,
+      "step": 723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2241853390832229e-05,
+      "loss": 0.085,
+      "step": 724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2222434698925728e-05,
+      "loss": 0.0916,
+      "step": 725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2203007187087485e-05,
+      "loss": 0.0892,
+      "step": 726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2183570932417324e-05,
+      "loss": 0.0726,
+      "step": 727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2164126012049766e-05,
+      "loss": 0.0924,
+      "step": 728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2144672503153728e-05,
+      "loss": 0.0666,
+      "step": 729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2125210482932204e-05,
+      "loss": 0.0716,
+      "step": 730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2105740028621971e-05,
+      "loss": 0.0593,
+      "step": 731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2086261217493276e-05,
+      "loss": 0.0822,
+      "step": 732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.206677412684953e-05,
+      "loss": 0.1001,
+      "step": 733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2047278834027006e-05,
+      "loss": 0.0839,
+      "step": 734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2027775416394523e-05,
+      "loss": 0.0941,
+      "step": 735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2008263951353143e-05,
+      "loss": 0.0696,
+      "step": 736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.198874451633587e-05,
+      "loss": 0.0889,
+      "step": 737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1969217188807334e-05,
+      "loss": 0.0728,
+      "step": 738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.194968204626349e-05,
+      "loss": 0.0891,
+      "step": 739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.193013916623131e-05,
+      "loss": 0.0585,
+      "step": 740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.191058862626846e-05,
+      "loss": 0.0811,
+      "step": 741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1891030503963022e-05,
+      "loss": 0.0806,
+      "step": 742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1871464876933156e-05,
+      "loss": 0.0866,
+      "step": 743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.185189182282681e-05,
+      "loss": 0.0919,
+      "step": 744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1832311419321414e-05,
+      "loss": 0.0688,
+      "step": 745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1812723744123553e-05,
+      "loss": 0.0819,
+      "step": 746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1793128874968675e-05,
+      "loss": 0.0857,
+      "step": 747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1773526889620782e-05,
+      "loss": 0.0874,
+      "step": 748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1753917865872111e-05,
+      "loss": 0.0915,
+      "step": 749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1734301881542836e-05,
+      "loss": 0.0774,
+      "step": 750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1714679014480752e-05,
+      "loss": 0.0598,
+      "step": 751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1695049342560969e-05,
+      "loss": 0.0685,
+      "step": 752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1675412943685605e-05,
+      "loss": 0.0818,
+      "step": 753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1655769895783469e-05,
+      "loss": 0.0755,
+      "step": 754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1636120276809764e-05,
+      "loss": 0.0809,
+      "step": 755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1616464164745768e-05,
+      "loss": 0.0835,
+      "step": 756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1596801637598532e-05,
+      "loss": 0.0994,
+      "step": 757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1577132773400552e-05,
+      "loss": 0.0767,
+      "step": 758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.155745765020949e-05,
+      "loss": 0.0808,
+      "step": 759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1537776346107834e-05,
+      "loss": 0.0757,
+      "step": 760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1518088939202613e-05,
+      "loss": 0.0901,
+      "step": 761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1498395507625066e-05,
+      "loss": 0.0935,
+      "step": 762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1478696129530347e-05,
+      "loss": 0.0842,
+      "step": 763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1458990883097205e-05,
+      "loss": 0.0769,
+      "step": 764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1439279846527682e-05,
+      "loss": 0.0823,
+      "step": 765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.14195630980468e-05,
+      "loss": 0.0793,
+      "step": 766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1399840715902243e-05,
+      "loss": 0.0738,
+      "step": 767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1380112778364058e-05,
+      "loss": 0.0834,
+      "step": 768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1360379363724338e-05,
+      "loss": 0.0816,
+      "step": 769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1340640550296906e-05,
+      "loss": 0.0762,
+      "step": 770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1320896416417026e-05,
+      "loss": 0.1129,
+      "step": 771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1301147040441056e-05,
+      "loss": 0.0711,
+      "step": 772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1281392500746178e-05,
+      "loss": 0.0804,
+      "step": 773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1261632875730052e-05,
+      "loss": 0.0819,
+      "step": 774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1241868243810525e-05,
+      "loss": 0.0673,
+      "step": 775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.122209868342532e-05,
+      "loss": 0.0754,
+      "step": 776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1202324273031706e-05,
+      "loss": 0.0901,
+      "step": 777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.118254509110621e-05,
+      "loss": 0.1003,
+      "step": 778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1162761216144293e-05,
+      "loss": 0.0957,
+      "step": 779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1142972726660038e-05,
+      "loss": 0.0937,
+      "step": 780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.112317970118584e-05,
+      "loss": 0.0779,
+      "step": 781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1103382218272107e-05,
+      "loss": 0.0804,
+      "step": 782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1083580356486925e-05,
+      "loss": 0.0921,
+      "step": 783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1063774194415762e-05,
+      "loss": 0.0757,
+      "step": 784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1043963810661144e-05,
+      "loss": 0.0827,
+      "step": 785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.102414928384237e-05,
+      "loss": 0.0916,
+      "step": 786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1004330692595159e-05,
+      "loss": 0.0871,
+      "step": 787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0984508115571377e-05,
+      "loss": 0.0871,
+      "step": 788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0964681631438703e-05,
+      "loss": 0.0848,
+      "step": 789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0944851318880314e-05,
+      "loss": 0.0828,
+      "step": 790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.092501725659459e-05,
+      "loss": 0.0935,
+      "step": 791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0905179523294791e-05,
+      "loss": 0.0785,
+      "step": 792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0885338197708741e-05,
+      "loss": 0.0846,
+      "step": 793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0865493358578525e-05,
+      "loss": 0.0601,
+      "step": 794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0845645084660169e-05,
+      "loss": 0.062,
+      "step": 795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.0764,
+      "step": 796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0805938547550976e-05,
+      "loss": 0.0867,
+      "step": 797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0786080441939107e-05,
+      "loss": 0.0812,
+      "step": 798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.076621921669639e-05,
+      "loss": 0.081,
+      "step": 799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0746354950643883e-05,
+      "loss": 0.0605,
+      "step": 800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0726487722614704e-05,
+      "loss": 0.0871,
+      "step": 801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0706617611453745e-05,
+      "loss": 0.0757,
+      "step": 802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0686744696017314e-05,
+      "loss": 0.0721,
+      "step": 803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.066686905517287e-05,
+      "loss": 0.0715,
+      "step": 804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0646990767798672e-05,
+      "loss": 0.0887,
+      "step": 805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0627109912783497e-05,
+      "loss": 0.0834,
+      "step": 806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0607226569026297e-05,
+      "loss": 0.0783,
+      "step": 807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0587340815435913e-05,
+      "loss": 0.0661,
+      "step": 808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0567452730930743e-05,
+      "loss": 0.0891,
+      "step": 809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0547562394438434e-05,
+      "loss": 0.0586,
+      "step": 810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0527669884895573e-05,
+      "loss": 0.083,
+      "step": 811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0507775281247376e-05,
+      "loss": 0.0794,
+      "step": 812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0487878662447361e-05,
+      "loss": 0.0768,
+      "step": 813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.046798010745705e-05,
+      "loss": 0.0605,
+      "step": 814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0448079695245643e-05,
+      "loss": 0.0593,
+      "step": 815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0428177504789713e-05,
+      "loss": 0.0526,
+      "step": 816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0408273615072893e-05,
+      "loss": 0.063,
+      "step": 817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0388368105085558e-05,
+      "loss": 0.0728,
+      "step": 818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.036846105382451e-05,
+      "loss": 0.0747,
+      "step": 819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.034855254029267e-05,
+      "loss": 0.0661,
+      "step": 820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0328642643498763e-05,
+      "loss": 0.0829,
+      "step": 821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0308731442457004e-05,
+      "loss": 0.0784,
+      "step": 822
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0288819016186782e-05,
+      "loss": 0.0592,
+      "step": 823
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0268905443712351e-05,
+      "loss": 0.0831,
+      "step": 824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.024899080406251e-05,
+      "loss": 0.0753,
+      "step": 825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0229075176270297e-05,
+      "loss": 0.0849,
+      "step": 826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.020915863937267e-05,
+      "loss": 0.0646,
+      "step": 827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0189241272410191e-05,
+      "loss": 0.0718,
+      "step": 828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0169323154426726e-05,
+      "loss": 0.0677,
+      "step": 829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0149404364469108e-05,
+      "loss": 0.0726,
+      "step": 830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0129484981586853e-05,
+      "loss": 0.069,
+      "step": 831
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0109565084831817e-05,
+      "loss": 0.067,
+      "step": 832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0089644753257897e-05,
+      "loss": 0.0669,
+      "step": 833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.006972406592072e-05,
+      "loss": 0.077,
+      "step": 834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.004980310187733e-05,
+      "loss": 0.0579,
+      "step": 835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0029881940185851e-05,
+      "loss": 0.051,
+      "step": 836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0009960659905212e-05,
+      "loss": 0.0714,
