diff --git "a/multimodal-vicuna-7b-v1.5-audio-damc/trainer_state.json" "b/multimodal-vicuna-7b-v1.5-audio-damc/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/multimodal-vicuna-7b-v1.5-audio-damc/trainer_state.json"
@@ -0,0 +1,16537 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998183469573115,
+  "global_step": 2752,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2048192771084338e-06,
+      "loss": 2.0439,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4096385542168676e-06,
+      "loss": 1.8574,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.614457831325301e-06,
+      "loss": 1.8833,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.819277108433735e-06,
+      "loss": 1.9565,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.024096385542169e-06,
+      "loss": 1.8735,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.228915662650602e-06,
+      "loss": 1.9663,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.433734939759036e-06,
+      "loss": 1.8081,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.63855421686747e-06,
+      "loss": 1.6899,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0843373493975904e-05,
+      "loss": 1.7153,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2048192771084338e-05,
+      "loss": 1.6328,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3253012048192772e-05,
+      "loss": 1.6201,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4457831325301205e-05,
+      "loss": 1.4883,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.566265060240964e-05,
+      "loss": 1.3945,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6867469879518073e-05,
+      "loss": 1.499,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8072289156626505e-05,
+      "loss": 1.4658,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.927710843373494e-05,
+      "loss": 1.5518,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0481927710843373e-05,
+      "loss": 1.3784,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.168674698795181e-05,
+      "loss": 1.3164,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.289156626506024e-05,
+      "loss": 1.2666,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4096385542168677e-05,
+      "loss": 1.4224,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.530120481927711e-05,
+      "loss": 1.3203,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6506024096385545e-05,
+      "loss": 1.3379,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7710843373493977e-05,
+      "loss": 1.2998,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.891566265060241e-05,
+      "loss": 1.2751,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.012048192771085e-05,
+      "loss": 1.2842,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.132530120481928e-05,
+      "loss": 1.2808,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.253012048192771e-05,
+      "loss": 1.2354,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3734939759036146e-05,
+      "loss": 1.3208,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4939759036144585e-05,
+      "loss": 1.2263,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.614457831325301e-05,
+      "loss": 1.2212,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.734939759036144e-05,
+      "loss": 1.1741,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.855421686746988e-05,
+      "loss": 1.2041,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9759036144578314e-05,
+      "loss": 1.311,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0963855421686746e-05,
+      "loss": 1.2632,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2168674698795186e-05,
+      "loss": 1.3064,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.337349397590362e-05,
+      "loss": 1.2432,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.457831325301205e-05,
+      "loss": 1.1257,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.578313253012048e-05,
+      "loss": 1.1885,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.698795180722892e-05,
+      "loss": 1.187,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8192771084337354e-05,
+      "loss": 1.2073,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9397590361445786e-05,
+      "loss": 1.2358,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.060240963855422e-05,
+      "loss": 1.1531,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.180722891566265e-05,
+      "loss": 1.1899,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.301204819277109e-05,
+      "loss": 1.2092,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4216867469879516e-05,
+      "loss": 1.1489,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5421686746987955e-05,
+      "loss": 1.2148,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6626506024096394e-05,
+      "loss": 1.1929,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.783132530120482e-05,
+      "loss": 1.1384,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.903614457831326e-05,
+      "loss": 1.23,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.02409638554217e-05,
+      "loss": 1.2004,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.144578313253012e-05,
+      "loss": 1.1489,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.265060240963856e-05,
+      "loss": 1.2632,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.385542168674698e-05,
+      "loss": 1.1238,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.506024096385543e-05,
+      "loss": 1.2134,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.626506024096386e-05,
+      "loss": 1.1541,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.746987951807229e-05,
+      "loss": 1.1604,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.867469879518072e-05,
+      "loss": 1.1702,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.987951807228917e-05,
+      "loss": 1.1111,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.108433734939759e-05,
+      "loss": 1.1614,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.228915662650602e-05,
+      "loss": 1.1436,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.349397590361447e-05,
+      "loss": 1.1484,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.469879518072289e-05,
+      "loss": 1.166,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.590361445783133e-05,
+      "loss": 1.1868,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.710843373493976e-05,
+      "loss": 1.1194,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.83132530120482e-05,
+      "loss": 1.1831,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.951807228915663e-05,
+      "loss": 1.158,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.072289156626507e-05,
+      "loss": 1.1299,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.192771084337349e-05,
+      "loss": 1.1057,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.313253012048194e-05,
+      "loss": 1.1917,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.433734939759037e-05,
+      "loss": 1.0925,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.55421686746988e-05,
+      "loss": 1.1851,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.674698795180724e-05,
+      "loss": 1.1191,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.795180722891567e-05,
+      "loss": 1.1292,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.91566265060241e-05,
+      "loss": 1.103,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.036144578313253e-05,
+      "loss": 1.1479,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.156626506024096e-05,
+      "loss": 1.1321,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.27710843373494e-05,
+      "loss": 1.2236,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.397590361445784e-05,
+      "loss": 1.1135,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.518072289156626e-05,
+      "loss": 1.1797,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.638554216867471e-05,
+      "loss": 1.1389,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.759036144578314e-05,
+      "loss": 1.1819,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.879518072289157e-05,
+      "loss": 1.1145,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 1.1489,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999996536281763e-05,
+      "loss": 1.0442,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999986145131847e-05,
+      "loss": 1.126,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999968826564655e-05,
+      "loss": 1.1072,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999944580604174e-05,
+      "loss": 1.1929,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999913407284002e-05,
+      "loss": 1.0999,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999875306647326e-05,
+      "loss": 1.1748,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999830278746937e-05,
+      "loss": 1.1099,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999778323645219e-05,
+      "loss": 1.1721,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999719441414155e-05,
+      "loss": 1.1577,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999653632135324e-05,
+      "loss": 1.1423,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999580895899907e-05,
+      "loss": 1.1143,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999501232808677e-05,
+      "loss": 1.1909,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99941464297201e-05,
+      "loss": 1.1111,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999321126509868e-05,
+      "loss": 1.165,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999220683551824e-05,
+      "loss": 1.1011,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999113314237035e-05,
+      "loss": 1.1221,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998999018714263e-05,
+      "loss": 1.1399,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998877797141863e-05,
+      "loss": 1.0342,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998749649687785e-05,
+      "loss": 1.116,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998614576529575e-05,
+      "loss": 1.1235,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998472577854376e-05,
+      "loss": 1.1587,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998323653858926e-05,
+      "loss": 1.0251,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998167804749557e-05,
+      "loss": 1.0447,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998005030742194e-05,
+      "loss": 1.0869,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997835332062362e-05,
+      "loss": 1.1162,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997658708945174e-05,
+      "loss": 1.0859,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997475161635337e-05,
+      "loss": 1.0815,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997284690387159e-05,
+      "loss": 1.1072,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99708729546453e-05,
+      "loss": 1.1169,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996882977140941e-05,
+      "loss": 1.1006,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996671735699473e-05,
+      "loss": 1.0823,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996453571432795e-05,
+      "loss": 1.085,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996228484643175e-05,
+      "loss": 1.1545,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995996475642466e-05,
+      "loss": 1.1104,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995757544752113e-05,
+      "loss": 1.1404,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995511692303152e-05,
+      "loss": 1.135,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995258918636209e-05,
+      "loss": 1.0867,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994999224101498e-05,
+      "loss": 1.1687,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994732609058823e-05,
+      "loss": 1.082,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994459073877576e-05,
+      "loss": 1.1306,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994178618936736e-05,
+      "loss": 1.1135,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993891244624869e-05,
+      "loss": 1.1421,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99359695134013e-05,
+      "loss": 1.0996,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993295739490258e-05,
+      "loss": 1.1143,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992987609492578e-05,
+      "loss": 1.1333,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992672561774e-05,
+      "loss": 1.1514,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992350596771019e-05,
+      "loss": 1.1089,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992021714929714e-05,
+      "loss": 1.0701,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991685916705747e-05,
+      "loss": 1.1533,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991343202564359e-05,
+      "loss": 1.0,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990993572980378e-05,
+      "loss": 1.186,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990637028438212e-05,
+      "loss": 1.1272,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990273569431849e-05,
+      "loss": 1.1333,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989903196464857e-05,
+      "loss": 1.1252,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98952591005038e-05,
+      "loss": 1.1121,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98914171071115e-05,
+      "loss": 1.0874,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988750598979463e-05,
+      "loss": 1.1528,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988352575397203e-05,
+      "loss": 1.0745,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987947640515827e-05,
+      "loss": 1.0791,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987535794896366e-05,
+      "loss": 1.0769,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987117039109427e-05,
+      "loss": 1.1165,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98669137373519e-05,
+      "loss": 1.1006,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98625879936341e-05,
+      "loss": 1.093,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985819316593415e-05,
+      "loss": 1.1252,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9853729260341e-05,
+      "loss": 1.1248,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984919628303935e-05,
+      "loss": 1.0352,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984459424030959e-05,
+      "loss": 1.0801,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983992313852774e-05,
+      "loss": 1.1211,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983518298416563e-05,
+      "loss": 1.0793,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983037378379063e-05,
+      "loss": 1.2002,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982549554406584e-05,
+      "loss": 1.1191,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982054827175e-05,
+      "loss": 1.1104,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981553197369751e-05,
+      "loss": 1.0959,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981044665685834e-05,
+      "loss": 1.0962,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980529232827817e-05,
+      "loss": 1.0791,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980006899509826e-05,
+      "loss": 1.072,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979477666455546e-05,
+      "loss": 1.0564,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978941534398223e-05,
+      "loss": 1.0559,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97839850408066e-05,
+      "loss": 1.0603,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97784857625522e-05,
+      "loss": 1.0962,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977291751683821e-05,
+      "loss": 1.1277,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976728031137936e-05,
+      "loss": 1.0925,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97615741539859e-05,
+      "loss": 1.0247,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97557990525637e-05,
+      "loss": 1.0283,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974995501511404e-05,
+      "loss": 1.0369,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974404204973376e-05,
+      "loss": 1.0713,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973806016461522e-05,
+      "loss": 1.1394,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973200936804624e-05,
+      "loss": 1.1436,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972588966841012e-05,
+      "loss": 1.0752,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971970107418563e-05,
+      "loss": 1.02,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971344359394696e-05,
+      "loss": 1.0967,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970711723636382e-05,
+      "loss": 1.0249,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970072201020126e-05,
+      "loss": 1.0342,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969425792431981e-05,
+      "loss": 1.075,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968772498767536e-05,
+      "loss": 1.1257,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968112320931921e-05,
+      "loss": 1.0908,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967445259839804e-05,
+      "loss": 1.0503,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96677131641539e-05,
+      "loss": 1.0605,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966090491592421e-05,
+      "loss": 1.0764,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965402786314169e-05,
+      "loss": 1.0784,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96470820153344e-05,
+      "loss": 1.0942,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964006738212575e-05,
+      "loss": 1.1055,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963298397323442e-05,
+      "loss": 1.1006,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962583179847436e-05,
+      "loss": 1.106,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961861086775484e-05,
+      "loss": 1.0388,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961132119108036e-05,
+      "loss": 1.0737,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960396277855067e-05,
+      "loss": 1.1201,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959653564036076e-05,
+      "loss": 1.0776,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958903978680086e-05,
+      "loss": 1.1497,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958147522825633e-05,
+      "loss": 1.0674,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957384197520782e-05,
+      "loss": 1.1106,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956614003823107e-05,
+      "loss": 1.0989,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955836942799703e-05,
+      "loss": 1.0254,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955053015527178e-05,
+      "loss": 1.0979,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954262223091653e-05,
+      "loss": 1.0825,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953464566588762e-05,
+      "loss": 1.1482,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952660047123648e-05,
+      "loss": 1.0925,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95184866581096e-05,
+      "loss": 1.1594,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951030423774859e-05,
+      "loss": 1.0852,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950205322149006e-05,
+      "loss": 1.0901,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949373362076573e-05,
+      "loss": 1.1458,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948534544710228e-05,
+      "loss": 1.1265,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947688871212141e-05,
+      "loss": 1.0703,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946836342753983e-05,
+      "loss": 1.1665,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945976960516921e-05,
+      "loss": 1.1299,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945110725691618e-05,
+      "loss": 1.1047,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944237639478231e-05,
+      "loss": 1.0332,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943357703086411e-05,
+      "loss": 1.0916,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942470917735298e-05,
+      "loss": 1.1023,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941577284653523e-05,
+      "loss": 1.0852,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940676805079201e-05,
+      "loss": 1.0498,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939769480259938e-05,
+      "loss": 1.0789,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938855311452817e-05,
+      "loss": 1.0952,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937934299924409e-05,
+      "loss": 1.1213,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937006446950767e-05,
+      "loss": 1.03,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936071753817415e-05,
+      "loss": 1.0972,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93513022181936e-05,
+      "loss": 1.1741,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934181852261084e-05,
+      "loss": 1.0803,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93322664645654e-05,
+      "loss": 1.134,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932264605729152e-05,
+      "loss": 1.0581,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931295731411818e-05,
+      "loss": 1.072,
+      "step": 224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930320024846899e-05,
+      "loss": 1.083,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929337487386224e-05,
+      "loss": 1.1118,
+      "step": 226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928348120391088e-05,
+      "loss": 1.0203,
+      "step": 227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927351925232245e-05,
+      "loss": 1.1201,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92634890328991e-05,
+      "loss": 1.0508,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925339055953758e-05,
+      "loss": 1.1008,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924322384622922e-05,
+      "loss": 1.1079,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923298890705984e-05,
+      "loss": 1.053,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922268575620981e-05,
+      "loss": 1.093,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921231440795404e-05,
+      "loss": 1.1033,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920187487666191e-05,
+      "loss": 1.0828,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919136717679722e-05,
+      "loss": 1.0469,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91807913229183e-05,
+      "loss": 1.074,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917014732967782e-05,
+      "loss": 1.1289,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915943521182291e-05,
+      "loss": 1.1013,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91486549841951e-05,
+      "loss": 1.1125,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913780666173022e-05,
+      "loss": 1.0635,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91268902594585e-05,
+      "loss": 1.04,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91159057925045e-05,
+      "loss": 1.0701,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910485327608701e-05,
+      "loss": 1.1331,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909373272551918e-05,
+      "loss": 1.0239,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908254415620839e-05,
+      "loss": 1.1399,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907128758365627e-05,
+      "loss": 1.0693,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905996302345864e-05,
+      "loss": 1.0432,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904857049130554e-05,
+      "loss": 1.0986,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903711000298116e-05,
+      "loss": 1.0845,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902558157436391e-05,
+      "loss": 1.1299,
+      "step": 251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901398522142625e-05,
+      "loss": 1.0095,
+      "step": 252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900232096023477e-05,
+      "loss": 1.1096,
+      "step": 253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899058880695018e-05,
+      "loss": 1.1094,
+      "step": 254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897878877782721e-05,
+      "loss": 1.0498,
+      "step": 255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896692088921466e-05,
+      "loss": 1.1226,
+      "step": 256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895498515755532e-05,
+      "loss": 1.1116,
+      "step": 257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894298159938605e-05,
+      "loss": 1.061,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893091023133756e-05,
+      "loss": 1.1021,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891877107013461e-05,
+      "loss": 1.1067,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890656413259585e-05,
+      "loss": 1.043,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889428943563382e-05,
+      "loss": 1.1252,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888194699625499e-05,
+      "loss": 1.0703,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886953683155964e-05,
+      "loss": 1.0559,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885705895874188e-05,
+      "loss": 1.0503,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884451339508966e-05,
+      "loss": 1.1304,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88319001579847e-05,
+      "loss": 1.0715,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881921926490245e-05,
+      "loss": 1.135,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880647073341219e-05,
+      "loss": 1.0471,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879365458117678e-05,
+      "loss": 1.0278,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878077082595287e-05,
+      "loss": 1.0667,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876781948559073e-05,
+      "loss": 1.0735,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87548005780343e-05,
+      "loss": 1.0583,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874171412132107e-05,
+      "loss": 1.1233,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87285601335822e-05,
+      "loss": 1.1401,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871533863304235e-05,
+      "loss": 1.0071,
+      "step": 276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870204963801973e-05,
+      "loss": 1.1274,