+      "step": 837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.990039340094793e-06,
+      "loss": 0.0589,
+      "step": 838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.97011805981415e-06,
+      "loss": 0.0613,
+      "step": 839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.950196898122677e-06,
+      "loss": 0.0489,
+      "step": 840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.930275934079281e-06,
+      "loss": 0.0657,
+      "step": 841
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.910355246742105e-06,
+      "loss": 0.0698,
+      "step": 842
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.890434915168187e-06,
+      "loss": 0.0691,
+      "step": 843
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.870515018413147e-06,
+      "loss": 0.0733,
+      "step": 844
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.850595635530894e-06,
+      "loss": 0.0618,
+      "step": 845
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.830676845573277e-06,
+      "loss": 0.0928,
+      "step": 846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.810758727589814e-06,
+      "loss": 0.0799,
+      "step": 847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.790841360627335e-06,
+      "loss": 0.0538,
+      "step": 848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.770924823729708e-06,
+      "loss": 0.0669,
+      "step": 849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.751009195937493e-06,
+      "loss": 0.1058,
+      "step": 850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.731094556287649e-06,
+      "loss": 0.0659,
+      "step": 851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.711180983813222e-06,
+      "loss": 0.0677,
+      "step": 852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.691268557542997e-06,
+      "loss": 0.093,
+      "step": 853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.67135735650124e-06,
+      "loss": 0.071,
+      "step": 854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.651447459707333e-06,
+      "loss": 0.0553,
+      "step": 855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.631538946175496e-06,
+      "loss": 0.0541,
+      "step": 856
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.611631894914445e-06,
+      "loss": 0.0654,
+      "step": 857
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.59172638492711e-06,
+      "loss": 0.0668,
+      "step": 858
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.571822495210288e-06,
+      "loss": 0.061,
+      "step": 859
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.551920304754359e-06,
+      "loss": 0.0612,
+      "step": 860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.532019892542953e-06,
+      "loss": 0.0678,
+      "step": 861
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.51212133755264e-06,
+      "loss": 0.0608,
+      "step": 862
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.492224718752629e-06,
+      "loss": 0.0617,
+      "step": 863
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.472330115104428e-06,
+      "loss": 0.0646,
+      "step": 864
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.452437605561571e-06,
+      "loss": 0.0683,
+      "step": 865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.43254726906926e-06,
+      "loss": 0.0643,
+      "step": 866
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.412659184564088e-06,
+      "loss": 0.0768,
+      "step": 867
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.392773430973705e-06,
+      "loss": 0.0636,
+      "step": 868
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.372890087216505e-06,
+      "loss": 0.0675,
+      "step": 869
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.35300923220133e-06,
+      "loss": 0.0754,
+      "step": 870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.333130944827132e-06,
+      "loss": 0.0565,
+      "step": 871
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.31325530398269e-06,
+      "loss": 0.0712,
+      "step": 872
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.29338238854626e-06,
+      "loss": 0.0648,
+      "step": 873
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.273512277385297e-06,
+      "loss": 0.0816,
+      "step": 874
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.25364504935612e-06,
+      "loss": 0.0739,
+      "step": 875
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.233780783303611e-06,
+      "loss": 0.0663,
+      "step": 876
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.213919558060897e-06,
+      "loss": 0.0724,
+      "step": 877
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.194061452449024e-06,
+      "loss": 0.0611,
+      "step": 878
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.0451,
+      "step": 879
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.154354915339836e-06,
+      "loss": 0.0607,
+      "step": 880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.13450664142148e-06,
+      "loss": 0.0635,
+      "step": 881
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.114661802291262e-06,
+      "loss": 0.0797,
+      "step": 882
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.094820476705209e-06,
+      "loss": 0.0552,
+      "step": 883
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.074982743405413e-06,
+      "loss": 0.0657,
+      "step": 884
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.055148681119688e-06,
+      "loss": 0.0617,
+      "step": 885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.035318368561302e-06,
+      "loss": 0.0636,
+      "step": 886
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.015491884428623e-06,
+      "loss": 0.0532,
+      "step": 887
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.995669307404844e-06,
+      "loss": 0.056,
+      "step": 888
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.975850716157634e-06,
+      "loss": 0.064,
+      "step": 889
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.956036189338857e-06,
+      "loss": 0.0856,
+      "step": 890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.936225805584241e-06,
+      "loss": 0.064,
+      "step": 891
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.916419643513075e-06,
+      "loss": 0.0797,
+      "step": 892
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.896617781727895e-06,
+      "loss": 0.068,
+      "step": 893
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.87682029881416e-06,
+      "loss": 0.0681,
+      "step": 894
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.857027273339967e-06,
+      "loss": 0.0736,
+      "step": 895
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.837238783855709e-06,
+      "loss": 0.0697,
+      "step": 896
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.817454908893794e-06,
+      "loss": 0.0675,
+      "step": 897
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.797675726968297e-06,
+      "loss": 0.0531,
+      "step": 898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.777901316574687e-06,
+      "loss": 0.0576,
+      "step": 899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.758131756189477e-06,
+      "loss": 0.0576,
+      "step": 900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.73836712426995e-06,
+      "loss": 0.0632,
+      "step": 901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.718607499253825e-06,
+      "loss": 0.0739,
+      "step": 902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.698852959558944e-06,
+      "loss": 0.0591,
+      "step": 903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.67910358358298e-06,
+      "loss": 0.0519,
+      "step": 904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.659359449703095e-06,
+      "loss": 0.0628,
+      "step": 905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.639620636275667e-06,
+      "loss": 0.0693,
+      "step": 906
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.619887221635943e-06,
+      "loss": 0.062,
+      "step": 907
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.600159284097756e-06,
+      "loss": 0.0705,
+      "step": 908
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.580436901953203e-06,
+      "loss": 0.0814,
+      "step": 909
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.56072015347232e-06,
+      "loss": 0.069,
+      "step": 910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.541009116902798e-06,
+      "loss": 0.0564,
+      "step": 911
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.521303870469655e-06,
+      "loss": 0.0672,
+      "step": 912
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.501604492374939e-06,
+      "loss": 0.0739,
+      "step": 913
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.48191106079739e-06,
+      "loss": 0.1009,
+      "step": 914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.462223653892169e-06,
+      "loss": 0.0469,
+      "step": 915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.442542349790515e-06,
+      "loss": 0.071,
+      "step": 916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.42286722659945e-06,
+      "loss": 0.0629,
+      "step": 917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.403198362401472e-06,
+      "loss": 0.0612,
+      "step": 918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.38353583525423e-06,
+      "loss": 0.0573,
+      "step": 919
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.363879723190241e-06,
+      "loss": 0.051,
+      "step": 920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.344230104216536e-06,
+      "loss": 0.0487,
+      "step": 921
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.324587056314402e-06,
+      "loss": 0.0579,
+      "step": 922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.304950657439034e-06,
+      "loss": 0.0534,
+      "step": 923
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.285320985519254e-06,
+      "loss": 0.0626,
+      "step": 924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.265698118457167e-06,
+      "loss": 0.0774,
+      "step": 925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.24608213412789e-06,
+      "loss": 0.0598,
+      "step": 926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.226473110379221e-06,
+      "loss": 0.0838,
+      "step": 927
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.206871125031324e-06,
+      "loss": 0.0661,
+      "step": 928
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.18727625587645e-06,
+      "loss": 0.0715,
+      "step": 929
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.167688580678587e-06,
+      "loss": 0.0689,
+      "step": 930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.148108177173191e-06,
+      "loss": 0.0521,
+      "step": 931
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.128535123066846e-06,
+      "loss": 0.0575,
+      "step": 932
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.10896949603698e-06,
+      "loss": 0.0607,
+      "step": 933
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.089411373731542e-06,
+      "loss": 0.0499,
+      "step": 934
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.069860833768692e-06,
+      "loss": 0.0822,
+      "step": 935
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.050317953736511e-06,
+      "loss": 0.0607,
+      "step": 936
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.030782811192668e-06,
+      "loss": 0.0467,
+      "step": 937
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.011255483664135e-06,
+      "loss": 0.0704,
+      "step": 938
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.991736048646859e-06,
+      "loss": 0.0633,
+      "step": 939
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.972224583605484e-06,
+      "loss": 0.0717,
+      "step": 940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.952721165972996e-06,
+      "loss": 0.0719,
+      "step": 941
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.93322587315047e-06,
+      "loss": 0.0747,
+      "step": 942
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.913738782506727e-06,
+      "loss": 0.0873,
+      "step": 943
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.89425997137803e-06,
+      "loss": 0.0648,
+      "step": 944
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.874789517067801e-06,
+      "loss": 0.0647,
+      "step": 945
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.855327496846277e-06,
+      "loss": 0.0767,
+      "step": 946
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.835873987950238e-06,
+      "loss": 0.0564,
+      "step": 947
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.81642906758268e-06,
+      "loss": 0.0498,
+      "step": 948
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.796992812912517e-06,
+      "loss": 0.0703,
+      "step": 949
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.777565301074274e-06,
+      "loss": 0.0517,
+      "step": 950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.758146609167773e-06,
+      "loss": 0.0613,
+      "step": 951
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.738736814257844e-06,
+      "loss": 0.0588,
+      "step": 952
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.719335993373989e-06,
+      "loss": 0.0609,
+      "step": 953
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.69994422351012e-06,
+      "loss": 0.0494,
+      "step": 954
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.680561581624213e-06,
+      "loss": 0.0546,
+      "step": 955
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.661188144638027e-06,
+      "loss": 0.0644,
+      "step": 956
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.64182398943678e-06,
+      "loss": 0.0641,
+      "step": 957
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.622469192868866e-06,
+      "loss": 0.0786,
+      "step": 958
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.603123831745536e-06,
+      "loss": 0.0593,
+      "step": 959
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.583787982840588e-06,
+      "loss": 0.0574,
+      "step": 960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.564461722890082e-06,
+      "loss": 0.0661,
+      "step": 961
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.0699,
+      "step": 962
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.525838276606016e-06,
+      "loss": 0.0471,
+      "step": 963
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.506541243553072e-06,
+      "loss": 0.0404,
+      "step": 964
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4872541060151945e-06,
+      "loss": 0.0597,
+      "step": 965
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.467976940535113e-06,
+      "loss": 0.0667,
+      "step": 966
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.448709823615995e-06,
+      "loss": 0.0618,
+      "step": 967
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.429452831721128e-06,
+      "loss": 0.0623,
+      "step": 968
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.410206041273605e-06,
+      "loss": 0.0769,
+      "step": 969
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.390969528656051e-06,
+      "loss": 0.0589,
+      "step": 970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.371743370210289e-06,
+      "loss": 0.0584,
+      "step": 971
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.352527642237064e-06,
+      "loss": 0.0507,
+      "step": 972
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.333322420995708e-06,
+      "loss": 0.0691,
+      "step": 973
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.31412778270387e-06,
+      "loss": 0.0597,
+      "step": 974
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.294943803537202e-06,
+      "loss": 0.0677,
+      "step": 975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.275770559629042e-06,
+      "loss": 0.0586,
+      "step": 976
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.256608127070137e-06,
+      "loss": 0.0721,
+      "step": 977
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2374565819083155e-06,
+      "loss": 0.0423,
+      "step": 978
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2183160001482075e-06,
+      "loss": 0.058,
+      "step": 979
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.199186457750931e-06,
+      "loss": 0.0751,
+      "step": 980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.180068030633798e-06,
+      "loss": 0.0601,
+      "step": 981
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.160960794669993e-06,
+      "loss": 0.0616,
+      "step": 982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.141864825688307e-06,
+      "loss": 0.0595,
+      "step": 983
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.122780199472808e-06,
+      "loss": 0.0511,
+      "step": 984
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1037069917625456e-06,
+      "loss": 0.0504,
+      "step": 985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.084645278251264e-06,
+      "loss": 0.0649,
+      "step": 986
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.065595134587078e-06,
+      "loss": 0.0634,
+      "step": 987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.046556636372202e-06,
+      "loss": 0.0647,
+      "step": 988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.027529859162616e-06,
+      "loss": 0.0488,
+      "step": 989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.008514878467805e-06,
+      "loss": 0.0489,
+      "step": 990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.989511769750421e-06,
+      "loss": 0.0687,
+      "step": 991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.97052060842601e-06,
+      "loss": 0.0668,
+      "step": 992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.951541469862706e-06,
+      "loss": 0.0637,
+      "step": 993
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.932574429380918e-06,
+      "loss": 0.0567,
+      "step": 994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.913619562253055e-06,
+      "loss": 0.0778,
+      "step": 995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.894676943703207e-06,
+      "loss": 0.0471,
+      "step": 996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.875746648906862e-06,
+      "loss": 0.0712,
+      "step": 997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.85682875299059e-06,
+      "loss": 0.0723,
+      "step": 998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.837923331031761e-06,
+      "loss": 0.0777,
+      "step": 999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.819030458058243e-06,
+      "loss": 0.0796,
+      "step": 1000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.800150209048098e-06,
+      "loss": 0.0655,
+      "step": 1001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7812826589292936e-06,
+      "loss": 0.0837,
+      "step": 1002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.76242788257939e-06,
+      "loss": 0.0505,
+      "step": 1003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.74358595482527e-06,
+      "loss": 0.0891,
+      "step": 1004
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.724756950442808e-06,
+      "loss": 0.0752,
+      "step": 1005
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.705940944156604e-06,
+      "loss": 0.0557,
+      "step": 1006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.687138010639667e-06,
+      "loss": 0.0687,
+      "step": 1007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.668348224513127e-06,
+      "loss": 0.0693,
+      "step": 1008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.649571660345945e-06,
+      "loss": 0.0605,
+      "step": 1009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.630808392654593e-06,
+      "loss": 0.0778,
+      "step": 1010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.612058495902791e-06,
+      "loss": 0.0795,
+      "step": 1011
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.593322044501185e-06,
+      "loss": 0.0762,
+      "step": 1012
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.574599112807071e-06,
+      "loss": 0.0679,
+      "step": 1013
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.555889775124077e-06,
+      "loss": 0.0766,
+      "step": 1014
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.537194105701896e-06,
+      "loss": 0.0645,
+      "step": 1015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5185121787359686e-06,
+      "loss": 0.0544,
+      "step": 1016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4998440683672e-06,
+      "loss": 0.0605,
+      "step": 1017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.48118984868167e-06,
+      "loss": 0.0668,
+      "step": 1018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.462549593710317e-06,
+      "loss": 0.0666,
+      "step": 1019
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.443923377428672e-06,
+      "loss": 0.0677,
+      "step": 1020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4253112737565426e-06,
+      "loss": 0.0697,
+      "step": 1021
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.406713356557739e-06,
+      "loss": 0.0598,
+      "step": 1022
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3881296996397624e-06,
+      "loss": 0.0642,
+      "step": 1023
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3695603767535275e-06,
+      "loss": 0.0724,
+      "step": 1024
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.351005461593063e-06,
+      "loss": 0.0522,
+      "step": 1025
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.332465027795208e-06,
+      "loss": 0.0602,
+      "step": 1026
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.313939148939348e-06,
+      "loss": 0.0692,
+      "step": 1027
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.295427898547091e-06,
+      "loss": 0.0582,
+      "step": 1028
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2769313500820025e-06,
+      "loss": 0.0572,
+      "step": 1029
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.258449576949292e-06,
+      "loss": 0.0653,
+      "step": 1030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.239982652495539e-06,
+      "loss": 0.0697,
+      "step": 1031
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.221530650008391e-06,
+      "loss": 0.0685,
+      "step": 1032
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.203093642716278e-06,
+      "loss": 0.0521,
+      "step": 1033
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.184671703788123e-06,
+      "loss": 0.0616,
+      "step": 1034
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.166264906333038e-06,
+      "loss": 0.0595,
+      "step": 1035
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.147873323400058e-06,
+      "loss": 0.0721,
+      "step": 1036
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.129497027977829e-06,
+      "loss": 0.0582,
+      "step": 1037
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.111136092994334e-06,
+      "loss": 0.0598,
+      "step": 1038
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.092790591316586e-06,
+      "loss": 0.059,
+      "step": 1039
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.074460595750362e-06,
+      "loss": 0.0546,
+      "step": 1040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.056146179039899e-06,
+      "loss": 0.0611,
+      "step": 1041
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.037847413867595e-06,
+      "loss": 0.0522,
+      "step": 1042
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.01956437285375e-06,
+      "loss": 0.0569,
+      "step": 1043
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0012971285562535e-06,
+      "loss": 0.0681,
+      "step": 1044
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.0727,
+      "step": 1045
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.964810320028129e-06,
+      "loss": 0.0707,
+      "step": 1046
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9465909005986765e-06,
+      "loss": 0.0721,
+      "step": 1047
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9283875674873524e-06,
+      "loss": 0.053,
+      "step": 1048
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.910200392935718e-06,
+      "loss": 0.0577,
+      "step": 1049
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.892029449121214e-06,
+      "loss": 0.0568,
+      "step": 1050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.873874808156856e-06,
+      "loss": 0.0635,
+      "step": 1051
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8557365420909725e-06,
+      "loss": 0.0669,
+      "step": 1052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8376147229069e-06,
+      "loss": 0.0545,
+      "step": 1053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.819509422522711e-06,
+      "loss": 0.0674,
+      "step": 1054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.801420712790905e-06,
+      "loss": 0.0704,
+      "step": 1055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.78334866549816e-06,
+      "loss": 0.0511,
+      "step": 1056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7652933523650205e-06,
+      "loss": 0.0637,
+      "step": 1057
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.747254845045609e-06,
+      "loss": 0.0676,
+      "step": 1058
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.729233215127378e-06,
+      "loss": 0.0579,
+      "step": 1059
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.711228534130766e-06,
+      "loss": 0.0523,
+      "step": 1060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.693240873508981e-06,
+      "loss": 0.0526,
+      "step": 1061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.675270304647665e-06,
+      "loss": 0.0657,
+      "step": 1062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.657316898864634e-06,
+      "loss": 0.0522,
+      "step": 1063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.639380727409594e-06,
+      "loss": 0.0811,
+      "step": 1064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.621461861463846e-06,