+      "step": 277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868869316692611e-05,
+      "loss": 1.0664,
+      "step": 278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867526923826668e-05,
+      "loss": 1.0854,
+      "step": 279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866177787064014e-05,
+      "loss": 1.0862,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86482190827386e-05,
+      "loss": 1.1191,
+      "step": 281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863459289334758e-05,
+      "loss": 1.1013,
+      "step": 282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862089932134602e-05,
+      "loss": 1.041,
+      "step": 283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860713838570616e-05,
+      "loss": 1.0952,
+      "step": 284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859331010549362e-05,
+      "loss": 1.1082,
+      "step": 285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85794144998673e-05,
+      "loss": 1.0457,
+      "step": 286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856545158807938e-05,
+      "loss": 1.1106,
+      "step": 287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85514213894753e-05,
+      "loss": 1.1252,
+      "step": 288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853732392349375e-05,
+      "loss": 1.0732,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852315920966653e-05,
+      "loss": 1.0574,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850892726761874e-05,
+      "loss": 1.0784,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849462811706849e-05,
+      "loss": 1.1177,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848026177782712e-05,
+      "loss": 1.0483,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846582826979899e-05,
+      "loss": 1.0371,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845132761298154e-05,
+      "loss": 1.0452,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843675982746525e-05,
+      "loss": 1.0186,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842212493343361e-05,
+      "loss": 1.0737,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840742295116305e-05,
+      "loss": 1.0164,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839265390102301e-05,
+      "loss": 1.0081,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837781780347584e-05,
+      "loss": 1.0627,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83629146790767e-05,
+      "loss": 1.0249,
+      "step": 301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834794454847374e-05,
+      "loss": 1.0664,
+      "step": 302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833290743240785e-05,
+      "loss": 1.0437,
+      "step": 303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831780335171279e-05,
+      "loss": 1.0647,
+      "step": 304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830263232731504e-05,
+      "loss": 1.0891,
+      "step": 305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828739438023389e-05,
+      "loss": 1.0779,
+      "step": 306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827208953158131e-05,
+      "loss": 1.063,
+      "step": 307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825671780256197e-05,
+      "loss": 1.0686,
+      "step": 308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824127921447321e-05,
+      "loss": 1.0605,
+      "step": 309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822577378870502e-05,
+      "loss": 1.0312,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82102015467399e-05,
+      "loss": 1.0,
+      "step": 311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819456251015308e-05,
+      "loss": 1.0603,
+      "step": 312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817885670061219e-05,
+      "loss": 1.0718,
+      "step": 313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816308413987747e-05,
+      "loss": 1.0493,
+      "step": 314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814724484980156e-05,
+      "loss": 1.0266,
+      "step": 315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813133885232962e-05,
+      "loss": 1.083,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811536616949921e-05,
+      "loss": 1.1155,
+      "step": 317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809932682344026e-05,
+      "loss": 1.0955,
+      "step": 318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80832208363751e-05,
+      "loss": 1.0215,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806704823061836e-05,
+      "loss": 1.0757,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805080902857699e-05,
+      "loss": 1.1023,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803450325275017e-05,
+      "loss": 1.0671,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801813092572937e-05,
+      "loss": 1.0249,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800169207019825e-05,
+      "loss": 1.0413,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798518670893262e-05,
+      "loss": 1.0491,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796861486480044e-05,
+      "loss": 1.063,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795197656076182e-05,
+      "loss": 1.1074,
+      "step": 327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793527181986888e-05,
+      "loss": 1.0339,
+      "step": 328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791850066526584e-05,
+      "loss": 1.083,
+      "step": 329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790166312018895e-05,
+      "loss": 1.1152,
+      "step": 330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788475920796638e-05,
+      "loss": 1.0818,
+      "step": 331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786778895201829e-05,
+      "loss": 1.0984,
+      "step": 332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785075237585677e-05,
+      "loss": 1.0598,
+      "step": 333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783364950308576e-05,
+      "loss": 1.0923,
+      "step": 334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78164803574011e-05,
+      "loss": 1.0464,
+      "step": 335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77992449625904e-05,
+      "loss": 1.1357,
+      "step": 336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778194334253308e-05,
+      "loss": 1.0803,
+      "step": 337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776457552120033e-05,
+      "loss": 1.0764,
+      "step": 338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774714152265504e-05,
+      "loss": 1.0496,
+      "step": 339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772964137105178e-05,
+      "loss": 1.0706,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771207509063682e-05,
+      "loss": 1.0923,
+      "step": 341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769444270574797e-05,
+      "loss": 1.0872,
+      "step": 342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767674424081472e-05,
+      "loss": 1.0613,
+      "step": 343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765897972035806e-05,
+      "loss": 1.0894,
+      "step": 344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764114916899048e-05,
+      "loss": 1.0232,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762325261141602e-05,
+      "loss": 1.0535,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76052900724301e-05,
+      "loss": 1.0425,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75872615769196e-05,
+      "loss": 1.0938,
+      "step": 348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75691671498628e-05,
+      "loss": 1.1011,
+      "step": 349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755100681632925e-05,
+      "loss": 1.0684,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753278060147989e-05,
+      "loss": 1.1584,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75144885305669e-05,
+      "loss": 1.0645,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749613062893373e-05,
+      "loss": 1.1558,
+      "step": 353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7477706922015e-05,
+      "loss": 1.051,
+      "step": 354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745921743533653e-05,
+      "loss": 1.1018,
+      "step": 355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744066219451526e-05,
+      "loss": 1.0439,
+      "step": 356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742204122525924e-05,
+      "loss": 1.0212,
+      "step": 357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740335455336761e-05,
+      "loss": 1.0818,
+      "step": 358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738460220473049e-05,
+      "loss": 1.0549,
+      "step": 359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736578420532904e-05,
+      "loss": 1.0603,
+      "step": 360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734690058123534e-05,
+      "loss": 1.0488,
+      "step": 361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732795135861244e-05,
+      "loss": 1.1084,
+      "step": 362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730893656371421e-05,
+      "loss": 1.0667,
+      "step": 363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728985622288542e-05,
+      "loss": 1.0667,
+      "step": 364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727071036256166e-05,
+      "loss": 1.0693,
+      "step": 365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725149900926924e-05,
+      "loss": 1.1062,
+      "step": 366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72322221896253e-05,
+      "loss": 1.0371,
+      "step": 367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721287993033757e-05,
+      "loss": 1.0159,
+      "step": 368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719347225820452e-05,
+      "loss": 1.0449,
+      "step": 369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717399920011528e-05,
+      "loss": 1.1187,
+      "step": 370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715446078304946e-05,
+      "loss": 1.0559,
+      "step": 371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713485703407731e-05,
+      "loss": 1.0815,
+      "step": 372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71151879803596e-05,
+      "loss": 1.0186,
+      "step": 373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709545364914754e-05,
+      "loss": 1.0593,
+      "step": 374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707565406778278e-05,
+      "loss": 1.0659,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705578926369739e-05,
+      "loss": 1.0017,
+      "step": 376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703585926441383e-05,
+      "loss": 1.0654,
+      "step": 377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701586409754484e-05,
+      "loss": 1.0115,
+      "step": 378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699580379079348e-05,
+      "loss": 1.1221,
+      "step": 379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697567837195302e-05,
+      "loss": 1.0908,
+      "step": 380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695548786890701e-05,
+      "loss": 1.0332,
+      "step": 381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693523230962913e-05,
+      "loss": 1.0664,
+      "step": 382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691491172218318e-05,
+      "loss": 1.0137,
+      "step": 383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689452613472308e-05,
+      "loss": 1.0342,
+      "step": 384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687407557549281e-05,
+      "loss": 1.0679,
+      "step": 385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685356007282638e-05,
+      "loss": 1.082,
+      "step": 386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683297965514773e-05,
+      "loss": 1.0579,
+      "step": 387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681233435097077e-05,
+      "loss": 1.1211,
+      "step": 388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679162418889931e-05,
+      "loss": 1.1326,
+      "step": 389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677084919762702e-05,
+      "loss": 1.0344,
+      "step": 390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675000940593738e-05,
+      "loss": 1.0627,
+      "step": 391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672910484270367e-05,
+      "loss": 1.1047,
+      "step": 392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670813553688888e-05,
+      "loss": 1.1042,
+      "step": 393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668710151754572e-05,
+      "loss": 1.1384,
+      "step": 394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666600281381658e-05,
+      "loss": 1.1394,
+      "step": 395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664483945493341e-05,
+      "loss": 1.0251,
+      "step": 396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662361147021779e-05,
+      "loss": 1.0518,
+      "step": 397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660231888908084e-05,
+      "loss": 1.0293,
+      "step": 398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658096174102314e-05,
+      "loss": 1.0151,
+      "step": 399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655954005563474e-05,
+      "loss": 1.0295,
+      "step": 400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653805386259513e-05,
+      "loss": 1.0686,
+      "step": 401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651650319167317e-05,
+      "loss": 1.0576,
+      "step": 402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649488807272703e-05,
+      "loss": 1.1189,
+      "step": 403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647320853570414e-05,
+      "loss": 1.095,
+      "step": 404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645146461064129e-05,
+      "loss": 1.0383,
+      "step": 405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642965632766436e-05,
+      "loss": 0.9788,
+      "step": 406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640778371698848e-05,
+      "loss": 1.0615,
+      "step": 407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638584680891787e-05,
+      "loss": 1.0837,
+      "step": 408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636384563384584e-05,
+      "loss": 1.0381,
+      "step": 409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634178022225473e-05,
+      "loss": 1.0679,
+      "step": 410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631965060471588e-05,
+      "loss": 1.0408,
+      "step": 411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62974568118896e-05,
+      "loss": 1.0759,
+      "step": 412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62751988745251e-05,
+      "loss": 1.0754,
+      "step": 413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62528768234605e-05,
+      "loss": 0.9739,
+      "step": 414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623049068962269e-05,
+      "loss": 0.978,
+      "step": 415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620804050402739e-05,
+      "loss": 0.9998,
+      "step": 416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618552629777903e-05,
+      "loss": 0.9805,
+      "step": 417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616294810207076e-05,
+      "loss": 1.0466,
+      "step": 418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61403059481844e-05,
+      "loss": 1.1343,
+      "step": 419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611759986749036e-05,
+      "loss": 1.0608,
+      "step": 420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60948298914476e-05,
+      "loss": 1.0339,
+      "step": 421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607199605160367e-05,
+      "loss": 1.1003,
+      "step": 422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604909837959455e-05,
+      "loss": 1.0413,
+      "step": 423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602613690714467e-05,
+      "loss": 1.0977,
+      "step": 424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600311166606686e-05,
+      "loss": 1.022,
+      "step": 425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59800226882623e-05,
+      "loss": 1.072,
+      "step": 426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595687000572049e-05,
+      "loss": 1.0332,
+      "step": 427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593365365051915e-05,
+      "loss": 1.0732,
+      "step": 428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591037365482425e-05,
+      "loss": 1.0752,
+      "step": 429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588703005088994e-05,
+      "loss": 0.9832,
+      "step": 430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58636228710585e-05,
+      "loss": 1.0415,
+      "step": 431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584015214776026e-05,
+      "loss": 1.0881,
+      "step": 432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58166179135136e-05,
+      "loss": 1.0334,
+      "step": 433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579302020092491e-05,
+      "loss": 1.0569,
+      "step": 434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576935904268854e-05,
+      "loss": 1.0173,
+      "step": 435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57456344715867e-05,
+      "loss": 1.0537,
+      "step": 436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57218465204895e-05,
+      "loss": 1.0737,
+      "step": 437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569799522235483e-05,
+      "loss": 1.0515,
+      "step": 438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567408061022838e-05,
+      "loss": 1.0747,
+      "step": 439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565010271724352e-05,
+      "loss": 1.0146,
+      "step": 440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562606157662133e-05,
+      "loss": 0.9951,
+      "step": 441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56019572216705e-05,
+      "loss": 1.0654,
+      "step": 442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557778968578728e-05,
+      "loss": 1.0466,
+      "step": 443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555355900245553e-05,
+      "loss": 1.1313,
+      "step": 444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552926520524653e-05,
+      "loss": 1.1235,
+      "step": 445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550490832781905e-05,
+      "loss": 1.0073,
+      "step": 446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548048840391918e-05,
+      "loss": 1.0588,
+      "step": 447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545600546738047e-05,
+      "loss": 0.9709,
+      "step": 448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54314595521237e-05,
+      "loss": 1.1116,
+      "step": 449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540685069215692e-05,
+      "loss": 1.1128,
+      "step": 450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53821789215754e-05,
+      "loss": 1.0776,
+      "step": 451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535744427456155e-05,
+      "loss": 1.0176,
+      "step": 452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533264678538492e-05,
+      "loss": 1.0471,
+      "step": 453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530778648840212e-05,
+      "loss": 1.0811,
+      "step": 454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528286341805675e-05,
+      "loss": 1.0371,
+      "step": 455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525787760887945e-05,
+      "loss": 1.0349,
+      "step": 456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523282909548773e-05,
+      "loss": 1.1292,
+      "step": 457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520771791258593e-05,
+      "loss": 1.1243,
+      "step": 458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518254409496536e-05,
+      "loss": 0.9939,
+      "step": 459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515730767750397e-05,
+      "loss": 1.1138,
+      "step": 460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51320086951665e-05,
+      "loss": 1.0386,
+      "step": 461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510664718300439e-05,
+      "loss": 1.0876,
+      "step": 462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508122317615569e-05,
+      "loss": 1.0442,
+      "step": 463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505573670984503e-05,
+      "loss": 1.0364,
+      "step": 464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503018781938357e-05,
+      "loss": 1.03,
+      "step": 465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500457654016901e-05,
+      "loss": 1.073,
+      "step": 466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497890290768541e-05,
+      "loss": 1.0715,
+      "step": 467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49531669575033e-05,
+      "loss": 1.0217,
+      "step": 468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492736872527947e-05,
+      "loss": 1.0203,
+      "step": 469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49015082467571e-05,
+      "loss": 1.0583,
+      "step": 470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48755855577655e-05,
+      "loss": 1.0247,
+      "step": 471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484960069422025e-05,
+      "loss": 1.0676,
+      "step": 472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482355369212306e-05,
+      "loss": 1.1235,
+      "step": 473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47974445875617e-05,
+      "loss": 1.0994,
+      "step": 474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477127341671e-05,
+      "loss": 1.0654,
+      "step": 475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474504021582781e-05,
+      "loss": 1.0232,
+      "step": 476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471874502126087e-05,
+      "loss": 1.054,
+      "step": 477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469238786944085e-05,
+      "loss": 1.0361,
+      "step": 478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466596879688525e-05,
+      "loss": 1.0256,
+      "step": 479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463948784019736e-05,
+      "loss": 1.0161,
+      "step": 480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46129450360662e-05,
+      "loss": 1.0894,
+      "step": 481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458634042126651e-05,
+      "loss": 1.1125,
+      "step": 482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455967403265861e-05,
+      "loss": 1.0605,
+      "step": 483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453294590718847e-05,
+      "loss": 1.0081,
+      "step": 484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450615608188756e-05,
+      "loss": 1.0544,
+      "step": 485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447930459387284e-05,
+      "loss": 1.0173,
+      "step": 486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445239148034673e-05,
+      "loss": 1.0129,
+      "step": 487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442541677859695e-05,
+      "loss": 1.0452,
+      "step": 488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439838052599667e-05,
+      "loss": 1.0583,
+      "step": 489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437128276000425e-05,
+      "loss": 0.9954,
+      "step": 490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434412351816328e-05,
+      "loss": 1.0107,
+      "step": 491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431690283810257e-05,
+      "loss": 1.0154,
+      "step": 492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428962075753602e-05,
+      "loss": 1.1128,
+      "step": 493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42622773142626e-05,
+      "loss": 1.0686,
+      "step": 494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423487254616631e-05,
+      "loss": 1.0708,
+      "step": 495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420740649121611e-05,
+      "loss": 0.967,
+      "step": 496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417987918746585e-05,
+      "loss": 1.0774,
+      "step": 497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415229067305429e-05,
+      "loss": 1.0417,
+      "step": 498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412464098620495e-05,
+      "loss": 1.1106,
+      "step": 499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409693016522612e-05,
+      "loss": 0.979,
+      "step": 500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40691582485108e-05,
+      "loss": 1.0039,
+      "step": 501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404132527453661e-05,
+      "loss": 1.0376,
+      "step": 502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40134312818658e-05,
+      "loss": 1.0552,
+      "step": 503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398547630914511e-05,
+      "loss": 1.0835,
+      "step": 504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395746039510585e-05,
+      "loss": 1.03,
+      "step": 505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392938357856366e-05,
+      "loss": 0.9919,
+      "step": 506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390124589841866e-05,
+      "loss": 0.9875,
+      "step": 507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387304739365523e-05,
+      "loss": 1.0154,
+      "step": 508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384478810334203e-05,
+      "loss": 1.0186,
+      "step": 509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381646806663196e-05,
+      "loss": 0.9712,
+      "step": 510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378808732276205e-05,
+      "loss": 1.0569,
+      "step": 511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37596459110535e-05,
+      "loss": 1.0049,
+      "step": 512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373114387091148e-05,
+      "loss": 1.0203,
+      "step": 513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370258124182525e-05,
+      "loss": 1.0654,
+      "step": 514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367395806336792e-05,
+      "loss": 1.1021,
+      "step": 515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364527437519658e-05,
+      "loss": 1.0657,
+      "step": 516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36165302170521e-05,
+      "loss": 1.0007,
+      "step": 517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358772562875914e-05,
+      "loss": 1.0608,
+      "step": 518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35588606502261e-05,
+      "loss": 1.0151,