+      "loss": 0.0542,
+      "step": 1065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.603560372140029e-06,
+      "loss": 0.0523,
+      "step": 1066
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.585676330481806e-06,
+      "loss": 0.0573,
+      "step": 1067
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.567809807463606e-06,
+      "loss": 0.0479,
+      "step": 1068
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5499608739903256e-06,
+      "loss": 0.0478,
+      "step": 1069
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.532129600897065e-06,
+      "loss": 0.0645,
+      "step": 1070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.514316058948828e-06,
+      "loss": 0.0688,
+      "step": 1071
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.496520318840266e-06,
+      "loss": 0.0655,
+      "step": 1072
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.478742451195358e-06,
+      "loss": 0.0582,
+      "step": 1073
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.46098252656718e-06,
+      "loss": 0.0622,
+      "step": 1074
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.443240615437586e-06,
+      "loss": 0.0649,
+      "step": 1075
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.425516788216942e-06,
+      "loss": 0.0471,
+      "step": 1076
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.40781111524385e-06,
+      "loss": 0.0639,
+      "step": 1077
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3901236667848586e-06,
+      "loss": 0.0646,
+      "step": 1078
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.372454513034208e-06,
+      "loss": 0.0519,
+      "step": 1079
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.354803724113506e-06,
+      "loss": 0.0593,
+      "step": 1080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.337171370071508e-06,
+      "loss": 0.0443,
+      "step": 1081
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3195575208837865e-06,
+      "loss": 0.0629,
+      "step": 1082
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.301962246452486e-06,
+      "loss": 0.0726,
+      "step": 1083
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.28438561660604e-06,
+      "loss": 0.0683,
+      "step": 1084
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.266827701098871e-06,
+      "loss": 0.0511,
+      "step": 1085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.249288569611155e-06,
+      "loss": 0.0679,
+      "step": 1086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.231768291748506e-06,
+      "loss": 0.0536,
+      "step": 1087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.214266937041721e-06,
+      "loss": 0.0593,
+      "step": 1088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.196784574946496e-06,
+      "loss": 0.0654,
+      "step": 1089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1793212748431565e-06,
+      "loss": 0.0866,
+      "step": 1090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.161877106036386e-06,
+      "loss": 0.053,
+      "step": 1091
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.144452137754921e-06,
+      "loss": 0.0696,
+      "step": 1092
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.127046439151322e-06,
+      "loss": 0.0608,
+      "step": 1093
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.109660079301668e-06,
+      "loss": 0.0619,
+      "step": 1094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.092293127205287e-06,
+      "loss": 0.0487,
+      "step": 1095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.074945651784491e-06,
+      "loss": 0.0568,
+      "step": 1096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.057617721884293e-06,
+      "loss": 0.0666,
+      "step": 1097
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.040309406272135e-06,
+      "loss": 0.0825,
+      "step": 1098
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.02302077363763e-06,
+      "loss": 0.075,
+      "step": 1099
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.005751892592266e-06,
+      "loss": 0.0663,
+      "step": 1100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.98850283166915e-06,
+      "loss": 0.0498,
+      "step": 1101
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9712736593227285e-06,
+      "loss": 0.0471,
+      "step": 1102
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.954064443928515e-06,
+      "loss": 0.0767,
+      "step": 1103
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.936875253782841e-06,
+      "loss": 0.0521,
+      "step": 1104
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.919706157102533e-06,
+      "loss": 0.0494,
+      "step": 1105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9025572220247074e-06,
+      "loss": 0.0665,
+      "step": 1106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.885428516606449e-06,
+      "loss": 0.0611,
+      "step": 1107
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.868320108824564e-06,
+      "loss": 0.0559,
+      "step": 1108
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.851232066575304e-06,
+      "loss": 0.0532,
+      "step": 1109
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.834164457674099e-06,
+      "loss": 0.0531,
+      "step": 1110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.817117349855297e-06,
+      "loss": 0.0596,
+      "step": 1111
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8000908107718625e-06,
+      "loss": 0.0684,
+      "step": 1112
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.783084907995156e-06,
+      "loss": 0.063,
+      "step": 1113
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.766099709014628e-06,
+      "loss": 0.0634,
+      "step": 1114
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.74913528123756e-06,
+      "loss": 0.0522,
+      "step": 1115
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.732191691988822e-06,
+      "loss": 0.0612,
+      "step": 1116
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.715269008510552e-06,
+      "loss": 0.0597,
+      "step": 1117
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.698367297961954e-06,
+      "loss": 0.0604,
+      "step": 1118
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.681486627418979e-06,
+      "loss": 0.0656,
+      "step": 1119
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.664627063874083e-06,
+      "loss": 0.0619,
+      "step": 1120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.647788674235961e-06,
+      "loss": 0.0743,
+      "step": 1121
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.630971525329274e-06,
+      "loss": 0.0485,
+      "step": 1122
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.614175683894384e-06,
+      "loss": 0.0691,
+      "step": 1123
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.597401216587104e-06,
+      "loss": 0.0716,
+      "step": 1124
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.58064818997841e-06,
+      "loss": 0.0459,
+      "step": 1125
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.563916670554196e-06,
+      "loss": 0.0554,
+      "step": 1126
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.547206724714996e-06,
+      "loss": 0.0534,
+      "step": 1127
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.07,
+      "step": 1128
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.513851818965449e-06,
+      "loss": 0.0471,
+      "step": 1129
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.497206991427036e-06,
+      "loss": 0.0541,
+      "step": 1130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.480584002216999e-06,
+      "loss": 0.0701,
+      "step": 1131
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.463982917305155e-06,
+      "loss": 0.0693,
+      "step": 1132
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.447403802574406e-06,
+      "loss": 0.0579,
+      "step": 1133
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.430846723820453e-06,
+      "loss": 0.0448,
+      "step": 1134
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4143117467515514e-06,
+      "loss": 0.0685,
+      "step": 1135
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.397798936988251e-06,
+      "loss": 0.0611,
+      "step": 1136
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.381308360063107e-06,
+      "loss": 0.0721,
+      "step": 1137
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3648400814204665e-06,
+      "loss": 0.0766,
+      "step": 1138
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.348394166416169e-06,
+      "loss": 0.0606,
+      "step": 1139
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3319706803173e-06,
+      "loss": 0.0483,
+      "step": 1140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.315569688301953e-06,
+      "loss": 0.0504,
+      "step": 1141
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.299191255458922e-06,
+      "loss": 0.0694,
+      "step": 1142
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.282835446787504e-06,
+      "loss": 0.0657,
+      "step": 1143
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.266502327197182e-06,
+      "loss": 0.0719,
+      "step": 1144
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.250191961507416e-06,
+      "loss": 0.0594,
+      "step": 1145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.233904414447355e-06,
+      "loss": 0.0582,
+      "step": 1146
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217639750655595e-06,
+      "loss": 0.0509,
+      "step": 1147
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.201398034679911e-06,
+      "loss": 0.0595,
+      "step": 1148
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.185179330977011e-06,
+      "loss": 0.0438,
+      "step": 1149
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.168983703912285e-06,
+      "loss": 0.0575,
+      "step": 1150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.152811217759529e-06,
+      "loss": 0.0618,
+      "step": 1151
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1366619367007086e-06,
+      "loss": 0.0573,
+      "step": 1152
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1205359248256946e-06,
+      "loss": 0.0616,
+      "step": 1153
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.104433246132015e-06,
+      "loss": 0.0583,
+      "step": 1154
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.088353964524593e-06,
+      "loss": 0.0597,
+      "step": 1155
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.072298143815508e-06,
+      "loss": 0.0559,
+      "step": 1156
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0562658477237235e-06,
+      "loss": 0.0726,
+      "step": 1157
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.040257139874848e-06,
+      "loss": 0.0652,
+      "step": 1158
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.024272083800877e-06,
+      "loss": 0.052,
+      "step": 1159
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.008310742939939e-06,
+      "loss": 0.0564,
+      "step": 1160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.992373180636051e-06,
+      "loss": 0.0594,
+      "step": 1161
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9764594601388565e-06,
+      "loss": 0.0636,
+      "step": 1162
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.960569644603395e-06,
+      "loss": 0.0832,
+      "step": 1163
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.944703797089814e-06,
+      "loss": 0.0829,
+      "step": 1164
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.928861980563164e-06,
+      "loss": 0.0464,
+      "step": 1165
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.913044257893114e-06,
+      "loss": 0.0753,
+      "step": 1166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8972506918537125e-06,
+      "loss": 0.0564,
+      "step": 1167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.881481345123158e-06,
+      "loss": 0.063,
+      "step": 1168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865736280283503e-06,
+      "loss": 0.0547,
+      "step": 1169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.850015559820465e-06,
+      "loss": 0.0652,
+      "step": 1170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.834319246123129e-06,
+      "loss": 0.065,
+      "step": 1171
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.818647401483724e-06,
+      "loss": 0.049,
+      "step": 1172
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.803000088097384e-06,
+      "loss": 0.0616,
+      "step": 1173
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7873773680618618e-06,
+      "loss": 0.0694,
+      "step": 1174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.771779303377342e-06,
+      "loss": 0.0598,
+      "step": 1175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7562059559461293e-06,
+      "loss": 0.0716,
+      "step": 1176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.740657387572464e-06,
+      "loss": 0.0474,
+      "step": 1177
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7251336599622323e-06,
+      "loss": 0.057,
+      "step": 1178
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7096348347227406e-06,
+      "loss": 0.048,
+      "step": 1179
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6941609733624707e-06,
+      "loss": 0.0584,
+      "step": 1180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6787121372908274e-06,
+      "loss": 0.0494,
+      "step": 1181
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6632883878179103e-06,
+      "loss": 0.071,
+      "step": 1182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.647889786154254e-06,
+      "loss": 0.0613,
+      "step": 1183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.632516393410589e-06,
+      "loss": 0.0466,
+      "step": 1184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6171682705976086e-06,
+      "loss": 0.0637,
+      "step": 1185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.601845478625714e-06,
+      "loss": 0.0583,
+      "step": 1186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5865480783047802e-06,
+      "loss": 0.07,
+      "step": 1187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5712761303439246e-06,
+      "loss": 0.0622,
+      "step": 1188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5560296953512296e-06,
+      "loss": 0.0658,
+      "step": 1189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5408088338335543e-06,
+      "loss": 0.0517,
+      "step": 1190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.525613606196249e-06,
+      "loss": 0.0583,
+      "step": 1191
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5104440727429378e-06,
+      "loss": 0.0609,
+      "step": 1192
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4953002936752876e-06,
+      "loss": 0.0499,
+      "step": 1193
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.480182329092732e-06,
+      "loss": 0.0478,
+      "step": 1194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4650902389922826e-06,
+      "loss": 0.0569,
+      "step": 1195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4500240832682454e-06,
+      "loss": 0.0615,
+      "step": 1196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4349839217120194e-06,
+      "loss": 0.0635,
+      "step": 1197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4199698140118354e-06,
+      "loss": 0.0603,
+      "step": 1198
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.404981819752524e-06,
+      "loss": 0.0578,
+      "step": 1199
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.390019998415297e-06,
+      "loss": 0.0502,
+      "step": 1200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3750844093774736e-06,
+      "loss": 0.0737,
+      "step": 1201
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.360175111912292e-06,
+      "loss": 0.063,
+      "step": 1202
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3452921651886315e-06,
+      "loss": 0.0596,
+      "step": 1203
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.330435628270806e-06,
+      "loss": 0.0569,
+      "step": 1204
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3156055601183156e-06,
+      "loss": 0.0742,
+      "step": 1205
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.300802019585615e-06,
+      "loss": 0.0562,
+      "step": 1206
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2860250654218916e-06,
+      "loss": 0.0644,
+      "step": 1207
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2712747562708115e-06,
+      "loss": 0.0612,
+      "step": 1208
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2565511506703007e-06,
+      "loss": 0.0546,
+      "step": 1209
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2418543070523134e-06,
+      "loss": 0.0539,
+      "step": 1210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.0539,
+      "step": 1211
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2125411389604397e-06,
+      "loss": 0.0691,
+      "step": 1212
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1979249308184955e-06,
+      "loss": 0.0588,
+      "step": 1213
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.18333571732249e-06,
+      "loss": 0.0727,
+      "step": 1214
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.168773556371034e-06,
+      "loss": 0.0572,
+      "step": 1215
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.154238505755367e-06,
+      "loss": 0.0721,
+      "step": 1216
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.139730623159144e-06,
+      "loss": 0.0673,
+      "step": 1217
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1252499661582016e-06,
+      "loss": 0.0627,
+      "step": 1218
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.110796592220322e-06,
+      "loss": 0.0856,
+      "step": 1219
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0963705587050286e-06,
+      "loss": 0.0603,
+      "step": 1220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.081971922863319e-06,
+      "loss": 0.0895,
+      "step": 1221
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.067600741837483e-06,
+      "loss": 0.0665,
+      "step": 1222
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.05325707266084e-06,
+      "loss": 0.0794,
+      "step": 1223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0389409722575258e-06,
+      "loss": 0.0684,
+      "step": 1224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0246524974422808e-06,
+      "loss": 0.0544,
+      "step": 1225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.010391704920187e-06,
+      "loss": 0.0552,
+      "step": 1226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9961586512864947e-06,
+      "loss": 0.0617,
+      "step": 1227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9819533930263434e-06,
+      "loss": 0.0457,
+      "step": 1228
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.967775986514585e-06,
+      "loss": 0.0671,
+      "step": 1229
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9536264880155286e-06,
+      "loss": 0.0571,
+      "step": 1230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.939504953682726e-06,
+      "loss": 0.0442,
+      "step": 1231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.925411439558764e-06,
+      "loss": 0.0519,
+      "step": 1232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9113460015750073e-06,
+      "loss": 0.0651,
+      "step": 1233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8973086955514194e-06,
+      "loss": 0.0581,
+      "step": 1234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.883299577196308e-06,
+      "loss": 0.0485,
+      "step": 1235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8693187021061164e-06,
+      "loss": 0.0631,
+      "step": 1236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8553661257652045e-06,
+      "loss": 0.0595,
+      "step": 1237
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8414419035456197e-06,
+      "loss": 0.0545,
+      "step": 1238
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8275460907068974e-06,
+      "loss": 0.0613,
+      "step": 1239
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8136787423958147e-06,
+      "loss": 0.064,
+      "step": 1240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7998399136461884e-06,
+      "loss": 0.0581,
+      "step": 1241
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7860296593786553e-06,
+      "loss": 0.0463,
+      "step": 1242
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7722480344004487e-06,
+      "loss": 0.0515,
+      "step": 1243
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7584950934051823e-06,
+      "loss": 0.067,
+      "step": 1244
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.744770890972648e-06,
+      "loss": 0.0496,
+      "step": 1245
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7310754815685627e-06,
+      "loss": 0.056,
+      "step": 1246
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7174089195443985e-06,
+      "loss": 0.0485,
+      "step": 1247
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7037712591371324e-06,
+      "loss": 0.0484,
+      "step": 1248
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6901625544690436e-06,
+      "loss": 0.0641,
+      "step": 1249
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6765828595474985e-06,
+      "loss": 0.0571,
+      "step": 1250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6630322282647336e-06,
+      "loss": 0.0438,
+      "step": 1251
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6495107143976573e-06,
+      "loss": 0.045,
+      "step": 1252
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6360183716075984e-06,
+      "loss": 0.0474,
+      "step": 1253
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.622555253440139e-06,
+      "loss": 0.0626,
+      "step": 1254
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.609121413324872e-06,
+      "loss": 0.0658,
+      "step": 1255
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5957169045751962e-06,
+      "loss": 0.0655,
+      "step": 1256
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5823417803881145e-06,
+      "loss": 0.0471,
+      "step": 1257
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5689960938440006e-06,
+      "loss": 0.0724,
+      "step": 1258
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.555679897906417e-06,
+      "loss": 0.0554,
+      "step": 1259
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5423932454218804e-06,
+      "loss": 0.063,
+      "step": 1260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5291361891196654e-06,
+      "loss": 0.0719,
+      "step": 1261
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.515908781611591e-06,
+      "loss": 0.0546,
+      "step": 1262
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5027110753918093e-06,
+      "loss": 0.0465,
+      "step": 1263
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4895431228366106e-06,
+      "loss": 0.0504,
+      "step": 1264
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4764049762041874e-06,
+      "loss": 0.05,
+      "step": 1265
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.463296687634462e-06,
+      "loss": 0.073,
+      "step": 1266
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.450218309148853e-06,
+      "loss": 0.0618,
+      "step": 1267
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.437169892650081e-06,
+      "loss": 0.0597,
+      "step": 1268
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.424151489921958e-06,
+      "loss": 0.0561,
+      "step": 1269
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4111631526291846e-06,
+      "loss": 0.0673,
+      "step": 1270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.398204932317141e-06,
+      "loss": 0.0507,
+      "step": 1271
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3852768804116955e-06,
+      "loss": 0.0752,
+      "step": 1272
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.372379048218979e-06,
+      "loss": 0.0512,
+      "step": 1273
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.359511486925199e-06,
+      "loss": 0.068,
+      "step": 1274
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3466742475964266e-06,
+      "loss": 0.0636,
+      "step": 1275
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3338673811783974e-06,
+      "loss": 0.061,
+      "step": 1276
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.32109093849632e-06,
+      "loss": 0.0418,
+      "step": 1277
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3083449702546425e-06,
+      "loss": 0.0718,
+      "step": 1278
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2956295270368965e-06,
+      "loss": 0.0744,
+      "step": 1279
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2829446593054495e-06,
+      "loss": 0.0828,
+      "step": 1280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2702904174013475e-06,
+      "loss": 0.0636,
+      "step": 1281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2576668515440824e-06,
+      "loss": 0.0489,
+      "step": 1282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2450740118314083e-06,
+      "loss": 0.0513,
+      "step": 1283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2325119482391466e-06,
+      "loss": 0.05,
+      "step": 1284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.219980710620967e-06,
+      "loss": 0.0573,
+      "step": 1285
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2074803487082164e-06,
+      "loss": 0.0681,
+      "step": 1286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1950109121097043e-06,
+      "loss": 0.0617,
+      "step": 1287
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1825724503115064e-06,
+      "loss": 0.0714,
+      "step": 1288
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1701650126767824e-06,
+      "loss": 0.0522,
+      "step": 1289
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1577886484455535e-06,
+      "loss": 0.0427,
+      "step": 1290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.145443406734542e-06,