+      "step": 519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352993532144504e-05,
+      "loss": 1.0239,
+      "step": 520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350094968249163e-05,
+      "loss": 1.0312,
+      "step": 521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347190377352512e-05,
+      "loss": 1.0735,
+      "step": 522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344279763478822e-05,
+      "loss": 1.0571,
+      "step": 523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341363130660714e-05,
+      "loss": 1.0459,
+      "step": 524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338440482939146e-05,
+      "loss": 1.1006,
+      "step": 525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335511824363407e-05,
+      "loss": 1.0735,
+      "step": 526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332577158991118e-05,
+      "loss": 1.0667,
+      "step": 527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32963649088822e-05,
+      "loss": 1.0579,
+      "step": 528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326689824128971e-05,
+      "loss": 0.9692,
+      "step": 529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32373716279594e-05,
+      "loss": 1.0247,
+      "step": 530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320778510980004e-05,
+      "loss": 1.0474,
+      "step": 531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317813872780337e-05,
+      "loss": 1.054,
+      "step": 532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314843252304404e-05,
+      "loss": 1.0195,
+      "step": 533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311866653667965e-05,
+      "loss": 1.0769,
+      "step": 534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30888408099506e-05,
+      "loss": 1.0457,
+      "step": 535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305895538418004e-05,
+      "loss": 1.1169,
+      "step": 536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302901030077384e-05,
+      "loss": 1.0454,
+      "step": 537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299900560122057e-05,
+      "loss": 1.0332,
+      "step": 538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296894132709133e-05,
+      "loss": 1.0303,
+      "step": 539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293881752003979e-05,
+      "loss": 1.041,
+      "step": 540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290863422180211e-05,
+      "loss": 1.0891,
+      "step": 541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287839147419686e-05,
+      "loss": 1.0347,
+      "step": 542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2848089319125e-05,
+      "loss": 1.0776,
+      "step": 543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281772779856976e-05,
+      "loss": 1.0183,
+      "step": 544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278730695459665e-05,
+      "loss": 1.0684,
+      "step": 545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275682682935336e-05,
+      "loss": 1.0874,
+      "step": 546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27262874650697e-05,
+      "loss": 1.0244,
+      "step": 547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269568890405763e-05,
+      "loss": 1.0681,
+      "step": 548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266503118871099e-05,
+      "loss": 1.1147,
+      "step": 549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263431436150572e-05,
+      "loss": 1.0459,
+      "step": 550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260353846499954e-05,
+      "loss": 0.9856,
+      "step": 551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257270354183211e-05,
+      "loss": 1.0991,
+      "step": 552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254180963472479e-05,
+      "loss": 1.0271,
+      "step": 553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251085678648072e-05,
+      "loss": 0.9893,
+      "step": 554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247984503998466e-05,
+      "loss": 1.0393,
+      "step": 555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244877443820299e-05,
+      "loss": 1.0876,
+      "step": 556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241764502418365e-05,
+      "loss": 1.0039,
+      "step": 557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238645684105605e-05,
+      "loss": 1.0569,
+      "step": 558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2355209932031e-05,
+      "loss": 1.0283,
+      "step": 559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23239043404007e-05,
+      "loss": 1.0383,
+      "step": 560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229254010953867e-05,
+      "loss": 0.968,
+      "step": 561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226111728289964e-05,
+      "loss": 0.9771,
+      "step": 562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222963590401953e-05,
+      "loss": 1.0503,
+      "step": 563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219809601651539e-05,
+      "loss": 1.0562,
+      "step": 564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216649766408535e-05,
+      "loss": 1.0278,
+      "step": 565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213484089050852e-05,
+      "loss": 0.9998,
+      "step": 566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210312573964494e-05,
+      "loss": 1.0327,
+      "step": 567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207135225543557e-05,
+      "loss": 1.0278,
+      "step": 568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203952048190217e-05,
+      "loss": 1.1001,
+      "step": 569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200763046314724e-05,
+      "loss": 1.0081,
+      "step": 570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197568224335401e-05,
+      "loss": 1.1584,
+      "step": 571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194367586678632e-05,
+      "loss": 1.0781,
+      "step": 572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191161137778861e-05,
+      "loss": 1.0247,
+      "step": 573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187948882078582e-05,
+      "loss": 1.0366,
+      "step": 574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184730824028333e-05,
+      "loss": 1.0364,
+      "step": 575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181506968086697e-05,
+      "loss": 1.0571,
+      "step": 576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178277318720279e-05,
+      "loss": 1.009,
+      "step": 577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17504188040372e-05,
+      "loss": 0.9922,
+      "step": 578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171800657619683e-05,
+      "loss": 1.0408,
+      "step": 579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168553654858834e-05,
+      "loss": 0.9854,
+      "step": 580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165300876619857e-05,
+      "loss": 1.0403,
+      "step": 581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162042327409436e-05,
+      "loss": 0.9824,
+      "step": 582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158778011742249e-05,
+      "loss": 1.0596,
+      "step": 583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155507934140961e-05,
+      "loss": 1.0417,
+      "step": 584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152232099136227e-05,
+      "loss": 1.0781,
+      "step": 585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148950511266674e-05,
+      "loss": 1.0117,
+      "step": 586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145663175078898e-05,
+      "loss": 1.0173,
+      "step": 587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142370095127465e-05,
+      "loss": 1.0376,
+      "step": 588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139071275974893e-05,
+      "loss": 1.0725,
+      "step": 589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135766722191654e-05,
+      "loss": 1.0691,
+      "step": 590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132456438356165e-05,
+      "loss": 1.0261,
+      "step": 591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.129140429054785e-05,
+      "loss": 1.0557,
+      "step": 592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125818698881798e-05,
+      "loss": 1.0044,
+      "step": 593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122491252439424e-05,
+      "loss": 0.9629,
+      "step": 594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119158094337795e-05,
+      "loss": 0.9795,
+      "step": 595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11581922919496e-05,
+      "loss": 1.0796,
+      "step": 596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112474661636872e-05,
+      "loss": 1.0405,
+      "step": 597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10912439629739e-05,
+      "loss": 0.9966,
+      "step": 598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105768437818263e-05,
+      "loss": 1.0942,
+      "step": 599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102406790849128e-05,
+      "loss": 1.0566,
+      "step": 600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099039460047505e-05,
+      "loss": 1.0295,
+      "step": 601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095666450078789e-05,
+      "loss": 1.0095,
+      "step": 602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092287765616241e-05,
+      "loss": 1.0991,
+      "step": 603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088903411340985e-05,
+      "loss": 1.1265,
+      "step": 604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085513391942002e-05,
+      "loss": 0.9988,
+      "step": 605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08211771211612e-05,
+      "loss": 1.0569,
+      "step": 606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07871637656801e-05,
+      "loss": 0.9946,
+      "step": 607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075309390010181e-05,
+      "loss": 1.0068,
+      "step": 608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071896757162969e-05,
+      "loss": 1.054,
+      "step": 609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068478482754532e-05,
+      "loss": 1.0784,
+      "step": 610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065054571520846e-05,
+      "loss": 1.0278,
+      "step": 611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061625028205698e-05,
+      "loss": 1.0989,
+      "step": 612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058189857560675e-05,
+      "loss": 1.0518,
+      "step": 613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054749064345165e-05,
+      "loss": 1.031,
+      "step": 614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051302653326339e-05,
+      "loss": 0.9648,
+      "step": 615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04785062927916e-05,
+      "loss": 1.0276,
+      "step": 616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044392996986361e-05,
+      "loss": 1.0837,
+      "step": 617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040929761238448e-05,
+      "loss": 1.0159,
+      "step": 618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03746092683369e-05,
+      "loss": 1.0454,
+      "step": 619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033986498578113e-05,
+      "loss": 1.0183,
+      "step": 620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030506481285495e-05,
+      "loss": 1.1167,
+      "step": 621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027020879777352e-05,
+      "loss": 1.011,
+      "step": 622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023529698882945e-05,
+      "loss": 1.0049,
+      "step": 623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020032943439258e-05,
+      "loss": 1.0232,
+      "step": 624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016530618291001e-05,
+      "loss": 0.9844,
+      "step": 625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013022728290603e-05,
+      "loss": 1.04,
+      "step": 626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0095092782982e-05,
+      "loss": 1.033,
+      "step": 627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005990273181631e-05,
+      "loss": 1.0479,
+      "step": 628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002465717816435e-05,
+      "loss": 1.0098,
+      "step": 629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998935617085837e-05,
+      "loss": 1.0364,
+      "step": 630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995399975880749e-05,
+      "loss": 1.0989,
+      "step": 631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991858799099755e-05,
+      "loss": 1.0181,
+      "step": 632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988312091649109e-05,
+      "loss": 1.0913,
+      "step": 633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984759858442733e-05,
+      "loss": 1.0076,
+      "step": 634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9812021044022e-05,
+      "loss": 1.1296,
+      "step": 635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97763883445673e-05,
+      "loss": 1.0493,
+      "step": 636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97407005354319e-05,
+      "loss": 1.0076,
+      "step": 637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.970495766606081e-05,
+      "loss": 1.01,
+      "step": 638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966915978597532e-05,
+      "loss": 1.0156,
+      "step": 639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963330694477296e-05,
+      "loss": 1.031,
+      "step": 640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959739919212734e-05,
+      "loss": 1.0986,
+      "step": 641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956143657778822e-05,
+      "loss": 1.0066,
+      "step": 642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952541915158136e-05,
+      "loss": 1.04,
+      "step": 643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948934696340843e-05,
+      "loss": 1.1367,
+      "step": 644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945322006324696e-05,
+      "loss": 0.9707,
+      "step": 645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941703850115038e-05,
+      "loss": 1.0085,
+      "step": 646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.938080232724773e-05,
+      "loss": 1.1328,
+      "step": 647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934451159174377e-05,
+      "loss": 1.0371,
+      "step": 648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930816634491887e-05,
+      "loss": 1.0127,
+      "step": 649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927176663712892e-05,
+      "loss": 0.9888,
+      "step": 650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923531251880524e-05,
+      "loss": 1.0261,
+      "step": 651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919880404045452e-05,
+      "loss": 1.0498,
+      "step": 652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916224125265883e-05,
+      "loss": 1.012,
+      "step": 653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912562420607544e-05,
+      "loss": 1.0659,
+      "step": 654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90889529514368e-05,
+      "loss": 1.0271,
+      "step": 655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905222753955045e-05,
+      "loss": 1.0576,
+      "step": 656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901544802129902e-05,
+      "loss": 1.0427,
+      "step": 657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897861444764003e-05,
+      "loss": 1.083,
+      "step": 658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894172686960593e-05,
+      "loss": 1.0808,
+      "step": 659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890478533830402e-05,
+      "loss": 1.0139,
+      "step": 660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886778990491631e-05,
+      "loss": 1.0352,
+      "step": 661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883074062069947e-05,
+      "loss": 1.0317,
+      "step": 662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879363753698486e-05,
+      "loss": 1.0427,
+      "step": 663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875648070517832e-05,
+      "loss": 1.0806,
+      "step": 664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871927017676014e-05,
+      "loss": 1.0173,
+      "step": 665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868200600328504e-05,
+      "loss": 1.0354,
+      "step": 666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86446882363821e-05,
+      "loss": 1.0796,
+      "step": 667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860731692775458e-05,
+      "loss": 1.0581,
+      "step": 668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856989212917995e-05,
+      "loss": 1.053,
+      "step": 669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85324138925098e-05,
+      "loss": 1.0403,
+      "step": 670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849488226966975e-05,
+      "loss": 1.0308,
+      "step": 671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84572973126594e-05,
+      "loss": 1.0115,
+      "step": 672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841965907355221e-05,
+      "loss": 0.9758,
+      "step": 673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838196760449549e-05,
+      "loss": 1.022,
+      "step": 674
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83442229577103e-05,
+      "loss": 0.9985,
+      "step": 675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.830642518549135e-05,
+      "loss": 1.0671,
+      "step": 676
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826857434020699e-05,
+      "loss": 1.011,
+      "step": 677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823067047429907e-05,
+      "loss": 1.0554,
+      "step": 678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819271364028293e-05,
+      "loss": 1.0469,
+      "step": 679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815470389074727e-05,
+      "loss": 1.0217,
+      "step": 680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811664127835413e-05,
+      "loss": 1.0222,
+      "step": 681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807852585583875e-05,
+      "loss": 0.9707,
+      "step": 682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80403576760096e-05,
+      "loss": 1.0107,
+      "step": 683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800213679174818e-05,
+      "loss": 1.043,
+      "step": 684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796386325600905e-05,
+      "loss": 1.0403,
+      "step": 685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79255371218197e-05,
+      "loss": 1.0188,
+      "step": 686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78871584422805e-05,
+      "loss": 1.0208,
+      "step": 687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784872727056463e-05,
+      "loss": 1.0557,
+      "step": 688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781024365991801e-05,
+      "loss": 1.0176,
+      "step": 689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.777170766365917e-05,
+      "loss": 1.0994,
+      "step": 690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773311933517923e-05,
+      "loss": 0.9768,
+      "step": 691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769447872794184e-05,
+      "loss": 1.0559,
+      "step": 692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765578589548309e-05,
+      "loss": 1.0408,
+      "step": 693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761704089141139e-05,
+      "loss": 1.051,
+      "step": 694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757824376940746e-05,
+      "loss": 1.0757,
+      "step": 695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753939458322421e-05,
+      "loss": 1.0403,
+      "step": 696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75004933866867e-05,
+      "loss": 1.072,
+      "step": 697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746154023369205e-05,
+      "loss": 1.0586,
+      "step": 698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742253517820933e-05,
+      "loss": 1.0388,
+      "step": 699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738347827427957e-05,
+      "loss": 1.0864,
+      "step": 700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734436957601563e-05,
+      "loss": 0.998,
+      "step": 701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.73052091376021e-05,
+      "loss": 1.0654,
+      "step": 702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726599701329526e-05,
+      "loss": 1.0554,
+      "step": 703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722673325742301e-05,
+      "loss": 1.053,
+      "step": 704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71874179243848e-05,
+      "loss": 1.0354,
+      "step": 705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714805106865152e-05,
+      "loss": 0.9756,
+      "step": 706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710863274476544e-05,
+      "loss": 0.9634,
+      "step": 707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706916300734016e-05,
+      "loss": 1.0527,
+      "step": 708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702964191106049e-05,
+      "loss": 1.0002,
+      "step": 709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69900695106824e-05,
+      "loss": 1.05,
+      "step": 710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695044586103296e-05,
+      "loss": 1.011,
+      "step": 711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691077101701024e-05,
+      "loss": 1.0225,
+      "step": 712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687104503358322e-05,
+      "loss": 1.0366,
+      "step": 713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683126796579172e-05,
+      "loss": 1.0024,
+      "step": 714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679143986874642e-05,
+      "loss": 1.0869,
+      "step": 715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67515607976286e-05,
+      "loss": 1.0015,
+      "step": 716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671163080769024e-05,
+      "loss": 1.0625,
+      "step": 717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66716499542538e-05,
+      "loss": 1.0471,
+      "step": 718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663161829271224e-05,
+      "loss": 1.031,
+      "step": 719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659153587852895e-05,
+      "loss": 1.0027,
+      "step": 720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65514027672376e-05,
+      "loss": 1.0237,
+      "step": 721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651121901444208e-05,
+      "loss": 1.0261,
+      "step": 722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64709846758165e-05,
+      "loss": 1.0217,
+      "step": 723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6430699807105e-05,
+      "loss": 0.9792,
+      "step": 724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639036446412177e-05,
+      "loss": 1.0576,
+      "step": 725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63499787027509e-05,
+      "loss": 1.072,
+      "step": 726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630954257894639e-05,
+      "loss": 1.0847,
+      "step": 727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626905614873191e-05,
+      "loss": 1.0015,
+      "step": 728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622851946820095e-05,
+      "loss": 1.0464,
+      "step": 729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618793259351654e-05,
+      "loss": 1.0618,
+      "step": 730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614729558091128e-05,
+      "loss": 1.0364,
+      "step": 731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610660848668723e-05,
+      "loss": 1.021,
+      "step": 732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606587136721586e-05,
+      "loss": 1.0046,
+      "step": 733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602508427893794e-05,
+      "loss": 0.9604,
+      "step": 734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.598424727836343e-05,
+      "loss": 1.074,
+      "step": 735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59433604220715e-05,
+      "loss": 1.0601,
+      "step": 736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590242376671035e-05,
+      "loss": 1.1042,
+      "step": 737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586143736899722e-05,
+      "loss": 1.0708,
+      "step": 738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582040128571821e-05,
+      "loss": 1.0737,
+      "step": 739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577931557372833e-05,
+      "loss": 1.0437,
+      "step": 740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573818028995129e-05,
+      "loss": 1.0637,
+      "step": 741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569699549137949e-05,
+      "loss": 1.04,
+      "step": 742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565576123507398e-05,
+      "loss": 1.0718,
+      "step": 743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561447757816428e-05,
+      "loss": 1.0173,
+      "step": 744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557314457784838e-05,
+      "loss": 1.0947,
+      "step": 745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553176229139261e-05,
+      "loss": 1.0491,
+      "step": 746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54903307761316e-05,
+      "loss": 1.0142,
+      "step": 747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544885008946822e-05,
+      "loss": 1.0093,
+      "step": 748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54073202888734e-05,
+      "loss": 1.0139,
+      "step": 749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536574143188618e-05,
+      "loss": 1.0137,
+      "step": 750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532411357611352e-05,
+      "loss": 1.0039,
+      "step": 751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528243677923028e-05,
+      "loss": 1.0344,
+      "step": 752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524071109897914e-05,
+      "loss": 1.0166,
+      "step": 753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519893659317051e-05,
+      "loss": 1.0312,
+      "step": 754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515711331968241e-05,
+      "loss": 1.0437,
+      "step": 755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511524133646049e-05,