+      "loss": 0.0444,
+      "step": 1291
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1331293365369444e-06,
+      "loss": 0.0484,
+      "step": 1292
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1208464867222544e-06,
+      "loss": 0.0598,
+      "step": 1293
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.0467,
+      "step": 1294
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0963746430998753e-06,
+      "loss": 0.0446,
+      "step": 1295
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.084185746410894e-06,
+      "loss": 0.0539,
+      "step": 1296
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0720282643418577e-06,
+      "loss": 0.0592,
+      "step": 1297
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0599022451408226e-06,
+      "loss": 0.0683,
+      "step": 1298
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0478077369309856e-06,
+      "loss": 0.0474,
+      "step": 1299
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0357447877104897e-06,
+      "loss": 0.0561,
+      "step": 1300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.023713445352232e-06,
+      "loss": 0.0514,
+      "step": 1301
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.011713757603675e-06,
+      "loss": 0.0619,
+      "step": 1302
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9997457720866554e-06,
+      "loss": 0.0679,
+      "step": 1303
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9878095362972037e-06,
+      "loss": 0.0452,
+      "step": 1304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9759050976053407e-06,
+      "loss": 0.0645,
+      "step": 1305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9640325032549023e-06,
+      "loss": 0.0489,
+      "step": 1306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9521918003633446e-06,
+      "loss": 0.0537,
+      "step": 1307
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.940383035921558e-06,
+      "loss": 0.0704,
+      "step": 1308
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9286062567936935e-06,
+      "loss": 0.0615,
+      "step": 1309
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.916861509716945e-06,
+      "loss": 0.0522,
+      "step": 1310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9051488413014063e-06,
+      "loss": 0.058,
+      "step": 1311
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8934682980298502e-06,
+      "loss": 0.0448,
+      "step": 1312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8818199262575598e-06,
+      "loss": 0.0421,
+      "step": 1313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8702037722121523e-06,
+      "loss": 0.0561,
+      "step": 1314
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8586198819933688e-06,
+      "loss": 0.062,
+      "step": 1315
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8470683015729273e-06,
+      "loss": 0.0693,
+      "step": 1316
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.835549076794303e-06,
+      "loss": 0.0711,
+      "step": 1317
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8240622533725816e-06,
+      "loss": 0.0674,
+      "step": 1318
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8126078768942512e-06,
+      "loss": 0.0562,
+      "step": 1319
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8011859928170338e-06,
+      "loss": 0.0743,
+      "step": 1320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7897966464697036e-06,
+      "loss": 0.066,
+      "step": 1321
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7784398830519002e-06,
+      "loss": 0.0647,
+      "step": 1322
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.767115747633965e-06,
+      "loss": 0.0597,
+      "step": 1323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7558242851567442e-06,
+      "loss": 0.0717,
+      "step": 1324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.744565540431421e-06,
+      "loss": 0.0728,
+      "step": 1325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7333395581393364e-06,
+      "loss": 0.046,
+      "step": 1326
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7221463828318074e-06,
+      "loss": 0.0685,
+      "step": 1327
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7109860589299554e-06,
+      "loss": 0.0498,
+      "step": 1328
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6998586307245313e-06,
+      "loss": 0.0641,
+      "step": 1329
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6887641423757328e-06,
+      "loss": 0.0558,
+      "step": 1330
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6777026379130323e-06,
+      "loss": 0.0692,
+      "step": 1331
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6666741612350034e-06,
+      "loss": 0.079,
+      "step": 1332
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6556787561091492e-06,
+      "loss": 0.0392,
+      "step": 1333
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6447164661717197e-06,
+      "loss": 0.0639,
+      "step": 1334
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6337873349275457e-06,
+      "loss": 0.0684,
+      "step": 1335
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6228914057498747e-06,
+      "loss": 0.0651,
+      "step": 1336
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.612028721880169e-06,
+      "loss": 0.0634,
+      "step": 1337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6011993264279735e-06,
+      "loss": 0.0488,
+      "step": 1338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5904032623707144e-06,
+      "loss": 0.0654,
+      "step": 1339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5796405725535401e-06,
+      "loss": 0.0587,
+      "step": 1340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5689112996891576e-06,
+      "loss": 0.0528,
+      "step": 1341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5582154863576415e-06,
+      "loss": 0.0566,
+      "step": 1342
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5475531750062955e-06,
+      "loss": 0.0617,
+      "step": 1343
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5369244079494561e-06,
+      "loss": 0.0666,
+      "step": 1344
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5263292273683405e-06,
+      "loss": 0.0535,
+      "step": 1345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5157676753108752e-06,
+      "loss": 0.0536,
+      "step": 1346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5052397936915264e-06,
+      "loss": 0.0457,
+      "step": 1347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4947456242911407e-06,
+      "loss": 0.0645,
+      "step": 1348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4842852087567727e-06,
+      "loss": 0.0584,
+      "step": 1349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4738585886015178e-06,
+      "loss": 0.0468,
+      "step": 1350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4634658052043581e-06,
+      "loss": 0.0669,
+      "step": 1351
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.453106899809985e-06,
+      "loss": 0.0624,
+      "step": 1352
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.442781913528647e-06,
+      "loss": 0.0612,
+      "step": 1353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4324908873359766e-06,
+      "loss": 0.0515,
+      "step": 1354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4222338620728405e-06,
+      "loss": 0.0671,
+      "step": 1355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4120108784451625e-06,
+      "loss": 0.0502,
+      "step": 1356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4018219770237718e-06,
+      "loss": 0.0562,
+      "step": 1357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3916671982442387e-06,
+      "loss": 0.0561,
+      "step": 1358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3815465824067154e-06,
+      "loss": 0.071,
+      "step": 1359
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3714601696757713e-06,
+      "loss": 0.0671,
+      "step": 1360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3614080000802488e-06,
+      "loss": 0.0699,
+      "step": 1361
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.351390113513078e-06,
+      "loss": 0.0429,
+      "step": 1362
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.341406549731148e-06,
+      "loss": 0.0447,
+      "step": 1363
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3314573483551253e-06,
+      "loss": 0.0499,
+      "step": 1364
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.321542548869308e-06,
+      "loss": 0.056,
+      "step": 1365
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3116621906214744e-06,
+      "loss": 0.0495,
+      "step": 1366
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3018163128227058e-06,
+      "loss": 0.042,
+      "step": 1367
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2920049545472602e-06,
+      "loss": 0.0697,
+      "step": 1368
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2822281547323867e-06,
+      "loss": 0.0571,
+      "step": 1369
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2724859521781996e-06,
+      "loss": 0.044,
+      "step": 1370
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.262778385547504e-06,
+      "loss": 0.0781,
+      "step": 1371
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2531054933656462e-06,
+      "loss": 0.0408,
+      "step": 1372
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2434673140203746e-06,
+      "loss": 0.0575,
+      "step": 1373
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2338638857616615e-06,
+      "loss": 0.0536,
+      "step": 1374
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.22429524670158e-06,
+      "loss": 0.041,
+      "step": 1375
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2147614348141334e-06,
+      "loss": 0.0581,
+      "step": 1376
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.0815,
+      "step": 1377
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.195798443761933e-06,
+      "loss": 0.0559,
+      "step": 1378
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1863693398535115e-06,
+      "loss": 0.0434,
+      "step": 1379
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1769752136300927e-06,
+      "loss": 0.0806,
+      "step": 1380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1676161023731115e-06,
+      "loss": 0.0751,
+      "step": 1381
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.158292043225039e-06,
+      "loss": 0.058,
+      "step": 1382
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1490030731892422e-06,
+      "loss": 0.0448,
+      "step": 1383
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.139749229129834e-06,
+      "loss": 0.0447,
+      "step": 1384
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1305305477715256e-06,
+      "loss": 0.0462,
+      "step": 1385
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1213470656994818e-06,
+      "loss": 0.0427,
+      "step": 1386
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1121988193591737e-06,
+      "loss": 0.0607,
+      "step": 1387
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1030858450562443e-06,
+      "loss": 0.0585,
+      "step": 1388
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0940081789563462e-06,
+      "loss": 0.0449,
+      "step": 1389
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0849658570850153e-06,
+      "loss": 0.0553,
+      "step": 1390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0759589153275162e-06,
+      "loss": 0.0675,
+      "step": 1391
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0669873894287052e-06,
+      "loss": 0.0531,
+      "step": 1392
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0580513149928961e-06,
+      "loss": 0.0551,
+      "step": 1393
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0491507274836922e-06,
+      "loss": 0.052,
+      "step": 1394
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0402856622238832e-06,
+      "loss": 0.0431,
+      "step": 1395
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0314561543952728e-06,
+      "loss": 0.0527,
+      "step": 1396
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0226622390385553e-06,
+      "loss": 0.0531,
+      "step": 1397
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.01390395105318e-06,
+      "loss": 0.0528,
+      "step": 1398
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0051813251971897e-06,
+      "loss": 0.0748,
+      "step": 1399
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.964943960871187e-07,
+      "loss": 0.0414,
+      "step": 1400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.878431981978177e-07,
+      "loss": 0.0418,
+      "step": 1401
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.792277658623461e-07,
+      "loss": 0.0383,
+      "step": 1402
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.70648133271821e-07,
+      "loss": 0.0653,
+      "step": 1403
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.621043344752834e-07,