+      "loss": 1.0481,
+      "step": 756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.507332070151784e-05,
+      "loss": 1.0044,
+      "step": 757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503135147293495e-05,
+      "loss": 1.0081,
+      "step": 758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498933370885967e-05,
+      "loss": 1.0166,
+      "step": 759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494726746750706e-05,
+      "loss": 1.0251,
+      "step": 760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490515280715937e-05,
+      "loss": 0.9927,
+      "step": 761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486298978616594e-05,
+      "loss": 0.9722,
+      "step": 762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482077846294308e-05,
+      "loss": 1.0139,
+      "step": 763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477851889597407e-05,
+      "loss": 1.0115,
+      "step": 764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473621114380899e-05,
+      "loss": 1.0405,
+      "step": 765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469385526506466e-05,
+      "loss": 0.989,
+      "step": 766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.465145131842466e-05,
+      "loss": 1.0867,
+      "step": 767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460899936263911e-05,
+      "loss": 1.0576,
+      "step": 768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456649945652464e-05,
+      "loss": 0.9868,
+      "step": 769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452395165896432e-05,
+      "loss": 1.0391,
+      "step": 770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448135602890762e-05,
+      "loss": 0.9805,
+      "step": 771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443871262537022e-05,
+      "loss": 0.9648,
+      "step": 772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4396021507434e-05,
+      "loss": 1.021,
+      "step": 773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4353282734247e-05,
+      "loss": 1.0222,
+      "step": 774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431049636502322e-05,
+      "loss": 0.998,
+      "step": 775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426766245904263e-05,
+      "loss": 0.9946,
+      "step": 776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422478107565107e-05,
+      "loss": 1.1062,
+      "step": 777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418185227426016e-05,
+      "loss": 0.9722,
+      "step": 778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.413887611434718e-05,
+      "loss": 1.0181,
+      "step": 779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.409585265545509e-05,
+      "loss": 1.0676,
+      "step": 780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405278195719233e-05,
+      "loss": 1.0835,
+      "step": 781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40096640792328e-05,
+      "loss": 1.0969,
+      "step": 782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396649908131577e-05,
+      "loss": 0.9885,
+      "step": 783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392328702324582e-05,
+      "loss": 1.0032,
+      "step": 784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388002796489267e-05,
+      "loss": 1.0625,
+      "step": 785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.383672196619123e-05,
+      "loss": 0.9792,
+      "step": 786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37933690871414e-05,
+      "loss": 0.9985,
+      "step": 787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374996938780804e-05,
+      "loss": 1.04,
+      "step": 788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370652292832086e-05,
+      "loss": 1.071,
+      "step": 789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36630297688744e-05,
+      "loss": 1.0188,
+      "step": 790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361948996972792e-05,
+      "loss": 1.0679,
+      "step": 791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357590359120518e-05,
+      "loss": 0.9568,
+      "step": 792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35322706936946e-05,
+      "loss": 1.0098,
+      "step": 793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348859133764901e-05,
+      "loss": 0.989,
+      "step": 794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34448655835856e-05,
+      "loss": 1.0244,
+      "step": 795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34010934920858e-05,
+      "loss": 1.0332,
+      "step": 796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335727512379534e-05,
+      "loss": 1.0818,
+      "step": 797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331341053942401e-05,
+      "loss": 1.022,
+      "step": 798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326949979974561e-05,
+      "loss": 1.01,
+      "step": 799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.322554296559792e-05,
+      "loss": 1.0103,
+      "step": 800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.318154009788257e-05,
+      "loss": 0.9866,
+      "step": 801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313749125756501e-05,
+      "loss": 1.0317,
+      "step": 802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30933965056743e-05,
+      "loss": 0.9978,
+      "step": 803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304925590330317e-05,
+      "loss": 1.0242,
+      "step": 804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300506951160788e-05,
+      "loss": 0.9641,
+      "step": 805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296083739180811e-05,
+      "loss": 1.0264,
+      "step": 806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29165596051869e-05,
+      "loss": 0.9888,
+      "step": 807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287223621309055e-05,
+      "loss": 1.0786,
+      "step": 808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282786727692855e-05,
+      "loss": 1.0522,
+      "step": 809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278345285817352e-05,
+      "loss": 1.0288,
+      "step": 810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27389930183611e-05,
+      "loss": 1.0808,
+      "step": 811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.269448781908977e-05,
+      "loss": 0.9844,
+      "step": 812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264993732202094e-05,
+      "loss": 1.0598,
+      "step": 813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260534158887876e-05,
+      "loss": 1.0759,
+      "step": 814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.256070068145007e-05,
+      "loss": 0.9827,
+      "step": 815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251601466158428e-05,
+      "loss": 1.0039,
+      "step": 816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247128359119324e-05,
+      "loss": 0.95,
+      "step": 817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.242650753225137e-05,
+      "loss": 1.0454,
+      "step": 818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.238168654679528e-05,
+      "loss": 1.0447,
+      "step": 819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233682069692388e-05,
+      "loss": 1.0303,
+      "step": 820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229191004479824e-05,
+      "loss": 0.9236,
+      "step": 821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22469546526415e-05,
+      "loss": 1.0518,
+      "step": 822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.220195458273878e-05,
+      "loss": 1.0583,
+      "step": 823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215690989743709e-05,
+      "loss": 1.04,
+      "step": 824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211182065914531e-05,
+      "loss": 1.0864,
+      "step": 825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.206668693033398e-05,
+      "loss": 1.0657,
+      "step": 826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.202150877353531e-05,
+      "loss": 1.0225,
+      "step": 827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197628625134307e-05,
+      "loss": 1.0095,
+      "step": 828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193101942641248e-05,
+      "loss": 0.9707,
+      "step": 829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188570836146014e-05,
+      "loss": 1.0122,
+      "step": 830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184035311926396e-05,
+      "loss": 0.9402,
+      "step": 831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17949537626631e-05,
+      "loss": 1.0315,
+      "step": 832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174951035455772e-05,
+      "loss": 1.0142,
+      "step": 833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170402295790914e-05,
+      "loss": 0.9509,
+      "step": 834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165849163573953e-05,
+      "loss": 1.0151,
+      "step": 835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161291645113199e-05,
+      "loss": 1.0754,
+      "step": 836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156729746723033e-05,
+      "loss": 1.0359,
+      "step": 837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15216347472391e-05,
+      "loss": 0.9644,
+      "step": 838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.14759283544234e-05,
+      "loss": 0.9827,
+      "step": 839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143017835210886e-05,
+      "loss": 1.0293,
+      "step": 840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138438480368153e-05,
+      "loss": 1.0312,
+      "step": 841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133854777258779e-05,
+      "loss": 1.021,
+      "step": 842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129266732233427e-05,
+      "loss": 1.0093,
+      "step": 843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.124674351648773e-05,
+      "loss": 0.9854,
+      "step": 844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120077641867505e-05,
+      "loss": 1.0742,
+      "step": 845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115476609258304e-05,
+      "loss": 1.0059,
+      "step": 846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110871260195842e-05,
+      "loss": 1.0408,
+      "step": 847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106261601060772e-05,
+      "loss": 1.0344,
+      "step": 848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10164763823972e-05,
+      "loss": 1.009,
+      "step": 849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09702937812527e-05,
+      "loss": 1.0679,
+      "step": 850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092406827115964e-05,
+      "loss": 1.0042,
+      "step": 851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087779991616287e-05,
+      "loss": 0.9739,
+      "step": 852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083148878036663e-05,
+      "loss": 0.9658,
+      "step": 853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078513492793438e-05,
+      "loss": 0.968,
+      "step": 854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073873842308883e-05,
+      "loss": 1.0625,
+      "step": 855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069229933011171e-05,
+      "loss": 1.0615,
+      "step": 856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.064581771334381e-05,
+      "loss": 1.0596,
+      "step": 857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059929363718482e-05,
+      "loss": 1.0229,
+      "step": 858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055272716609326e-05,
+      "loss": 1.0515,
+      "step": 859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050611836458637e-05,
+      "loss": 1.0881,
+      "step": 860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.045946729724007e-05,
+      "loss": 1.0391,
+      "step": 861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041277402868882e-05,
+      "loss": 1.0181,
+      "step": 862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036603862362553e-05,
+      "loss": 1.0217,
+      "step": 863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031926114680153e-05,
+      "loss": 1.0994,
+      "step": 864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027244166302642e-05,
+      "loss": 1.0061,
+      "step": 865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022558023716799e-05,
+      "loss": 1.042,
+      "step": 866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.017867693415214e-05,
+      "loss": 1.0273,
+      "step": 867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013173181896283e-05,
+      "loss": 1.0039,
+      "step": 868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00847449566419e-05,
+      "loss": 1.0447,
+      "step": 869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.003771641228904e-05,
+      "loss": 0.9729,
+      "step": 870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999064625106174e-05,
+      "loss": 1.0215,
+      "step": 871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994353453817507e-05,
+      "loss": 0.9971,
+      "step": 872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989638133890174e-05,
+      "loss": 1.0391,
+      "step": 873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984918671857189e-05,
+      "loss": 1.0623,
+      "step": 874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980195074257307e-05,
+      "loss": 0.9998,
+      "step": 875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975467347635012e-05,
+      "loss": 0.9897,
+      "step": 876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97073549854051e-05,
+      "loss": 1.0376,
+      "step": 877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965999533529718e-05,
+      "loss": 1.0613,
+      "step": 878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961259459164255e-05,
+      "loss": 1.0359,
+      "step": 879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.956515282011434e-05,
+      "loss": 0.9866,
+      "step": 880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951767008644251e-05,
+      "loss": 0.9719,
+      "step": 881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947014645641379e-05,
+      "loss": 1.0715,
+      "step": 882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942258199587158e-05,
+      "loss": 1.0317,
+      "step": 883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.937497677071584e-05,
+      "loss": 0.9685,
+      "step": 884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932733084690296e-05,
+      "loss": 1.0962,
+      "step": 885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92796442904458e-05,
+      "loss": 1.0154,
+      "step": 886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923191716741347e-05,
+      "loss": 1.0347,
+      "step": 887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91841495439313e-05,
+      "loss": 1.0669,
+      "step": 888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.913634148618072e-05,
+      "loss": 1.0291,
+      "step": 889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908849306039918e-05,
+      "loss": 0.9807,
+      "step": 890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904060433288006e-05,
+      "loss": 1.0364,
+      "step": 891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89926753699726e-05,
+      "loss": 1.0068,
+      "step": 892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894470623808176e-05,
+      "loss": 1.061,
+      "step": 893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889669700366817e-05,
+      "loss": 1.011,
+      "step": 894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884864773324802e-05,
+      "loss": 1.0063,
+      "step": 895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.880055849339294e-05,
+      "loss": 0.9514,
+      "step": 896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.875242935073e-05,
+      "loss": 1.0354,
+      "step": 897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.870426037194145e-05,
+      "loss": 1.0166,
+      "step": 898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865605162376486e-05,
+      "loss": 1.0183,
+      "step": 899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860780317299281e-05,
+      "loss": 0.9553,
+      "step": 900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855951508647295e-05,
+      "loss": 0.9849,
+      "step": 901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851118743110775e-05,
+      "loss": 0.9934,
+      "step": 902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846282027385461e-05,
+      "loss": 1.0183,
+      "step": 903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.841441368172558e-05,
+      "loss": 0.9905,
+      "step": 904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836596772178741e-05,
+      "loss": 1.0203,
+      "step": 905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.831748246116136e-05,
+      "loss": 1.0586,
+      "step": 906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826895796702312e-05,
+      "loss": 1.0435,
+      "step": 907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822039430660276e-05,
+      "loss": 1.0688,
+      "step": 908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817179154718463e-05,
+      "loss": 1.011,
+      "step": 909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.812314975610721e-05,
+      "loss": 0.9758,
+      "step": 910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807446900076314e-05,
+      "loss": 1.0181,
+      "step": 911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802574934859895e-05,
+      "loss": 1.0173,
+      "step": 912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797699086711507e-05,
+      "loss": 1.0518,
+      "step": 913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79281936238658e-05,
+      "loss": 1.0349,
+      "step": 914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787935768645911e-05,
+      "loss": 1.01,
+      "step": 915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783048312255653e-05,
+      "loss": 1.0146,
+      "step": 916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778156999987317e-05,
+      "loss": 0.9199,
+      "step": 917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773261838617753e-05,
+      "loss": 1.0288,
+      "step": 918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768362834929146e-05,
+      "loss": 0.9556,
+      "step": 919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763459995709004e-05,
+      "loss": 1.0608,
+      "step": 920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758553327750146e-05,
+      "loss": 0.9556,
+      "step": 921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.753642837850698e-05,
+      "loss": 1.0154,
+      "step": 922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748728532814087e-05,
+      "loss": 0.9644,
+      "step": 923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743810419449014e-05,
+      "loss": 1.021,
+      "step": 924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.738888504569463e-05,
+      "loss": 1.1455,
+      "step": 925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733962794994688e-05,
+      "loss": 1.0381,
+      "step": 926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729033297549195e-05,
+      "loss": 1.0515,
+      "step": 927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.724100019062738e-05,
+      "loss": 1.0208,
+      "step": 928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719162966370316e-05,
+      "loss": 1.064,
+      "step": 929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.714222146312151e-05,
+      "loss": 1.0461,
+      "step": 930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709277565733686e-05,
+      "loss": 0.9473,
+      "step": 931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704329231485575e-05,
+      "loss": 0.9629,
+      "step": 932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699377150423672e-05,
+      "loss": 0.9937,
+      "step": 933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694421329409021e-05,
+      "loss": 0.958,
+      "step": 934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689461775307851e-05,
+      "loss": 0.9844,
+      "step": 935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.684498494991561e-05,
+      "loss": 1.0122,
+      "step": 936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679531495336711e-05,
+      "loss": 0.9678,
+      "step": 937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674560783225017e-05,
+      "loss": 1.0054,
+      "step": 938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.669586365543341e-05,
+      "loss": 1.0405,
+      "step": 939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66460824918367e-05,
+      "loss": 1.0342,
+      "step": 940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659626441043124e-05,
+      "loss": 1.0583,
+      "step": 941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654640948023934e-05,
+      "loss": 1.0083,
+      "step": 942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649651777033437e-05,
+      "loss": 1.084,
+      "step": 943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644658934984066e-05,
+      "loss": 1.0447,
+      "step": 944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639662428793342e-05,
+      "loss": 1.0583,
+      "step": 945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634662265383858e-05,
+      "loss": 1.0635,
+      "step": 946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629658451683279e-05,
+      "loss": 1.05,
+      "step": 947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624650994624325e-05,
+      "loss": 0.9512,
+      "step": 948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619639901144764e-05,
+      "loss": 0.9834,
+      "step": 949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614625178187402e-05,
+      "loss": 1.041,
+      "step": 950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.609606832700074e-05,
+      "loss": 1.0452,
+      "step": 951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.604584871635634e-05,
+      "loss": 1.0359,
+      "step": 952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599559301951944e-05,
+      "loss": 1.0093,
+      "step": 953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.59453013061187e-05,
+      "loss": 1.0566,
+      "step": 954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589497364583262e-05,
+      "loss": 0.9873,
+      "step": 955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.584461010838955e-05,
+      "loss": 0.9685,
+      "step": 956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579421076356753e-05,
+      "loss": 1.123,
+      "step": 957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574377568119421e-05,
+      "loss": 0.938,
+      "step": 958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.569330493114675e-05,
+      "loss": 0.991,
+      "step": 959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564279858335174e-05,
+      "loss": 0.999,
+      "step": 960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559225670778508e-05,
+      "loss": 0.97,
+      "step": 961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55416793744719e-05,
+      "loss": 1.0793,
+      "step": 962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549106665348644e-05,
+      "loss": 1.0032,
+      "step": 963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544041861495201e-05,
+      "loss": 1.0083,
+      "step": 964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53897353290408e-05,
+      "loss": 0.9854,
+      "step": 965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.533901686597386e-05,
+      "loss": 0.9937,
+      "step": 966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528826329602099e-05,
+      "loss": 0.9556,
+      "step": 967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523747468950061e-05,
+      "loss": 1.071,
+      "step": 968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.518665111677967e-05,
+      "loss": 0.9709,
+      "step": 969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513579264827361e-05,
+      "loss": 1.04,
+      "step": 970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.508489935444618e-05,
+      "loss": 1.0237,
+      "step": 971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.503397130580941e-05,
+      "loss": 1.0085,
+      "step": 972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.498300857292341e-05,
+      "loss": 0.9812,
+      "step": 973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.493201122639649e-05,
+      "loss": 1.0203,
+      "step": 974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488097933688474e-05,
+      "loss": 1.021,
+      "step": 975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482991297509224e-05,
+      "loss": 0.9885,
+      "step": 976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.477881221177076e-05,
+      "loss": 1.0178,
+      "step": 977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472767711771978e-05,
+      "loss": 0.9849,
+      "step": 978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.467650776378633e-05,
+      "loss": 1.0273,
+      "step": 979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462530422086486e-05,
+      "loss": 1.012,
+      "step": 980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457406655989728e-05,
+      "loss": 1.0317,
+      "step": 981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452279485187268e-05,
+      "loss": 1.0493,
+      "step": 982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447148916782736e-05,
+      "loss": 1.0369,
+      "step": 983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442014957884472e-05,
+      "loss": 0.9656,
+      "step": 984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43687761560551e-05,
+      "loss": 1.0388,
+      "step": 985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431736897063569e-05,
+      "loss": 0.9553,
+      "step": 986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.426592809381054e-05,
+      "loss": 1.0125,
+      "step": 987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.421445359685031e-05,
+      "loss": 1.0332,
+      "step": 988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416294555107227e-05,
+      "loss": 0.9631,
+      "step": 989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411140402784014e-05,
+      "loss": 0.9954,
+      "step": 990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405982909856407e-05,
+      "loss": 1.0242,
+      "step": 991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.400822083470046e-05,
+      "loss": 0.9871,
+      "step": 992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395657930775191e-05,
+      "loss": 1.0879,
+      "step": 993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390490458926707e-05,
+      "loss": 0.9685,
+      "step": 994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385319675084066e-05,
+      "loss": 1.0271,
+      "step": 995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38014558641132e-05,
+      "loss": 1.0178,
+      "step": 996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374968200077102e-05,
+      "loss": 0.9856,
+      "step": 997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369787523254617e-05,
+      "loss": 0.946,
+      "step": 998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364603563121627e-05,
+      "loss": 1.0432,
+      "step": 999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359416326860442e-05,
+      "loss": 1.0154,
+      "step": 1000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354225821657914e-05,
+      "loss": 0.9729,
+      "step": 1001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.349032054705417e-05,
+      "loss": 1.0391,
+      "step": 1002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343835033198854e-05,
+      "loss": 0.9795,
+      "step": 1003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.33863476433863e-05,
+      "loss": 1.0491,
+      "step": 1004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333431255329653e-05,
+      "loss": 0.9915,
+      "step": 1005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.328224513381316e-05,
+      "loss": 1.0198,
+      "step": 1006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323014545707496e-05,
+      "loss": 0.9534,
+      "step": 1007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.317801359526537e-05,
+      "loss": 1.0322,
+      "step": 1008
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312584962061243e-05,
+      "loss": 1.0117,
+      "step": 1009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.307365360538865e-05,
+      "loss": 0.9836,
+      "step": 1010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302142562191092e-05,
+      "loss": 0.9941,
+      "step": 1011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296916574254049e-05,
+      "loss": 1.021,
+      "step": 1012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291687403968276e-05,
+      "loss": 1.0161,
+      "step": 1013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.286455058578718e-05,
+      "loss": 0.9163,
+      "step": 1014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.281219545334727e-05,
+      "loss": 1.0208,
+      "step": 1015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.275980871490038e-05,
+      "loss": 1.0896,
+      "step": 1016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.270739044302766e-05,
+      "loss": 1.0349,
+      "step": 1017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.265494071035401e-05,
+      "loss": 1.0715,
+      "step": 1018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.260245958954782e-05,
+      "loss": 1.1155,
+      "step": 1019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254994715332102e-05,
+      "loss": 0.9941,
+      "step": 1020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249740347442895e-05,
+      "loss": 0.9692,
+      "step": 1021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244482862567018e-05,
+      "loss": 0.9885,
+      "step": 1022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23922226798865e-05,
+      "loss": 0.9053,
+      "step": 1023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23395857099628e-05,
+      "loss": 1.0022,
+      "step": 1024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228691778882693e-05,
+      "loss": 1.0842,
+      "step": 1025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.22342189894496e-05,
+      "loss": 1.01,
+      "step": 1026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218148938484434e-05,
+      "loss": 0.9922,
+      "step": 1027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.212872904806736e-05,
+      "loss": 0.9915,
+      "step": 1028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.207593805221742e-05,
+      "loss": 0.9653,
+      "step": 1029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.202311647043579e-05,
+      "loss": 1.0703,
+      "step": 1030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197026437590609e-05,
+      "loss": 1.0288,
+      "step": 1031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.191738184185422e-05,
+      "loss": 1.0381,
+      "step": 1032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186446894154826e-05,
+      "loss": 0.9937,
+      "step": 1033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181152574829837e-05,
+      "loss": 0.9614,
+      "step": 1034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175855233545668e-05,
+      "loss": 0.9502,
+      "step": 1035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.170554877641716e-05,
+      "loss": 1.0469,
+      "step": 1036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165251514461557e-05,
+      "loss": 1.0557,
+      "step": 1037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159945151352935e-05,
+      "loss": 1.0491,
+      "step": 1038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.154635795667748e-05,
+      "loss": 0.97,
+      "step": 1039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14932345476204e-05,
+      "loss": 1.0215,
+      "step": 1040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144008135995991e-05,
+      "loss": 1.1038,
+      "step": 1041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.138689846733911e-05,
+      "loss": 1.0549,
+      "step": 1042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.13336859434422e-05,
+      "loss": 0.9995,
+      "step": 1043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.128044386199445e-05,
+      "loss": 1.04,
+      "step": 1044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.122717229676208e-05,
+      "loss": 1.0002,
+      "step": 1045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11738713215522e-05,
+      "loss": 1.0388,
+      "step": 1046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112054101021262e-05,
+      "loss": 1.0227,
+      "step": 1047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106718143663178e-05,
+      "loss": 0.9951,
+      "step": 1048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101379267473873e-05,
+      "loss": 1.0481,
+      "step": 1049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096037479850291e-05,
+      "loss": 0.9717,
+      "step": 1050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.090692788193409e-05,
+      "loss": 1.0625,
+      "step": 1051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085345199908235e-05,
+      "loss": 1.0105,
+      "step": 1052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079994722403777e-05,
+      "loss": 0.9771,
+      "step": 1053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.074641363093058e-05,
+      "loss": 1.0442,
+      "step": 1054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.069285129393089e-05,
+      "loss": 0.9688,
+      "step": 1055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063926028724861e-05,
+      "loss": 0.9697,
+      "step": 1056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058564068513343e-05,
+      "loss": 0.9941,
+      "step": 1057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053199256187464e-05,
+      "loss": 0.9937,
+      "step": 1058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.047831599180099e-05,
+      "loss": 0.9968,
+      "step": 1059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.042461104928072e-05,
+      "loss": 0.9128,
+      "step": 1060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037087780872133e-05,
+      "loss": 0.9792,
+      "step": 1061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031711634456954e-05,
+      "loss": 0.9456,
+      "step": 1062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02633267313112e-05,
+      "loss": 0.9539,
+      "step": 1063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02095090434711e-05,
+      "loss": 1.0061,
+      "step": 1064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015566335561296e-05,
+      "loss": 1.0134,
+      "step": 1065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.010178974233935e-05,
+      "loss": 1.0088,
+      "step": 1066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.004788827829143e-05,
+      "loss": 1.0256,
+      "step": 1067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9993959038149e-05,
+      "loss": 1.0872,
+      "step": 1068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994000209663036e-05,
+      "loss": 0.9465,
+      "step": 1069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988601752849213e-05,
+      "loss": 1.0752,
+      "step": 1070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983200540852927e-05,
+      "loss": 0.9797,
+      "step": 1071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97779658115749e-05,
+      "loss": 0.9478,
+      "step": 1072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972389881250015e-05,
+      "loss": 1.0129,
+      "step": 1073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.966980448621419e-05,
+      "loss": 1.0481,
+      "step": 1074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961568290766403e-05,
+      "loss": 0.9517,
+      "step": 1075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.95615341518344e-05,
+      "loss": 1.1074,
+      "step": 1076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950735829374772e-05,
+      "loss": 0.939,
+      "step": 1077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9453155408464e-05,
+      "loss": 1.0679,
+      "step": 1078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939892557108058e-05,
+      "loss": 0.9541,
+      "step": 1079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.934466885673225e-05,
+      "loss": 1.0068,
+      "step": 1080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9290385340591e-05,
+      "loss": 0.9939,
+      "step": 1081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923607509786593e-05,
+      "loss": 0.9922,
+      "step": 1082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.918173820380321e-05,
+      "loss": 0.9558,
+      "step": 1083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91273747336859e-05,
+      "loss": 0.9731,
+      "step": 1084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907298476283392e-05,
+      "loss": 1.0254,
+      "step": 1085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901856836660386e-05,
+      "loss": 0.9534,
+      "step": 1086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.896412562038897e-05,
+      "loss": 1.0083,
+      "step": 1087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890965659961897e-05,
+      "loss": 1.0129,
+      "step": 1088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885516137975998e-05,
+      "loss": 1.0281,
+      "step": 1089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880064003631445e-05,
+      "loss": 1.0923,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.874609264482102e-05,
+      "loss": 0.9998,
+      "step": 1091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.86915192808544e-05,
+      "loss": 1.0708,
+      "step": 1092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863692002002529e-05,
+      "loss": 1.0068,
+      "step": 1093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858229493798026e-05,
+      "loss": 1.011,
+      "step": 1094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85276441104017e-05,
+      "loss": 0.9827,
+      "step": 1095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.84729676130076e-05,
+      "loss": 0.9482,
+      "step": 1096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.841826552155157e-05,
+      "loss": 0.9192,
+      "step": 1097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.836353791182266e-05,
+      "loss": 1.0176,
+      "step": 1098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830878485964528e-05,
+      "loss": 1.0129,
+      "step": 1099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82540064408791e-05,
+      "loss": 1.0315,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.81992027314189e-05,
+      "loss": 0.9788,
+      "step": 1101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814437380719453e-05,
+      "loss": 1.0349,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808951974417078e-05,
+      "loss": 1.0229,
+      "step": 1103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803464061834725e-05,
+      "loss": 1.0129,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.797973650575826e-05,
+      "loss": 0.9641,
+      "step": 1105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792480748247278e-05,
+      "loss": 1.0042,
+      "step": 1106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786985362459427e-05,
+      "loss": 1.0444,
+      "step": 1107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781487500826059e-05,
+      "loss": 0.9971,
+      "step": 1108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.775987170964393e-05,
+      "loss": 1.0271,
+      "step": 1109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770484380495064e-05,
+      "loss": 1.021,
+      "step": 1110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76497913704212e-05,
+      "loss": 0.9768,
+      "step": 1111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759471448233007e-05,
+      "loss": 0.9248,
+      "step": 1112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753961321698555e-05,
+      "loss": 0.9968,
+      "step": 1113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.748448765072977e-05,
+      "loss": 1.0681,
+      "step": 1114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742933785993847e-05,
+      "loss": 0.9229,
+      "step": 1115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7374163921021e-05,
+      "loss": 0.9939,
+      "step": 1116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.731896591042016e-05,
+      "loss": 1.0867,
+      "step": 1117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.72637439046121e-05,
+      "loss": 0.9417,
+      "step": 1118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720849798010618e-05,
+      "loss": 1.0491,
+      "step": 1119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715322821344494e-05,
+      "loss": 1.0073,
+      "step": 1120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.709793468120395e-05,
+      "loss": 1.0266,
+      "step": 1121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704261745999168e-05,
+      "loss": 0.9536,
+      "step": 1122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698727662644944e-05,
+      "loss": 0.9629,
+      "step": 1123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693191225725125e-05,
+      "loss": 1.0015,
+      "step": 1124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.687652442910376e-05,
+      "loss": 1.0425,
+      "step": 1125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682111321874608e-05,
+      "loss": 0.9133,
+      "step": 1126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.676567870294975e-05,
+      "loss": 0.9861,
+      "step": 1127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671022095851856e-05,
+      "loss": 1.0183,
+      "step": 1128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665474006228855e-05,
+      "loss": 1.0464,
+      "step": 1129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65992360911278e-05,
+      "loss": 1.0017,
+      "step": 1130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654370912193633e-05,
+      "loss": 1.051,
+      "step": 1131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.648815923164603e-05,
+      "loss": 0.9668,
+      "step": 1132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643258649722061e-05,
+      "loss": 0.9331,
+      "step": 1133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637699099565537e-05,
+      "loss": 1.04,
+      "step": 1134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.632137280397719e-05,
+      "loss": 0.9849,
+      "step": 1135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626573199924433e-05,
+      "loss": 1.0093,
+      "step": 1136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621006865854644e-05,
+      "loss": 1.0195,
+      "step": 1137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615438285900438e-05,
+      "loss": 0.9922,
+      "step": 1138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.609867467777012e-05,
+      "loss": 1.0496,
+      "step": 1139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60429441920266e-05,
+      "loss": 1.0254,
+      "step": 1140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.598719147898773e-05,
+      "loss": 0.9287,
+      "step": 1141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59314166158982e-05,
+      "loss": 1.022,
+      "step": 1142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.587561968003332e-05,
+      "loss": 1.1118,
+      "step": 1143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58198007486991e-05,
+      "loss": 1.0186,
+      "step": 1144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.576395989923193e-05,
+      "loss": 0.9771,
+      "step": 1145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57080972089986e-05,
+      "loss": 1.0837,
+      "step": 1146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.565221275539615e-05,
+      "loss": 1.05,
+      "step": 1147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.559630661585178e-05,
+      "loss": 0.9929,
+      "step": 1148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554037886782276e-05,
+      "loss": 1.0483,
+      "step": 1149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.548442958879624e-05,
+      "loss": 0.9768,
+      "step": 1150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.542845885628925e-05,
+      "loss": 0.9993,
+      "step": 1151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.537246674784855e-05,
+      "loss": 0.9805,
+      "step": 1152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531645334105046e-05,
+      "loss": 0.9436,
+      "step": 1153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.526041871350086e-05,
+      "loss": 0.9524,
+      "step": 1154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.520436294283503e-05,
+      "loss": 1.0608,
+      "step": 1155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514828610671751e-05,
+      "loss": 0.9607,
+      "step": 1156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.509218828284203e-05,
+      "loss": 1.012,
+      "step": 1157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.503606954893144e-05,
+      "loss": 1.0342,
+      "step": 1158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497992998273751e-05,
+      "loss": 0.9312,
+      "step": 1159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.492376966204093e-05,
+      "loss": 0.9463,
+      "step": 1160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486758866465106e-05,
+      "loss": 0.9744,
+      "step": 1161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.481138706840599e-05,
+      "loss": 0.9763,
+      "step": 1162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475516495117234e-05,
+      "loss": 0.9656,
+      "step": 1163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.469892239084508e-05,
+      "loss": 0.9895,
+      "step": 1164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.464265946534763e-05,
+      "loss": 0.9934,
+      "step": 1165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.458637625263149e-05,
+      "loss": 0.9878,
+      "step": 1166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453007283067638e-05,
+      "loss": 1.0117,
+      "step": 1167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.447374927748997e-05,
+      "loss": 1.0027,
+      "step": 1168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.441740567110782e-05,
+      "loss": 0.9719,
+      "step": 1169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436104208959329e-05,
+      "loss": 1.0054,
+      "step": 1170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43046586110374e-05,
+      "loss": 0.9587,
+      "step": 1171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.424825531355873e-05,
+      "loss": 1.0046,
+      "step": 1172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.419183227530336e-05,
+      "loss": 1.0015,
+      "step": 1173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.413538957444468e-05,
+      "loss": 0.9939,
+      "step": 1174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.407892728918332e-05,
+      "loss": 1.043,
+      "step": 1175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.402244549774707e-05,
+      "loss": 0.948,
+      "step": 1176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.396594427839075e-05,
+      "loss": 1.0269,
+      "step": 1177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.390942370939606e-05,
+      "loss": 0.9868,
+      "step": 1178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.385288386907155e-05,
+      "loss": 1.0178,
+      "step": 1179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.379632483575242e-05,
+      "loss": 1.0037,
+      "step": 1180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373974668780053e-05,
+      "loss": 1.0591,
+      "step": 1181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.368314950360415e-05,
+      "loss": 1.0334,
+      "step": 1182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362653336157798e-05,
+      "loss": 0.999,
+      "step": 1183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.356989834016296e-05,
+      "loss": 0.9617,
+      "step": 1184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.35132445178262e-05,
+      "loss": 1.0195,
+      "step": 1185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.345657197306083e-05,
+      "loss": 0.9834,
+      "step": 1186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.339988078438597e-05,
+      "loss": 0.97,
+      "step": 1187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.334317103034652e-05,
+      "loss": 0.9746,
+      "step": 1188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.328644278951313e-05,
+      "loss": 1.0281,
+      "step": 1189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.322969614048207e-05,
+      "loss": 1.0076,
+      "step": 1190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317293116187509e-05,
+      "loss": 0.9795,
+      "step": 1191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.311614793233932e-05,
+      "loss": 0.9646,
+      "step": 1192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.305934653054725e-05,
+      "loss": 0.9565,
+      "step": 1193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.300252703519647e-05,
+      "loss": 0.9993,
+      "step": 1194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.294568952500968e-05,
+      "loss": 1.0107,
+      "step": 1195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288883407873452e-05,
+      "loss": 0.9485,
+      "step": 1196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.283196077514351e-05,
+      "loss": 0.9292,
+      "step": 1197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.277506969303386e-05,
+      "loss": 1.0088,
+      "step": 1198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.271816091122748e-05,
+      "loss": 1.0547,
+      "step": 1199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266123450857071e-05,
+      "loss": 1.0474,
+      "step": 1200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26042905639344e-05,
+      "loss": 0.9575,
+      "step": 1201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.254732915621365e-05,
+      "loss": 1.033,
+      "step": 1202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249035036432775e-05,
+      "loss": 0.9915,
+      "step": 1203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243335426722014e-05,
+      "loss": 1.0015,
+      "step": 1204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.237634094385813e-05,
+      "loss": 0.9829,
+      "step": 1205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.231931047323299e-05,
+      "loss": 0.9333,
+      "step": 1206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.226226293435973e-05,
+      "loss": 1.019,
+      "step": 1207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220519840627694e-05,
+      "loss": 1.04,
+      "step": 1208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214811696804682e-05,
+      "loss": 0.9766,
+      "step": 1209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2091018698755e-05,
+      "loss": 0.9395,
+      "step": 1210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203390367751038e-05,
+      "loss": 0.9641,
+      "step": 1211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.197677198344508e-05,
+      "loss": 0.9875,
+      "step": 1212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.191962369571439e-05,
+      "loss": 0.9597,
+      "step": 1213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18624588934965e-05,
+      "loss": 1.0413,
+      "step": 1214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180527765599251e-05,
+      "loss": 1.0825,
+      "step": 1215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.174808006242634e-05,
+      "loss": 1.0244,
+      "step": 1216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.169086619204447e-05,
+      "loss": 1.0471,
+      "step": 1217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163363612411604e-05,
+      "loss": 1.0415,
+      "step": 1218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.157638993793256e-05,
+      "loss": 1.009,
+      "step": 1219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15191277128079e-05,
+      "loss": 0.9246,
+      "step": 1220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.146184952807815e-05,
+      "loss": 1.0464,
+      "step": 1221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.140455546310148e-05,
+      "loss": 0.9329,
+      "step": 1222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.134724559725811e-05,
+      "loss": 1.0181,
+      "step": 1223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.128992000995016e-05,
+      "loss": 0.9768,
+      "step": 1224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.123257878060146e-05,
+      "loss": 1.0278,
+      "step": 1225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.117522198865755e-05,
+      "loss": 0.9998,