+      "loss": 0.0468,
+      "step": 1404
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.535964033795709e-07,
+      "loss": 0.0393,
+      "step": 1405
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.451243737491656e-07,
+      "loss": 0.0463,
+      "step": 1406
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.366882792060827e-07,
+      "loss": 0.0546,
+      "step": 1407
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.282881532297205e-07,
+      "loss": 0.0699,
+      "step": 1408
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.199240291567335e-07,
+      "loss": 0.0528,
+      "step": 1409
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.115959401808983e-07,
+      "loss": 0.065,
+      "step": 1410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.033039193529858e-07,
+      "loss": 0.049,
+      "step": 1411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.950479995806216e-07,
+      "loss": 0.0518,
+      "step": 1412
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.868282136281703e-07,
+      "loss": 0.0409,
+      "step": 1413
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.786445941165878e-07,
+      "loss": 0.037,
+      "step": 1414
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.704971735233048e-07,
+      "loss": 0.0394,
+      "step": 1415
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.623859841820902e-07,
+      "loss": 0.044,
+      "step": 1416
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.543110582829272e-07,
+      "loss": 0.056,
+      "step": 1417
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.462724278718882e-07,
+      "loss": 0.064,
+      "step": 1418
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.38270124850995e-07,
+      "loss": 0.084,
+      "step": 1419
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.303041809781087e-07,
+      "loss": 0.0408,
+      "step": 1420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.223746278667944e-07,
+      "loss": 0.0529,
+      "step": 1421
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.144814969861936e-07,
+      "loss": 0.0571,
+      "step": 1422
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.066248196609073e-07,
+      "loss": 0.0502,
+      "step": 1423
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.988046270708615e-07,
+      "loss": 0.0439,
+      "step": 1424
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.91020950251199e-07,
+      "loss": 0.0636,
+      "step": 1425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83273820092133e-07,
+      "loss": 0.0573,
+      "step": 1426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.755632673388524e-07,
+      "loss": 0.0514,
+      "step": 1427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.678893225913742e-07,
+      "loss": 0.0587,
+      "step": 1428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.60252016304438e-07,
+      "loss": 0.0695,
+      "step": 1429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.526513787873834e-07,
+      "loss": 0.0399,
+      "step": 1430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.450874402040176e-07,
+      "loss": 0.0418,
+      "step": 1431
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.375602305725138e-07,
+      "loss": 0.0556,
+      "step": 1432
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.3006977976528e-07,
+      "loss": 0.0585,
+      "step": 1433
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.22616117508842e-07,
+      "loss": 0.0542,
+      "step": 1434
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.151992733837276e-07,
+      "loss": 0.0523,
+      "step": 1435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.078192768243486e-07,
+      "loss": 0.0642,
+      "step": 1436
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.004761571188856e-07,
+      "loss": 0.0692,
+      "step": 1437
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.931699434091676e-07,
+      "loss": 0.0489,
+      "step": 1438
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.85900664690562e-07,
+      "loss": 0.0693,
+      "step": 1439
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.786683498118518e-07,
+      "loss": 0.0504,
+      "step": 1440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.714730274751303e-07,
+      "loss": 0.0378,
+      "step": 1441
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.643147262356809e-07,
+      "loss": 0.0536,
+      "step": 1442
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.571934745018627e-07,
+      "loss": 0.0619,
+      "step": 1443
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.501093005350023e-07,
+      "loss": 0.0528,
+      "step": 1444
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.430622324492853e-07,
+      "loss": 0.0501,
+      "step": 1445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.360522982116301e-07,
+      "loss": 0.0505,
+      "step": 1446
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.290795256415927e-07,
+      "loss": 0.0631,
+      "step": 1447
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.221439424112463e-07,
+      "loss": 0.0431,
+      "step": 1448
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.152455760450749e-07,
+      "loss": 0.0626,
+      "step": 1449
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.083844539198691e-07,
+      "loss": 0.0562,
+      "step": 1450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.015606032646026e-07,
+      "loss": 0.0785,
+      "step": 1451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.947740511603461e-07,
+      "loss": 0.0458,
+      "step": 1452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.880248245401354e-07,
+      "loss": 0.0663,
+      "step": 1453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.813129501888859e-07,
+      "loss": 0.07,
+      "step": 1454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.746384547432738e-07,
+      "loss": 0.0753,
+      "step": 1455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.680013646916316e-07,
+      "loss": 0.0505,
+      "step": 1456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.614017063738519e-07,
+      "loss": 0.0543,
+      "step": 1457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.54839505981265e-07,
+      "loss": 0.0735,
+      "step": 1458
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.483147895565589e-07,
+      "loss": 0.0554,
+      "step": 1459
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.0665,
+      "step": 1460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.353779120376102e-07,
+      "loss": 0.0541,
+      "step": 1461
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.289658022845323e-07,
+      "loss": 0.0795,
+      "step": 1462
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.22591279181447e-07,
+      "loss": 0.0708,
+      "step": 1463
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.162543680262267e-07,
+      "loss": 0.0357,
+      "step": 1464
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.099550939674691e-07,
+      "loss": 0.047,
+      "step": 1465
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.036934820044126e-07,
+      "loss": 0.0578,
+      "step": 1466
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.974695569868238e-07,
+      "loss": 0.0611,
+      "step": 1467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.9128334361491e-07,
+      "loss": 0.0453,
+      "step": 1468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.85134866439212e-07,
+      "loss": 0.0666,
+      "step": 1469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.790241498605175e-07,
+      "loss": 0.0497,
+      "step": 1470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.729512181297524e-07,
+      "loss": 0.0534,
+      "step": 1471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.669160953478913e-07,
+      "loss": 0.0584,
+      "step": 1472
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6091880546586307e-07,
+      "loss": 0.0515,
+      "step": 1473
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.549593722844492e-07,
+      "loss": 0.0729,
+      "step": 1474
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4903781945419556e-07,
+      "loss": 0.0506,
+      "step": 1475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.431541704753173e-07,
+      "loss": 0.0455,
+      "step": 1476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3730844869760535e-07,
+      "loss": 0.0496,
+      "step": 1477
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3150067732032895e-07,
+      "loss": 0.0689,
+      "step": 1478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2573087939215217e-07,
+      "loss": 0.0435,
+      "step": 1479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.199990778110363e-07,
+      "loss": 0.0607,
+      "step": 1480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1430529532414886e-07,
+      "loss": 0.0649,
+      "step": 1481
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0864955452778244e-07,
+      "loss": 0.0516,
+      "step": 1482
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0303187786724486e-07,
+      "loss": 0.0416,
+      "step": 1483
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.97452287636797e-07,
+      "loss": 0.0468,
+      "step": 1484
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.919108059795407e-07,
+      "loss": 0.0622,
+      "step": 1485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.864074548873431e-07,
+      "loss": 0.0449,
+      "step": 1486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.809422562007525e-07,
+      "loss": 0.0676,
+      "step": 1487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.755152316088928e-07,
+      "loss": 0.0494,
+      "step": 1488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.701264026494067e-07,
+      "loss": 0.0607,
+      "step": 1489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6477579070833934e-07,
+      "loss": 0.0487,
+      "step": 1490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5946341702007836e-07,
+      "loss": 0.0601,
+      "step": 1491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5418930266725606e-07,
+      "loss": 0.0572,
+      "step": 1492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4895346858066723e-07,
+      "loss": 0.0537,
+      "step": 1493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.437559355391917e-07,
+      "loss": 0.0629,
+      "step": 1494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.385967241697041e-07,
+      "loss": 0.0461,
+      "step": 1495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3347585494699966e-07,
+      "loss": 0.0558,
+      "step": 1496
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.283933481937085e-07,
+      "loss": 0.0887,
+      "step": 1497
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2334922408021385e-07,
+      "loss": 0.0471,
+      "step": 1498
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1834350262457624e-07,
+      "loss": 0.0577,
+      "step": 1499
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1337620369245034e-07,
+      "loss": 0.0532,
+      "step": 1500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0844734699700727e-07,
+      "loss": 0.0556,
+      "step": 1501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0355695209886125e-07,
+      "loss": 0.0505,
+      "step": 1502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.987050384059798e-07,
+      "loss": 0.0478,
+      "step": 1503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.93891625173619e-07,
+      "loss": 0.0691,
+      "step": 1504
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8911673150424314e-07,
+      "loss": 0.0704,
+      "step": 1505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.843803763474462e-07,
+      "loss": 0.0631,
+      "step": 1506
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.796825784998791e-07,
+      "loss": 0.0491,
+      "step": 1507
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.750233566051719e-07,
+      "loss": 0.0571,
+      "step": 1508
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7040272915387024e-07,
+      "loss": 0.0562,
+      "step": 1509
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6582071448334466e-07,
+      "loss": 0.0565,
+      "step": 1510
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.61277330777735e-07,
+      "loss": 0.0483,
+      "step": 1511
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5677259606786686e-07,
+      "loss": 0.0615,
+      "step": 1512
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5230652823118204e-07,
+      "loss": 0.0457,
+      "step": 1513
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.478791449916773e-07,
+      "loss": 0.0494,
+      "step": 1514
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.434904639198155e-07,
+      "loss": 0.0583,
+      "step": 1515
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3914050243247446e-07,
+      "loss": 0.0656,
+      "step": 1516
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3482927779286624e-07,