+      "step": 1226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.111784971358556e-05,
+      "loss": 0.9924,
+      "step": 1227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.106046203487405e-05,
+      "loss": 1.0281,
+      "step": 1228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1003059032032916e-05,
+      "loss": 0.9749,
+      "step": 1229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.094564078459328e-05,
+      "loss": 0.9106,
+      "step": 1230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.08882073721074e-05,
+      "loss": 1.0305,
+      "step": 1231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.083075887414853e-05,
+      "loss": 0.9407,
+      "step": 1232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.077329537031087e-05,
+      "loss": 1.0869,
+      "step": 1233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0715816940209325e-05,
+      "loss": 0.9521,
+      "step": 1234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.065832366347956e-05,
+      "loss": 1.0264,
+      "step": 1235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0600815619777774e-05,
+      "loss": 1.0447,
+      "step": 1236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.054329288878061e-05,
+      "loss": 1.0488,
+      "step": 1237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.048575555018512e-05,
+      "loss": 0.9507,
+      "step": 1238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042820368370854e-05,
+      "loss": 1.0361,
+      "step": 1239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0370637369088226e-05,
+      "loss": 1.052,
+      "step": 1240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0313056686081607e-05,
+      "loss": 0.9919,
+      "step": 1241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.025546171446599e-05,
+      "loss": 0.9753,
+      "step": 1242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0197852534038434e-05,
+      "loss": 1.0085,
+      "step": 1243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.014022922461576e-05,
+      "loss": 0.98,
+      "step": 1244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.008259186603433e-05,
+      "loss": 1.0625,
+      "step": 1245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.002494053814997e-05,
+      "loss": 1.0591,
+      "step": 1246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9967275320837855e-05,
+      "loss": 0.9463,
+      "step": 1247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.990959629399242e-05,
+      "loss": 0.9387,
+      "step": 1248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985190353752722e-05,
+      "loss": 0.9656,
+      "step": 1249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.979419713137484e-05,
+      "loss": 0.9363,
+      "step": 1250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.973647715548676e-05,
+      "loss": 1.0222,
+      "step": 1251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.967874368983328e-05,
+      "loss": 0.9866,
+      "step": 1252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.962099681440341e-05,
+      "loss": 0.9531,
+      "step": 1253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9563236609204654e-05,
+      "loss": 0.979,
+      "step": 1254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9505463154263084e-05,
+      "loss": 0.9688,
+      "step": 1255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944767652962309e-05,
+      "loss": 0.9519,
+      "step": 1256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.938987681534729e-05,
+      "loss": 0.9609,
+      "step": 1257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.933206409151646e-05,
+      "loss": 1.0132,
+      "step": 1258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.92742384382294e-05,
+      "loss": 1.0134,
+      "step": 1259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.921639993560279e-05,
+      "loss": 0.9834,
+      "step": 1260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.915854866377119e-05,
+      "loss": 0.9868,
+      "step": 1261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.910068470288677e-05,
+      "loss": 1.0486,
+      "step": 1262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.90428081331193e-05,
+      "loss": 1.0256,
+      "step": 1263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.898491903465606e-05,
+      "loss": 1.0439,
+      "step": 1264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8927017487701656e-05,
+      "loss": 1.0159,
+      "step": 1265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.886910357247792e-05,
+      "loss": 1.019,
+      "step": 1266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.881117736922389e-05,
+      "loss": 0.9724,
+      "step": 1267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8753238958195534e-05,
+      "loss": 0.9441,
+      "step": 1268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.869528841966583e-05,
+      "loss": 0.9209,
+      "step": 1269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.863732583392449e-05,
+      "loss": 1.0083,
+      "step": 1270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.857935128127793e-05,
+      "loss": 0.98,
+      "step": 1271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8521364842049175e-05,
+      "loss": 1.0229,
+      "step": 1272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8463366596577706e-05,
+      "loss": 1.0093,
+      "step": 1273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.840535662521933e-05,
+      "loss": 1.0088,
+      "step": 1274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8347335008346146e-05,
+      "loss": 1.0542,
+      "step": 1275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.828930182634638e-05,
+      "loss": 0.9597,
+      "step": 1276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.823125715962421e-05,
+      "loss": 0.9146,
+      "step": 1277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.817320108859984e-05,
+      "loss": 1.019,
+      "step": 1278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81151336937092e-05,
+      "loss": 1.0186,
+      "step": 1279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.805705505540392e-05,
+      "loss": 0.928,
+      "step": 1280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7998965254151235e-05,
+      "loss": 0.97,
+      "step": 1281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794086437043382e-05,
+      "loss": 1.0403,
+      "step": 1282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7882752484749694e-05,
+      "loss": 0.9741,
+      "step": 1283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.782462967761216e-05,
+      "loss": 0.9951,
+      "step": 1284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7766496029549624e-05,
+      "loss": 0.989,
+      "step": 1285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.770835162110551e-05,
+      "loss": 0.9724,
+      "step": 1286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.765019653283815e-05,
+      "loss": 0.9861,
+      "step": 1287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7592030845320676e-05,
+      "loss": 0.9763,
+      "step": 1288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.753385463914094e-05,
+      "loss": 0.9814,
+      "step": 1289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.747566799490132e-05,
+      "loss": 0.9521,
+      "step": 1290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.741747099321866e-05,
+      "loss": 0.9697,
+      "step": 1291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7359263714724175e-05,
+      "loss": 0.9497,
+      "step": 1292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7301046240063325e-05,
+      "loss": 0.9988,
+      "step": 1293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.724281864989567e-05,
+      "loss": 1.0613,
+      "step": 1294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.718458102489479e-05,
+      "loss": 0.9021,
+      "step": 1295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.712633344574816e-05,
+      "loss": 1.0498,
+      "step": 1296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7068075993157096e-05,
+      "loss": 0.9697,
+      "step": 1297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7009808747836546e-05,
+      "loss": 1.0457,
+      "step": 1298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.695153179051504e-05,
+      "loss": 1.0269,
+      "step": 1299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.689324520193454e-05,
+      "loss": 0.9924,
+      "step": 1300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6834949062850395e-05,
+      "loss": 0.9719,
+      "step": 1301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.677664345403117e-05,
+      "loss": 1.0166,
+      "step": 1302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.671832845625853e-05,
+      "loss": 0.9802,
+      "step": 1303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.666000415032717e-05,
+      "loss": 0.9033,
+      "step": 1304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6601670617044666e-05,
+      "loss": 1.0312,
+      "step": 1305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6543327937231404e-05,
+      "loss": 0.9688,
+      "step": 1306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.648497619172042e-05,
+      "loss": 1.0244,
+      "step": 1307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.64266154613573e-05,
+      "loss": 1.0076,
+      "step": 1308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.636824582700012e-05,
+      "loss": 0.9539,
+      "step": 1309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.630986736951925e-05,
+      "loss": 1.0391,
+      "step": 1310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625148016979731e-05,
+      "loss": 1.0254,
+      "step": 1311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6193084308729013e-05,
+      "loss": 0.9961,
+      "step": 1312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6134679867221096e-05,
+      "loss": 0.9417,
+      "step": 1313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.607626692619216e-05,
+      "loss": 1.0283,
+      "step": 1314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6017845566572594e-05,
+      "loss": 0.9924,
+      "step": 1315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5959415869304445e-05,
+      "loss": 0.9709,
+      "step": 1316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5900977915341326e-05,
+      "loss": 0.9392,
+      "step": 1317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.584253178564829e-05,
+      "loss": 0.9697,
+      "step": 1318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5784077561201676e-05,
+      "loss": 0.9194,
+      "step": 1319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57256153229891e-05,
+      "loss": 0.9453,
+      "step": 1320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5667145152009235e-05,
+      "loss": 0.9651,
+      "step": 1321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5608667129271754e-05,
+      "loss": 0.9724,
+      "step": 1322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5550181335797223e-05,
+      "loss": 0.9702,
+      "step": 1323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.549168785261698e-05,
+      "loss": 0.8958,
+      "step": 1324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5433186760772973e-05,
+      "loss": 0.9585,
+      "step": 1325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.537467814131774e-05,
+      "loss": 0.9587,
+      "step": 1326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5316162075314226e-05,
+      "loss": 1.0444,
+      "step": 1327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525763864383571e-05,
+      "loss": 1.0452,
+      "step": 1328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.519910792796564e-05,
+      "loss": 0.9561,
+      "step": 1329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5140570008797586e-05,
+      "loss": 0.9968,
+      "step": 1330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.508202496743511e-05,
+      "loss": 1.0745,
+      "step": 1331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.50234728849916e-05,
+      "loss": 1.0042,
+      "step": 1332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4964913842590214e-05,
+      "loss": 0.9692,
+      "step": 1333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.49063479213638e-05,
+      "loss": 0.9587,
+      "step": 1334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.484777520245466e-05,
+      "loss": 0.9888,
+      "step": 1335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.478919576701459e-05,
+      "loss": 0.9512,
+      "step": 1336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4730609696204615e-05,
+      "loss": 1.0278,
+      "step": 1337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.467201707119502e-05,
+      "loss": 1.0024,
+      "step": 1338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4613417973165106e-05,
+      "loss": 0.9365,
+      "step": 1339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.455481248330322e-05,
+      "loss": 1.05,
+      "step": 1340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4496200682806495e-05,
+      "loss": 0.9927,
+      "step": 1341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.443758265288086e-05,
+      "loss": 0.9314,
+      "step": 1342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4378958474740826e-05,
+      "loss": 1.0322,
+      "step": 1343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4320328229609475e-05,
+      "loss": 1.0005,
+      "step": 1344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.426169199871824e-05,
+      "loss": 0.9646,
+      "step": 1345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.42030498633069e-05,
+      "loss": 1.0298,
+      "step": 1346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.414440190462335e-05,
+      "loss": 0.9788,
+      "step": 1347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.408574820392364e-05,
+      "loss": 0.9917,
+      "step": 1348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4027088842471685e-05,
+      "loss": 1.0457,
+      "step": 1349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.39684239015393e-05,
+      "loss": 0.9905,
+      "step": 1350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390975346240602e-05,
+      "loss": 1.0327,
+      "step": 1351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3851077606359e-05,
+      "loss": 0.9817,
+      "step": 1352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.379239641469289e-05,
+      "loss": 0.9788,
+      "step": 1353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.373370996870972e-05,
+      "loss": 0.9839,
+      "step": 1354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.367501834971882e-05,
+      "loss": 0.9338,
+      "step": 1355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.361632163903668e-05,
+      "loss": 0.9783,
+      "step": 1356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355761991798688e-05,
+      "loss": 0.9978,
+      "step": 1357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3498913267899864e-05,
+      "loss": 1.0127,
+      "step": 1358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3440201770112964e-05,
+      "loss": 0.9705,
+      "step": 1359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.338148550597023e-05,
+      "loss": 1.0305,
+      "step": 1360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3322764556822294e-05,
+      "loss": 0.9092,
+      "step": 1361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.326403900402627e-05,
+      "loss": 0.9858,
+      "step": 1362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3205308928945674e-05,
+      "loss": 0.9827,
+      "step": 1363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.314657441295028e-05,
+      "loss": 0.9429,
+      "step": 1364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.308783553741602e-05,
+      "loss": 0.9922,
+      "step": 1365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3029092383724846e-05,
+      "loss": 0.9561,
+      "step": 1366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.297034503326466e-05,
+      "loss": 1.1038,
+      "step": 1367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.291159356742917e-05,
+      "loss": 0.9575,
+      "step": 1368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.285283806761778e-05,
+      "loss": 1.0176,
+      "step": 1369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27940786152355e-05,
+      "loss": 0.9998,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27353152916928e-05,
+      "loss": 0.9736,
+      "step": 1371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2676548178405516e-05,
+      "loss": 0.959,
+      "step": 1372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.261777735679472e-05,
+      "loss": 0.9126,
+      "step": 1373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255900290828666e-05,
+      "loss": 0.9849,
+      "step": 1374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250022491431259e-05,
+      "loss": 1.0059,
+      "step": 1375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2441443456308656e-05,
+      "loss": 1.0315,
+      "step": 1376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.238265861571584e-05,
+      "loss": 1.0032,
+      "step": 1377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.232387047397978e-05,
+      "loss": 0.9675,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.226507911255071e-05,
+      "loss": 0.9277,
+      "step": 1379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2206284612883304e-05,
+      "loss": 0.939,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2147487056436585e-05,
+      "loss": 0.8936,
+      "step": 1381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.208868652467385e-05,
+      "loss": 0.9995,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.202988309906246e-05,
+      "loss": 1.0063,
+      "step": 1383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.197107686107382e-05,
+      "loss": 0.9658,
+      "step": 1384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.191226789218324e-05,
+      "loss": 1.0173,
+      "step": 1385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1853456273869794e-05,
+      "loss": 0.9802,
+      "step": 1386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1794642087616216e-05,
+      "loss": 1.0701,
+      "step": 1387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.173582541490886e-05,
+      "loss": 0.9373,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167700633723742e-05,
+      "loss": 0.9685,
+      "step": 1389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1618184936094995e-05,
+      "loss": 0.9595,
+      "step": 1390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1559361292977915e-05,
+      "loss": 0.876,
+      "step": 1391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150053548938557e-05,
+      "loss": 0.9673,
+      "step": 1392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1441707606820365e-05,
+      "loss": 0.9053,
+      "step": 1393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.138287772678759e-05,
+      "loss": 0.9897,
+      "step": 1394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1324045930795305e-05,
+      "loss": 1.0269,
+      "step": 1395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12652123003542e-05,
+      "loss": 0.9775,
+      "step": 1396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1206376916977526e-05,
+      "loss": 1.0083,
+      "step": 1397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114753986218095e-05,
+      "loss": 0.9744,
+      "step": 1398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.108870121748248e-05,
+      "loss": 0.9807,
+      "step": 1399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.102986106440232e-05,
+      "loss": 0.9858,
+      "step": 1400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.097101948446271e-05,
+      "loss": 0.9346,
+      "step": 1401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.091217655918796e-05,
+      "loss": 0.99,
+      "step": 1402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085333237010418e-05,
+      "loss": 0.9905,
+      "step": 1403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.079448699873923e-05,
+      "loss": 1.0493,
+      "step": 1404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.073564052662265e-05,
+      "loss": 0.947,
+      "step": 1405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.067679303528545e-05,
+      "loss": 0.9341,
+      "step": 1406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0617944606260114e-05,
+      "loss": 0.981,
+      "step": 1407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0559095321080376e-05,
+      "loss": 0.9731,
+      "step": 1408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0500245261281175e-05,
+      "loss": 0.9792,
+      "step": 1409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.044139450839851e-05,
+      "loss": 1.0535,
+      "step": 1410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.038254314396936e-05,
+      "loss": 1.009,
+      "step": 1411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032369124953156e-05,
+      "loss": 1.0173,
+      "step": 1412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0264838906623624e-05,
+      "loss": 1.0034,
+      "step": 1413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.020598619678477e-05,
+      "loss": 0.9446,
+      "step": 1414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0147133201554644e-05,
+      "loss": 0.9893,
+      "step": 1415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.008828000247334e-05,
+      "loss": 1.0093,
+      "step": 1416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.002942668108121e-05,
+      "loss": 0.938,
+      "step": 1417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.99705733189188e-05,
+      "loss": 0.9548,
+      "step": 1418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.991171999752668e-05,
+      "loss": 0.9602,
+      "step": 1419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.985286679844537e-05,
+      "loss": 0.9631,
+      "step": 1420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.979401380321524e-05,
+      "loss": 0.948,
+      "step": 1421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.973516109337639e-05,
+      "loss": 0.9539,
+      "step": 1422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.967630875046845e-05,
+      "loss": 0.999,
+      "step": 1423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.961745685603065e-05,
+      "loss": 0.9141,
+      "step": 1424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.955860549160151e-05,
+      "loss": 0.9602,
+      "step": 1425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.949975473871884e-05,
+      "loss": 1.0405,
+      "step": 1426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.944090467891963e-05,
+      "loss": 0.9487,
+      "step": 1427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.938205539373989e-05,
+      "loss": 0.9578,
+      "step": 1428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9323206964714554e-05,
+      "loss": 0.9185,
+      "step": 1429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9264359473377356e-05,
+      "loss": 0.9199,
+      "step": 1430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9205513001260775e-05,
+      "loss": 0.9712,
+      "step": 1431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914666762989583e-05,
+      "loss": 0.9585,
+      "step": 1432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.908782344081204e-05,
+      "loss": 1.0667,
+      "step": 1433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9028980515537286e-05,
+      "loss": 1.001,
+      "step": 1434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.897013893559771e-05,
+      "loss": 0.9915,
+      "step": 1435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.891129878251754e-05,
+      "loss": 0.9685,
+      "step": 1436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885246013781906e-05,
+      "loss": 1.0208,
+      "step": 1437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.879362308302249e-05,
+      "loss": 1.0076,
+      "step": 1438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8734787699645824e-05,
+      "loss": 0.9009,
+      "step": 1439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8675954069204706e-05,
+      "loss": 1.0037,
+      "step": 1440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8617122273212414e-05,
+      "loss": 1.0344,
+      "step": 1441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.855829239317965e-05,
+      "loss": 0.9519,
+      "step": 1442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849946451061443e-05,
+      "loss": 0.9697,
+      "step": 1443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8440638707022096e-05,
+      "loss": 0.9763,
+      "step": 1444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.838181506390501e-05,
+      "loss": 1.0015,
+      "step": 1445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8322993662762595e-05,
+      "loss": 1.0134,
+      "step": 1446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.826417458509115e-05,
+      "loss": 1.0403,
+      "step": 1447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820535791238378e-05,
+      "loss": 0.9905,
+      "step": 1448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.81465437261302e-05,
+      "loss": 1.0276,
+      "step": 1449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8087732107816755e-05,
+      "loss": 0.9644,
+      "step": 1450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8028923138926176e-05,
+      "loss": 0.9668,
+      "step": 1451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797011690093756e-05,
+      "loss": 0.9192,
+      "step": 1452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.791131347532618e-05,
+      "loss": 1.0217,
+      "step": 1453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7852512943563427e-05,
+      "loss": 1.031,
+      "step": 1454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779371538711672e-05,
+      "loss": 0.9336,
+      "step": 1455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.773492088744931e-05,
+      "loss": 1.0173,
+      "step": 1456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.767612952602023e-05,
+      "loss": 1.0505,
+      "step": 1457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.761734138428417e-05,
+      "loss": 0.9944,
+      "step": 1458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7558556543691355e-05,
+      "loss": 0.8896,
+      "step": 1459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749977508568742e-05,
+      "loss": 0.9773,
+      "step": 1460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7440997091713347e-05,
+      "loss": 1.0146,
+      "step": 1461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738222264320529e-05,
+      "loss": 1.0034,
+      "step": 1462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73234518215945e-05,
+      "loss": 0.9714,
+      "step": 1463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.726468470830721e-05,
+      "loss": 0.9482,
+      "step": 1464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7205921384764505e-05,
+      "loss": 0.925,
+      "step": 1465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.714716193238221e-05,
+      "loss": 0.957,
+      "step": 1466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7088406432570833e-05,
+      "loss": 0.9326,
+      "step": 1467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.702965496673534e-05,
+      "loss": 0.8945,
+      "step": 1468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.697090761627515e-05,
+      "loss": 0.9575,
+      "step": 1469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.691216446258401e-05,
+      "loss": 0.9465,
+      "step": 1470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.685342558704973e-05,
+      "loss": 0.9514,
+      "step": 1471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6794691071054345e-05,
+      "loss": 0.9929,
+      "step": 1472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6735960995973755e-05,
+      "loss": 1.009,
+      "step": 1473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6677235443177724e-05,
+      "loss": 0.9651,
+      "step": 1474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661851449402978e-05,
+      "loss": 0.9707,
+      "step": 1475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.655979822988705e-05,
+      "loss": 1.0022,
+      "step": 1476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.650108673210015e-05,
+      "loss": 0.9592,
+      "step": 1477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644238008201314e-05,
+      "loss": 0.9436,
+      "step": 1478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.638367836096332e-05,
+      "loss": 0.9724,
+      "step": 1479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.632498165028118e-05,
+      "loss": 0.9238,
+      "step": 1480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.626629003129029e-05,
+      "loss": 1.0183,
+      "step": 1481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.620760358530712e-05,
+      "loss": 0.9092,
+      "step": 1482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6148922393641e-05,
+      "loss": 0.9143,
+      "step": 1483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609024653759397e-05,
+      "loss": 1.0242,
+      "step": 1484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60315760984607e-05,
+      "loss": 0.9648,
+      "step": 1485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.597291115752831e-05,
+      "loss": 0.9795,
+      "step": 1486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.591425179607639e-05,
+      "loss": 1.0222,
+      "step": 1487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.585559809537666e-05,
+      "loss": 0.9167,
+      "step": 1488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5796950136693125e-05,
+      "loss": 1.0835,
+      "step": 1489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5738308001281775e-05,
+      "loss": 1.0503,
+      "step": 1490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5679671770390536e-05,
+      "loss": 0.9592,
+      "step": 1491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562104152525918e-05,
+      "loss": 0.949,
+      "step": 1492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.556241734711916e-05,
+      "loss": 0.9766,
+      "step": 1493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.550379931719351e-05,
+      "loss": 0.9846,
+      "step": 1494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5445187516696794e-05,
+      "loss": 0.978,
+      "step": 1495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5386582026834906e-05,
+      "loss": 0.9824,
+      "step": 1496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5327982928804994e-05,
+      "loss": 0.9114,
+      "step": 1497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.52693903037954e-05,
+      "loss": 0.9902,
+      "step": 1498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5210804232985426e-05,
+      "loss": 1.0212,
+      "step": 1499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5152224797545335e-05,
+      "loss": 0.9905,
+      "step": 1500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.50936520786362e-05,
+      "loss": 0.9014,
+      "step": 1501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5035086157409784e-05,
+      "loss": 0.9587,
+      "step": 1502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4976527115008405e-05,
+      "loss": 0.9768,
+      "step": 1503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.491797503256492e-05,
+      "loss": 0.9761,
+      "step": 1504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4859429991202426e-05,
+      "loss": 0.9951,
+      "step": 1505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.480089207203438e-05,
+      "loss": 0.8889,
+      "step": 1506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.474236135616432e-05,
+      "loss": 0.9744,
+      "step": 1507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468383792468578e-05,
+      "loss": 1.0518,
+      "step": 1508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462532185868227e-05,
+      "loss": 0.9746,
+      "step": 1509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4566813239227045e-05,
+      "loss": 0.9473,
+      "step": 1510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4508312147383036e-05,
+      "loss": 0.9932,
+      "step": 1511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.444981866420278e-05,
+      "loss": 0.9619,
+      "step": 1512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.439133287072826e-05,
+      "loss": 0.9609,
+      "step": 1513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.433285484799078e-05,
+      "loss": 0.9719,
+      "step": 1514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.427438467701091e-05,
+      "loss": 1.0051,
+      "step": 1515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.421592243879833e-05,
+      "loss": 0.9583,
+      "step": 1516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4157468214351714e-05,
+      "loss": 1.0156,
+      "step": 1517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.409902208465867e-05,
+      "loss": 0.9351,
+      "step": 1518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.404058413069556e-05,
+      "loss": 1.0369,
+      "step": 1519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.398215443342741e-05,
+      "loss": 0.9883,
+      "step": 1520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.392373307380787e-05,
+      "loss": 0.958,
+      "step": 1521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.386532013277892e-05,
+      "loss": 1.0703,
+      "step": 1522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3806915691271e-05,
+      "loss": 0.9468,
+      "step": 1523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.37485198302027e-05,
+      "loss": 0.927,
+      "step": 1524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.369013263048076e-05,
+      "loss": 1.0127,
+      "step": 1525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.363175417299989e-05,
+      "loss": 0.9148,
+      "step": 1526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3573384538642706e-05,
+      "loss": 0.9756,
+      "step": 1527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3515023808279586e-05,
+      "loss": 0.9998,
+      "step": 1528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345667206276861e-05,
+      "loss": 0.9475,
+      "step": 1529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3398329382955346e-05,
+      "loss": 1.0427,
+      "step": 1530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.333999584967284e-05,
+      "loss": 0.9277,
+      "step": 1531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3281671543741474e-05,
+      "loss": 0.9614,
+      "step": 1532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.322335654596884e-05,
+      "loss": 0.9866,
+      "step": 1533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.31650509371496e-05,
+      "loss": 0.9475,
+      "step": 1534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3106754798065464e-05,
+      "loss": 0.9695,
+      "step": 1535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.304846820948497e-05,
+      "loss": 1.0376,
+      "step": 1536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2990191252163446e-05,
+      "loss": 1.0359,
+      "step": 1537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2931924006842895e-05,
+      "loss": 0.9902,
+      "step": 1538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2873666554251845e-05,
+      "loss": 1.0046,
+      "step": 1539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2815418975105236e-05,
+      "loss": 0.9614,
+      "step": 1540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.275718135010435e-05,
+      "loss": 0.9146,
+      "step": 1541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269895375993668e-05,
+      "loss": 0.9768,
+      "step": 1542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264073628527583e-05,
+      "loss": 0.9666,
+      "step": 1543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.258252900678135e-05,
+      "loss": 0.9866,
+      "step": 1544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.252433200509869e-05,
+      "loss": 0.8914,
+      "step": 1545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.246614536085907e-05,
+      "loss": 0.9512,
+      "step": 1546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.240796915467933e-05,
+      "loss": 1.0046,
+      "step": 1547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2349803467161866e-05,
+      "loss": 0.9373,
+      "step": 1548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.229164837889451e-05,
+      "loss": 0.9839,
+      "step": 1549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2233503970450394e-05,
+      "loss": 0.929,
+      "step": 1550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.217537032238784e-05,
+      "loss": 0.9932,
+      "step": 1551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2117247515250304e-05,
+      "loss": 0.9011,
+      "step": 1552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205913562956619e-05,
+      "loss": 0.9514,
+      "step": 1553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.200103474584876e-05,
+      "loss": 1.0979,
+      "step": 1554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1942944944596074e-05,
+      "loss": 0.9934,
+      "step": 1555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1884866306290823e-05,
+      "loss": 0.9905,
+      "step": 1556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1826798911400186e-05,
+      "loss": 1.0105,
+      "step": 1557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.176874284037581e-05,
+      "loss": 0.8647,
+      "step": 1558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.171069817365365e-05,
+      "loss": 1.0178,
+      "step": 1559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1652664991653865e-05,
+      "loss": 1.0125,
+      "step": 1560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1594643374780686e-05,
+      "loss": 0.9946,
+      "step": 1561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1536633403422306e-05,
+      "loss": 1.0017,
+      "step": 1562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.147863515795083e-05,
+      "loss": 0.9009,
+      "step": 1563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1420648718722085e-05,
+      "loss": 0.9756,
+      "step": 1564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.136267416607552e-05,
+      "loss": 1.0215,
+      "step": 1565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.130471158033418e-05,
+      "loss": 0.9155,
+      "step": 1566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.124676104180447e-05,
+      "loss": 0.9258,
+      "step": 1567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1188822630776127e-05,
+      "loss": 0.9834,
+      "step": 1568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1130896427522084e-05,
+      "loss": 0.9678,
+      "step": 1569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.107298251229836e-05,
+      "loss": 0.9363,
+      "step": 1570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1015080965343946e-05,
+      "loss": 0.9348,
+      "step": 1571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.09571918668807e-05,
+      "loss": 0.9573,
+      "step": 1572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0899315297113257e-05,
+      "loss": 0.907,
+      "step": 1573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.084145133622883e-05,
+      "loss": 0.9104,
+      "step": 1574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.078360006439722e-05,
+      "loss": 0.9304,
+      "step": 1575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.072576156177062e-05,
+      "loss": 1.0037,
+      "step": 1576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0667935908483554e-05,
+      "loss": 1.0176,
+      "step": 1577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0610123184652724e-05,
+      "loss": 1.0334,
+      "step": 1578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0552323470376916e-05,
+      "loss": 0.9492,
+      "step": 1579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049453684573693e-05,
+      "loss": 0.999,
+      "step": 1580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0436763390795365e-05,
+      "loss": 1.0286,
+      "step": 1581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.037900318559661e-05,
+      "loss": 0.9246,
+      "step": 1582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0321256310166724e-05,
+      "loss": 0.9348,
+      "step": 1583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026352284451325e-05,
+      "loss": 0.9348,
+      "step": 1584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0205802868625165e-05,
+      "loss": 1.0017,
+      "step": 1585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0148096462472784e-05,
+      "loss": 0.9709,
+      "step": 1586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0090403706007586e-05,
+      "loss": 1.0349,
+      "step": 1587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.003272467916214e-05,
+      "loss": 1.0061,
+      "step": 1588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.997505946185003e-05,
+      "loss": 0.9587,
+      "step": 1589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991740813396568e-05,
+      "loss": 0.9963,
+      "step": 1590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.985977077538426e-05,
+      "loss": 0.926,
+      "step": 1591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.980214746596159e-05,
+      "loss": 0.9033,
+      "step": 1592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.974453828553404e-05,
+      "loss": 0.9343,
+      "step": 1593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9686943313918405e-05,
+      "loss": 0.9187,
+      "step": 1594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9629362630911786e-05,
+      "loss": 0.9521,
+      "step": 1595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9571796316291476e-05,
+      "loss": 1.0095,
+      "step": 1596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9514244449814884e-05,
+      "loss": 0.9927,
+      "step": 1597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9456707111219394e-05,
+      "loss": 0.9458,
+      "step": 1598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.939918438022224e-05,
+      "loss": 1.0122,
+      "step": 1599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.934167633652045e-05,
+      "loss": 0.9695,
+      "step": 1600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9284183059790686e-05,
+      "loss": 0.9858,
+      "step": 1601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.922670462968914e-05,
+      "loss": 0.9768,
+      "step": 1602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.916924112585146e-05,
+      "loss": 1.0291,
+      "step": 1603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.911179262789261e-05,
+      "loss": 0.9536,
+      "step": 1604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9054359215406724e-05,
+      "loss": 0.9536,
+      "step": 1605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899694096796709e-05,
+      "loss": 0.9648,
+      "step": 1606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893953796512596e-05,
+      "loss": 1.0332,
+      "step": 1607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8882150286414455e-05,
+      "loss": 0.9822,
+      "step": 1608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.882477801134247e-05,
+      "loss": 0.9407,
+      "step": 1609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.876742121939857e-05,
+      "loss": 0.9048,
+      "step": 1610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871007999004986e-05,
+      "loss": 0.9634,
+      "step": 1611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86527544027419e-05,
+      "loss": 1.032,
+      "step": 1612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.859544453689853e-05,
+      "loss": 1.0137,
+      "step": 1613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.853815047192187e-05,
+      "loss": 0.9744,
+      "step": 1614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8480872287192114e-05,
+      "loss": 0.9829,
+      "step": 1615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.842361006206745e-05,
+      "loss": 0.9133,
+      "step": 1616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.836636387588397e-05,
+      "loss": 0.9438,
+      "step": 1617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8309133807955535e-05,
+      "loss": 0.9751,
+      "step": 1618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.825191993757368e-05,
+      "loss": 0.9231,
+      "step": 1619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8194722344007485e-05,
+      "loss": 0.9138,
+      "step": 1620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8137541106503515e-05,
+      "loss": 0.9702,
+      "step": 1621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.808037630428561e-05,
+      "loss": 1.0002,
+      "step": 1622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8023228016554915e-05,
+      "loss": 0.9949,
+      "step": 1623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.796609632248964e-05,
+      "loss": 0.9482,
+      "step": 1624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7908981301245025e-05,
+      "loss": 0.9924,
+      "step": 1625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7851883031953196e-05,
+      "loss": 0.9451,
+      "step": 1626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.779480159372307e-05,
+      "loss": 0.9346,
+      "step": 1627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.773773706564029e-05,
+      "loss": 0.9619,
+      "step": 1628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.768068952676701e-05,
+      "loss": 0.9297,
+      "step": 1629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.762365905614187e-05,
+      "loss": 0.9592,
+      "step": 1630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.756664573277987e-05,
+      "loss": 0.9495,
+      "step": 1631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.750964963567225e-05,
+      "loss": 1.0181,
+      "step": 1632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.745267084378636e-05,
+      "loss": 0.9749,
+      "step": 1633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7395709436065615e-05,
+      "loss": 0.9614,
+      "step": 1634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7338765491429305e-05,
+      "loss": 0.991,
+      "step": 1635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.728183908877254e-05,
+      "loss": 0.9553,
+      "step": 1636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.722493030696614e-05,
+      "loss": 0.8525,
+      "step": 1637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7168039224856504e-05,
+      "loss": 0.9343,
+      "step": 1638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.711116592126548e-05,
+      "loss": 0.9763,
+      "step": 1639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.705431047499033e-05,
+      "loss": 0.9648,
+      "step": 1640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6997472964803544e-05,
+      "loss": 1.0173,
+      "step": 1641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.694065346945278e-05,
+      "loss": 0.9209,
+      "step": 1642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68838520676607e-05,
+      "loss": 0.9375,
+      "step": 1643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6827068838124936e-05,
+      "loss": 0.9482,
+      "step": 1644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.677030385951795e-05,
+      "loss": 0.9795,
+      "step": 1645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.671355721048687e-05,
+      "loss": 0.8928,
+      "step": 1646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.665682896965349e-05,
+      "loss": 0.9924,
+      "step": 1647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.660011921561405e-05,
+      "loss": 0.9539,
+      "step": 1648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.654342802693918e-05,
+      "loss": 0.9219,
+      "step": 1649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.648675548217381e-05,
+      "loss": 0.9351,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.643010165983705e-05,
+      "loss": 0.9529,
+      "step": 1651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.637346663842204e-05,
+      "loss": 1.0144,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.631685049639586e-05,
+      "loss": 1.001,
+      "step": 1653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6260253312199486e-05,
+      "loss": 0.9507,
+      "step": 1654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.620367516424759e-05,
+      "loss": 0.9651,
+      "step": 1655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.614711613092846e-05,
+      "loss": 0.9443,
+      "step": 1656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.609057629060394e-05,
+      "loss": 0.9736,
+      "step": 1657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.603405572160925e-05,
+      "loss": 1.0581,
+      "step": 1658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.597755450225294e-05,
+      "loss": 0.9304,
+      "step": 1659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5921072710816694e-05,
+      "loss": 0.9341,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5864610425555346e-05,
+      "loss": 1.0039,
+      "step": 1661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.580816772469665e-05,
+      "loss": 0.9807,
+      "step": 1662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.575174468644127e-05,
+      "loss": 0.9407,
+      "step": 1663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5695341388962614e-05,
+      "loss": 0.9846,
+      "step": 1664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5638957910406724e-05,
+      "loss": 0.9509,
+      "step": 1665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5582594328892184e-05,
+      "loss": 0.9463,
+      "step": 1666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.552625072251004e-05,
+      "loss": 0.9678,
+      "step": 1667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.546992716932364e-05,
+      "loss": 0.9634,
+      "step": 1668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5413623747368515e-05,
+      "loss": 0.9304,
+      "step": 1669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.53573405346524e-05,
+      "loss": 1.021,
+      "step": 1670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.530107760915493e-05,
+      "loss": 0.9426,
+      "step": 1671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.524483504882768e-05,
+      "loss": 1.0061,
+      "step": 1672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.518861293159401e-05,
+      "loss": 1.0449,
+      "step": 1673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.513241133534895e-05,
+      "loss": 0.9578,
+      "step": 1674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5076230337959095e-05,
+      "loss": 0.9292,
+      "step": 1675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5020070017262516e-05,
+      "loss": 0.9253,
+      "step": 1676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.496393045106858e-05,
+      "loss": 0.8911,
+      "step": 1677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.490781171715799e-05,
+      "loss": 0.9739,
+      "step": 1678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4851713893282524e-05,
+      "loss": 0.9917,
+      "step": 1679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4795637057164984e-05,
+      "loss": 0.9121,
+      "step": 1680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.473958128649915e-05,
+      "loss": 1.0222,
+      "step": 1681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.468354665894955e-05,
+      "loss": 0.9087,
+      "step": 1682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.462753325215146e-05,
+      "loss": 1.0151,
+      "step": 1683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.457154114371076e-05,
+      "loss": 0.9556,
+      "step": 1684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.451557041120378e-05,
+      "loss": 0.9458,
+      "step": 1685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445962113217726e-05,
+      "loss": 1.0088,
+      "step": 1686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.440369338414823e-05,
+      "loss": 0.9631,
+      "step": 1687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.434778724460387e-05,
+      "loss": 0.9934,
+      "step": 1688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.429190279100141e-05,
+      "loss": 0.9457,
+      "step": 1689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.423604010076807e-05,
+      "loss": 0.9783,
+      "step": 1690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.41801992513009e-05,
+      "loss": 0.9231,
+      "step": 1691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.412438031996666e-05,
+      "loss": 0.9661,
+      "step": 1692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.406858338410181e-05,