+      "loss": 0.0665,
+      "step": 1517
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3055680711047356e-07,
+      "loss": 0.0488,
+      "step": 1518
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2632310734097994e-07,
+      "loss": 0.0768,
+      "step": 1519
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2212819528620199e-07,
+      "loss": 0.0589,
+      "step": 1520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.179720875940272e-07,
+      "loss": 0.0533,
+      "step": 1521
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1385480075834076e-07,
+      "loss": 0.0491,
+      "step": 1522
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0977635111896656e-07,
+      "loss": 0.0449,
+      "step": 1523
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0573675486159738e-07,
+      "loss": 0.0428,
+      "step": 1524
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0173602801773496e-07,
+      "loss": 0.0582,
+      "step": 1525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.97774186464621e-07,
+      "loss": 0.0623,
+      "step": 1526
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9385124592518068e-07,
+      "loss": 0.0418,
+      "step": 1527
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8996722196795715e-07,
+      "loss": 0.0615,
+      "step": 1528
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8612213000704705e-07,
+      "loss": 0.0887,
+      "step": 1529
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8231598530204287e-07,
+      "loss": 0.0585,
+      "step": 1530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7854880295797406e-07,
+      "loss": 0.061,
+      "step": 1531
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.748205979252393e-07,
+      "loss": 0.0435,
+      "step": 1532
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.711313849995555e-07,
+      "loss": 0.0513,
+      "step": 1533
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6748117882189886e-07,
+      "loss": 0.0505,
+      "step": 1534
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6386999387843717e-07,
+      "loss": 0.0451,
+      "step": 1535
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6029784450048325e-07,
+      "loss": 0.0591,
+      "step": 1536
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.567647448644327e-07,
+      "loss": 0.0599,
+      "step": 1537
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5327070899170736e-07,
+      "loss": 0.0618,
+      "step": 1538
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4981575074870636e-07,
+      "loss": 0.0658,
+      "step": 1539
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4639988384673843e-07,
+      "loss": 0.0428,
+      "step": 1540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4302312184197975e-07,
+      "loss": 0.0583,
+      "step": 1541
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.396854781354129e-07,
+      "loss": 0.0538,
+      "step": 1542
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.0729,
+      "step": 1543
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3312759844451352e-07,
+      "loss": 0.0376,
+      "step": 1544
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2990738848571493e-07,
+      "loss": 0.0392,
+      "step": 1545
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2672634887607616e-07,
+      "loss": 0.062,
+      "step": 1546
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.235844922398355e-07,
+      "loss": 0.0567,
+      "step": 1547
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2048183104573563e-07,
+      "loss": 0.0629,
+      "step": 1548
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1741837760696595e-07,
+      "loss": 0.0553,
+      "step": 1549
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1439414408111471e-07,
+      "loss": 0.0469,
+      "step": 1550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.114091424701258e-07,
+      "loss": 0.0535,
+      "step": 1551
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0846338462024542e-07,
+      "loss": 0.0526,
+      "step": 1552
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.055568822219799e-07,
+      "loss": 0.0626,
+      "step": 1553
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0268964681004356e-07,
+      "loss": 0.0417,
+      "step": 1554
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.986168976331866e-08,
+      "loss": 0.0585,
+      "step": 1555
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.707302230480553e-08,
+      "loss": 0.0592,
+      "step": 1556
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.432365550158251e-08,
+      "loss": 0.0675,
+      "step": 1557
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.16136002647583e-08,
+      "loss": 0.0584,
+      "step": 1558
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.89428673494308e-08,
+      "loss": 0.0486,
+      "step": 1559
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.631146735464279e-08,
+      "loss": 0.0583,
+      "step": 1560
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.3719410723343e-08,
+      "loss": 0.052,
+      "step": 1561
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.116670774234059e-08,
+      "loss": 0.0648,
+      "step": 1562
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.865336854226525e-08,
+      "loss": 0.0599,
+      "step": 1563
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.617940309753047e-08,
+      "loss": 0.0588,
+      "step": 1564
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.374482122628923e-08,
+      "loss": 0.0413,
+      "step": 1565
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.134963259040173e-08,
+      "loss": 0.0804,
+      "step": 1566
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.899384669538434e-08,
+      "loss": 0.0612,
+      "step": 1567
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.66774728903885e-08,
+      "loss": 0.0409,
+      "step": 1568
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.440052036815081e-08,
+      "loss": 0.075,
+      "step": 1569
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.216299816496186e-08,
+      "loss": 0.058,
+      "step": 1570
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.996491516062964e-08,
+      "loss": 0.0829,
+      "step": 1571
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7806280078444024e-08,
+      "loss": 0.07,
+      "step": 1572
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.568710148514123e-08,
+      "loss": 0.0383,
+      "step": 1573
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.360738779087382e-08,
+      "loss": 0.0439,
+      "step": 1574
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.156714724917078e-08,
+      "loss": 0.0503,
+      "step": 1575
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9566387956909713e-08,
+      "loss": 0.0511,
+      "step": 1576
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.760511785428468e-08,
+      "loss": 0.0505,
+      "step": 1577
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.568334472477287e-08,
+      "loss": 0.0631,
+      "step": 1578
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.380107619510243e-08,
+      "loss": 0.0595,
+      "step": 1579
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.195831973522468e-08,
+      "loss": 0.0586,
+      "step": 1580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.015508265828527e-08,
+      "loss": 0.0582,
+      "step": 1581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8391372120591964e-08,
+      "loss": 0.0601,
+      "step": 1582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.666719512158912e-08,
+      "loss": 0.0605,
+      "step": 1583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.498255850382659e-08,
+      "loss": 0.0696,
+      "step": 1584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.333746895293755e-08,
+      "loss": 0.0612,
+      "step": 1585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.173193299760735e-08,
+      "loss": 0.0604,
+      "step": 1586
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.016595700954916e-08,
+      "loss": 0.068,
+      "step": 1587
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.86395472034795e-08,
+      "loss": 0.0478,
+      "step": 1588
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.715270963709382e-08,
+      "loss": 0.0582,
+      "step": 1589
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5705450211038764e-08,
+      "loss": 0.0615,
+      "step": 1590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.429777466889438e-08,
+      "loss": 0.0779,
+      "step": 1591
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2929688597147505e-08,
+      "loss": 0.0458,
+      "step": 1592
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.160119742517064e-08,
+      "loss": 0.045,
+      "step": 1593
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.031230642520088e-08,
+      "loss": 0.067,
+      "step": 1594
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9063020712316582e-08,
+      "loss": 0.0642,
+      "step": 1595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.785334524442073e-08,
+      "loss": 0.0516,
+      "step": 1596
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6683284822219814e-08,
+      "loss": 0.0549,
+      "step": 1597
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.555284408920388e-08,
+      "loss": 0.0508,
+      "step": 1598
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4462027531627621e-08,
+      "loss": 0.05,
+      "step": 1599
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3410839478493753e-08,
+      "loss": 0.0448,
+      "step": 1600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2399284101538566e-08,
+      "loss": 0.0556,
+      "step": 1601
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1427365415209723e-08,
+      "loss": 0.0582,
+      "step": 1602
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0495087276654049e-08,
+      "loss": 0.0722,
+      "step": 1603
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.602453385699762e-09,
+      "loss": 0.0654,
+      "step": 1604
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.74946728484538e-09,
+      "loss": 0.0486,
+      "step": 1605
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.936132359243066e-09,
+      "loss": 0.0503,
+      "step": 1606
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.162451836685291e-09,
+      "loss": 0.0516,
+      "step": 1607
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.428428787593755e-09,
+      "loss": 0.0434,
+      "step": 1608
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7340661250038235e-09,
+      "loss": 0.0446,
+      "step": 1609
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.079366604555658e-09,
+      "loss": 0.0518,
+      "step": 1610
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.464332824483108e-09,
+      "loss": 0.045,
+      "step": 1611
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.888967225604834e-09,
+      "loss": 0.054,
+      "step": 1612
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.353272091309867e-09,
+      "loss": 0.0743,
+      "step": 1613
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8572495475509553e-09,
+      "loss": 0.046,
+      "step": 1614
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.400901562840119e-09,
+      "loss": 0.0793,
+      "step": 1615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.984229948235328e-09,
+      "loss": 0.035,
+      "step": 1616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6072363573338412e-09,
+      "loss": 0.0519,
+      "step": 1617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2699222862699868e-09,
+      "loss": 0.0559,
+      "step": 1618
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.722890737029478e-10,
+      "loss": 0.0643,
+      "step": 1619
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.143379008200946e-10,
+      "loss": 0.0642,
+      "step": 1620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.960697913203305e-10,
+      "loss": 0.0362,
+      "step": 1621
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.174856114229741e-10,
+      "loss": 0.053,
+      "step": 1622
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7858606985443616e-10,
+      "loss": 0.0692,
+      "step": 1623
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.937171784933028e-11,
+      "loss": 0.0524,
+      "step": 1624
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9842949149362358e-11,
+      "loss": 0.0488,
+      "step": 1625
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.0335,
+      "step": 1626
+    },
+    {
+      "epoch": 2.0,
+      "step": 1626,
+      "total_flos": 43698468323328.0,
+      "train_loss": 0.08158567177560204,
+      "train_runtime": 9449.8527,
+      "train_samples_per_second": 11.005,
+      "train_steps_per_second": 0.172
+    }
+  ],
+  "max_steps": 1626,
+  "num_train_epochs": 2,
+  "total_flos": 43698468323328.0,
+  "trial_name": null,
+  "trial_params": null
+}