+      "loss": 0.9587,
+      "step": 1693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.401280852101228e-05,
+      "loss": 0.9128,
+      "step": 1694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3957055807973416e-05,
+      "loss": 0.9163,
+      "step": 1695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.390132532222991e-05,
+      "loss": 0.9055,
+      "step": 1696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.384561714099563e-05,
+      "loss": 0.9922,
+      "step": 1697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.378993134145356e-05,
+      "loss": 0.9905,
+      "step": 1698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.373426800075569e-05,
+      "loss": 0.991,
+      "step": 1699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3678627196022825e-05,
+      "loss": 0.9944,
+      "step": 1700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.362300900434463e-05,
+      "loss": 0.9485,
+      "step": 1701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.35674135027794e-05,
+      "loss": 0.9116,
+      "step": 1702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.351184076835398e-05,
+      "loss": 1.0125,
+      "step": 1703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.345629087806369e-05,
+      "loss": 0.9248,
+      "step": 1704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.340076390887221e-05,
+      "loss": 1.0068,
+      "step": 1705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.334525993771144e-05,
+      "loss": 0.9941,
+      "step": 1706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.328977904148143e-05,
+      "loss": 0.9541,
+      "step": 1707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3234321297050266e-05,
+      "loss": 0.9246,
+      "step": 1708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3178886781253924e-05,
+      "loss": 0.9114,
+      "step": 1709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.312347557089624e-05,
+      "loss": 1.0898,
+      "step": 1710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.306808774274876e-05,
+      "loss": 1.0337,
+      "step": 1711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.301272337355058e-05,
+      "loss": 0.9573,
+      "step": 1712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.295738254000835e-05,
+      "loss": 0.9968,
+      "step": 1713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.290206531879607e-05,
+      "loss": 0.9473,
+      "step": 1714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.284677178655507e-05,
+      "loss": 0.9993,
+      "step": 1715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.279150201989384e-05,
+      "loss": 1.0227,
+      "step": 1716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.273625609538791e-05,
+      "loss": 0.9971,
+      "step": 1717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.268103408957984e-05,
+      "loss": 0.9878,
+      "step": 1718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.262583607897901e-05,
+      "loss": 0.9302,
+      "step": 1719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.257066214006154e-05,
+      "loss": 1.0002,
+      "step": 1720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.251551234927025e-05,
+      "loss": 0.9951,
+      "step": 1721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.246038678301446e-05,
+      "loss": 0.9858,
+      "step": 1722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.240528551766994e-05,
+      "loss": 0.9456,
+      "step": 1723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2350208629578796e-05,
+      "loss": 0.9827,
+      "step": 1724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.229515619504936e-05,
+      "loss": 0.8923,
+      "step": 1725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.224012829035607e-05,
+      "loss": 1.0071,
+      "step": 1726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.218512499173941e-05,
+      "loss": 0.9543,
+      "step": 1727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2130146375405746e-05,
+      "loss": 0.9756,
+      "step": 1728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.207519251752723e-05,
+      "loss": 0.9546,
+      "step": 1729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2020263494241754e-05,
+      "loss": 0.9033,
+      "step": 1730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.196535938165276e-05,
+      "loss": 0.9392,
+      "step": 1731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1910480255829237e-05,
+      "loss": 0.9197,
+      "step": 1732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1855626192805487e-05,
+      "loss": 0.95,
+      "step": 1733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.180079726858111e-05,
+      "loss": 0.9163,
+      "step": 1734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.174599355912092e-05,
+      "loss": 0.9768,
+      "step": 1735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.169121514035473e-05,
+      "loss": 0.9365,
+      "step": 1736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.163646208817735e-05,
+      "loss": 0.978,
+      "step": 1737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.158173447844845e-05,
+      "loss": 0.9683,
+      "step": 1738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.152703238699242e-05,
+      "loss": 1.0005,
+      "step": 1739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.147235588959831e-05,
+      "loss": 0.9998,
+      "step": 1740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.141770506201974e-05,
+      "loss": 0.9758,
+      "step": 1741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.136307997997472e-05,
+      "loss": 0.9587,
+      "step": 1742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1308480719145594e-05,
+      "loss": 1.0134,
+      "step": 1743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.125390735517898e-05,
+      "loss": 0.9685,
+      "step": 1744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.119935996368556e-05,
+      "loss": 1.0496,
+      "step": 1745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1144838620240036e-05,
+      "loss": 0.9109,
+      "step": 1746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.109034340038106e-05,
+      "loss": 0.9937,
+      "step": 1747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1035874379611044e-05,
+      "loss": 0.9849,
+      "step": 1748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.098143163339615e-05,
+      "loss": 1.0242,
+      "step": 1749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.09270152371661e-05,
+      "loss": 0.9485,
+      "step": 1750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0872625266314106e-05,
+      "loss": 0.9741,
+      "step": 1751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0818261796196804e-05,
+      "loss": 1.0454,
+      "step": 1752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.076392490213409e-05,
+      "loss": 0.9988,
+      "step": 1753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.070961465940902e-05,
+      "loss": 0.9578,
+      "step": 1754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.065533114326776e-05,
+      "loss": 1.0459,
+      "step": 1755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.060107442891943e-05,
+      "loss": 0.9397,
+      "step": 1756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0546844591536013e-05,
+      "loss": 0.9431,
+      "step": 1757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.049264170625227e-05,
+      "loss": 1.0098,
+      "step": 1758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.043846584816561e-05,
+      "loss": 0.9265,
+      "step": 1759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.038431709233598e-05,
+      "loss": 0.9907,
+      "step": 1760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0330195513785808e-05,
+      "loss": 0.9573,
+      "step": 1761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0276101187499863e-05,
+      "loss": 1.0115,
+      "step": 1762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0222034188425118e-05,
+      "loss": 0.989,
+      "step": 1763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0167994591470737e-05,
+      "loss": 0.9512,
+      "step": 1764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0113982471507874e-05,
+      "loss": 0.8838,
+      "step": 1765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0059997903369656e-05,
+      "loss": 0.9434,
+      "step": 1766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.000604096185101e-05,
+      "loss": 0.9727,
+      "step": 1767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9952111721708576e-05,
+      "loss": 0.9836,
+      "step": 1768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9898210257660665e-05,
+      "loss": 1.0203,
+      "step": 1769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.984433664438704e-05,
+      "loss": 0.9243,
+      "step": 1770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9790490956528915e-05,
+      "loss": 0.9861,
+      "step": 1771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.973667326868882e-05,
+      "loss": 0.9707,
+      "step": 1772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9682883655430466e-05,
+      "loss": 0.9236,
+      "step": 1773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9629122191278675e-05,
+      "loss": 0.9839,
+      "step": 1774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9575388950719286e-05,
+      "loss": 0.9028,
+      "step": 1775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9521684008199013e-05,
+      "loss": 0.9524,
+      "step": 1776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.946800743812537e-05,
+      "loss": 0.9209,
+      "step": 1777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9414359314866557e-05,
+      "loss": 0.9011,
+      "step": 1778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9360739712751396e-05,
+      "loss": 0.9353,
+      "step": 1779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9307148706069142e-05,
+      "loss": 0.8689,
+      "step": 1780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9253586369069445e-05,
+      "loss": 0.999,
+      "step": 1781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920005277596225e-05,
+      "loss": 0.9077,
+      "step": 1782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.914654800091768e-05,
+      "loss": 0.9458,
+      "step": 1783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.90930721180659e-05,
+      "loss": 0.9038,
+      "step": 1784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9039625201497108e-05,
+      "loss": 0.9551,
+      "step": 1785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.898620732526127e-05,
+      "loss": 0.9548,
+      "step": 1786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8932818563368218e-05,
+      "loss": 0.9617,
+      "step": 1787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8879458989787406e-05,
+      "loss": 0.9785,
+      "step": 1788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8826128678447805e-05,
+      "loss": 0.9597,
+      "step": 1789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8772827703237914e-05,
+      "loss": 0.9766,
+      "step": 1790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8719556138005572e-05,
+      "loss": 0.9768,
+      "step": 1791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8666314056557815e-05,
+      "loss": 0.8738,
+      "step": 1792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8613101532660892e-05,
+      "loss": 0.9404,
+      "step": 1793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8559918640040094e-05,
+      "loss": 0.9819,
+      "step": 1794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8506765452379603e-05,
+      "loss": 1.0061,
+      "step": 1795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8453642043322516e-05,
+      "loss": 0.9546,
+      "step": 1796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8400548486470657e-05,
+      "loss": 1.0146,
+      "step": 1797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8347484855384444e-05,
+      "loss": 0.9902,
+      "step": 1798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.829445122358285e-05,
+      "loss": 0.9497,
+      "step": 1799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.824144766454333e-05,
+      "loss": 0.9241,
+      "step": 1800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8188474251701646e-05,
+      "loss": 0.9099,
+      "step": 1801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8135531058451748e-05,
+      "loss": 0.939,
+      "step": 1802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.808261815814579e-05,
+      "loss": 0.9319,
+      "step": 1803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8029735624093933e-05,
+      "loss": 0.9021,
+      "step": 1804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7976883529564223e-05,
+      "loss": 0.9443,
+      "step": 1805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7924061947782576e-05,
+      "loss": 0.9641,
+      "step": 1806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7871270951932655e-05,
+      "loss": 0.9436,
+      "step": 1807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7818510615155667e-05,
+      "loss": 0.9619,
+      "step": 1808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7765781010550408e-05,
+      "loss": 0.9814,
+      "step": 1809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.771308221117309e-05,
+      "loss": 0.9463,
+      "step": 1810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7660414290037202e-05,
+      "loss": 0.9424,
+      "step": 1811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7607777320113494e-05,
+      "loss": 1.0203,
+      "step": 1812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7555171374329837e-05,
+      "loss": 1.0229,
+      "step": 1813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7502596525571078e-05,
+      "loss": 0.9619,
+      "step": 1814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7450052846678988e-05,
+      "loss": 0.9434,
+      "step": 1815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7397540410452206e-05,
+      "loss": 0.897,
+      "step": 1816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7345059289646008e-05,
+      "loss": 0.9817,
+      "step": 1817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.729260955697233e-05,
+      "loss": 0.9409,
+      "step": 1818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7240191285099637e-05,
+      "loss": 0.9397,
+      "step": 1819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7187804546652738e-05,
+      "loss": 0.9519,
+      "step": 1820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.713544941421282e-05,
+      "loss": 0.9194,
+      "step": 1821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7083125960317268e-05,
+      "loss": 0.988,
+      "step": 1822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7030834257459515e-05,
+      "loss": 0.9746,
+      "step": 1823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6978574378089082e-05,
+      "loss": 1.0381,
+      "step": 1824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.692634639461138e-05,
+      "loss": 0.9546,
+      "step": 1825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.687415037938758e-05,
+      "loss": 0.9495,
+      "step": 1826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6821986404734623e-05,
+      "loss": 1.0378,
+      "step": 1827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6769854542925043e-05,
+      "loss": 1.0095,
+      "step": 1828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6717754866186843e-05,
+      "loss": 0.9756,
+      "step": 1829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.666568744670348e-05,
+      "loss": 0.928,
+      "step": 1830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6613652356613716e-05,
+      "loss": 0.9958,
+      "step": 1831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.656164966801149e-05,
+      "loss": 0.9785,
+      "step": 1832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6509679452945844e-05,
+      "loss": 0.9951,
+      "step": 1833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6457741783420886e-05,
+      "loss": 0.9248,
+      "step": 1834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6405836731395595e-05,
+      "loss": 1.0354,
+      "step": 1835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.635396436878374e-05,
+      "loss": 0.9287,
+      "step": 1836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6302124767453828e-05,
+      "loss": 0.8965,
+      "step": 1837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.625031799922899e-05,
+      "loss": 0.9934,
+      "step": 1838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6198544135886816e-05,
+      "loss": 0.9272,
+      "step": 1839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6146803249159336e-05,
+      "loss": 0.979,
+      "step": 1840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6095095410732927e-05,
+      "loss": 0.9517,
+      "step": 1841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6043420692248098e-05,
+      "loss": 0.9739,
+      "step": 1842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.599177916529954e-05,
+      "loss": 0.9507,
+      "step": 1843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5940170901435946e-05,
+      "loss": 0.9573,
+      "step": 1844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5888595972159864e-05,
+      "loss": 0.9124,
+      "step": 1845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5837054448927734e-05,
+      "loss": 0.9897,
+      "step": 1846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5785546403149697e-05,
+      "loss": 0.9465,
+      "step": 1847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5734071906189476e-05,
+      "loss": 0.9497,
+      "step": 1848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5682631029364313e-05,
+      "loss": 0.9695,
+      "step": 1849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5631223843944935e-05,
+      "loss": 0.9382,
+      "step": 1850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5579850421155293e-05,
+      "loss": 0.9622,
+      "step": 1851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5528510832172647e-05,
+      "loss": 0.9551,
+      "step": 1852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5477205148127347e-05,
+      "loss": 0.9536,
+      "step": 1853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5425933440102734e-05,
+      "loss": 0.9263,
+      "step": 1854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5374695779135138e-05,
+      "loss": 1.0081,
+      "step": 1855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.532349223621369e-05,
+      "loss": 0.9509,
+      "step": 1856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.527232288228022e-05,
+      "loss": 0.9385,
+      "step": 1857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.522118778822924e-05,
+      "loss": 0.8916,
+      "step": 1858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.517008702490778e-05,
+      "loss": 1.0051,
+      "step": 1859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5119020663115266e-05,
+      "loss": 1.0117,
+      "step": 1860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5067988773603524e-05,
+      "loss": 0.9558,
+      "step": 1861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5016991427076587e-05,
+      "loss": 0.9551,
+      "step": 1862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4966028694190607e-05,
+      "loss": 0.9739,
+      "step": 1863
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4915100645553812e-05,
+      "loss": 0.9285,
+      "step": 1864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4864207351726397e-05,
+      "loss": 0.9575,
+      "step": 1865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.481334888322035e-05,
+      "loss": 0.9021,
+      "step": 1866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4762525310499413e-05,
+      "loss": 1.0186,
+      "step": 1867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4711736703979018e-05,
+      "loss": 0.9753,
+      "step": 1868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4660983134026155e-05,
+      "loss": 0.948,
+      "step": 1869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.461026467095921e-05,
+      "loss": 0.9907,
+      "step": 1870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.455958138504799e-05,
+      "loss": 0.9404,
+      "step": 1871
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4508933346513564e-05,
+      "loss": 0.9519,
+      "step": 1872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.445832062552811e-05,
+      "loss": 0.9548,
+      "step": 1873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.440774329221492e-05,
+      "loss": 0.9958,
+      "step": 1874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4357201416648274e-05,
+      "loss": 0.9805,
+      "step": 1875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4306695068853257e-05,
+      "loss": 1.0171,
+      "step": 1876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4256224318805792e-05,
+      "loss": 1.0134,
+      "step": 1877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.420578923643248e-05,
+      "loss": 0.9849,
+      "step": 1878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4155389891610454e-05,
+      "loss": 0.8936,
+      "step": 1879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4105026354167377e-05,
+      "loss": 0.9814,
+      "step": 1880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4054698693881313e-05,
+      "loss": 0.8696,
+      "step": 1881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.400440698048056e-05,
+      "loss": 0.9653,
+      "step": 1882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3954151283643678e-05,
+      "loss": 0.9736,
+      "step": 1883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3903931672999293e-05,
+      "loss": 1.0088,
+      "step": 1884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3853748218126e-05,
+      "loss": 0.9385,
+      "step": 1885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3803600988552373e-05,
+      "loss": 0.9456,
+      "step": 1886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3753490053756765e-05,
+      "loss": 0.9236,
+      "step": 1887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.370341548316722e-05,
+      "loss": 0.9463,
+      "step": 1888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.365337734616142e-05,
+      "loss": 0.9587,
+      "step": 1889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.36033757120666e-05,
+      "loss": 0.906,
+      "step": 1890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3553410650159347e-05,
+      "loss": 0.9392,
+      "step": 1891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3503482229665635e-05,
+      "loss": 0.9045,
+      "step": 1892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3453590519760675e-05,
+      "loss": 0.9502,
+      "step": 1893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3403735589568772e-05,
+      "loss": 1.0007,
+      "step": 1894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.33539175081633e-05,
+      "loss": 0.9333,
+      "step": 1895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.33041363445666e-05,
+      "loss": 1.0337,
+      "step": 1896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3254392167749817e-05,
+      "loss": 0.926,
+      "step": 1897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3204685046632884e-05,
+      "loss": 0.9556,
+      "step": 1898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3155015050084406e-05,
+      "loss": 0.946,
+      "step": 1899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3105382246921513e-05,
+      "loss": 1.0215,
+      "step": 1900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3055786705909803e-05,
+      "loss": 0.9761,
+      "step": 1901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3006228495763295e-05,
+      "loss": 0.9888,
+      "step": 1902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.295670768514427e-05,
+      "loss": 0.9678,
+      "step": 1903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2907224342663146e-05,
+      "loss": 0.9497,
+      "step": 1904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2857778536878488e-05,
+      "loss": 1.0083,
+      "step": 1905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2808370336296842e-05,
+      "loss": 0.9883,
+      "step": 1906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2758999809372616e-05,
+      "loss": 0.9404,
+      "step": 1907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2709667024508058e-05,
+      "loss": 0.9492,
+      "step": 1908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2660372050053136e-05,
+      "loss": 0.9355,
+      "step": 1909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2611114954305373e-05,
+      "loss": 0.9656,
+      "step": 1910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.256189580550987e-05,
+      "loss": 0.9814,
+      "step": 1911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2512714671859148e-05,
+      "loss": 0.9668,
+      "step": 1912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2463571621493007e-05,
+      "loss": 0.958,
+      "step": 1913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.241446672249854e-05,
+      "loss": 0.9297,
+      "step": 1914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2365400042909972e-05,
+      "loss": 0.97,
+      "step": 1915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2316371650708534e-05,
+      "loss": 0.967,
+      "step": 1916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.226738161382248e-05,
+      "loss": 0.9663,
+      "step": 1917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2218430000126854e-05,
+      "loss": 0.9233,
+      "step": 1918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2169516877443485e-05,
+      "loss": 0.9971,
+      "step": 1919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.212064231354091e-05,
+      "loss": 0.9976,
+      "step": 1920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2071806376134207e-05,
+      "loss": 0.9817,
+      "step": 1921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2023009132884938e-05,
+      "loss": 0.95,
+      "step": 1922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.197425065140107e-05,
+      "loss": 0.9653,
+      "step": 1923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1925530999236878e-05,
+      "loss": 0.9438,
+      "step": 1924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1876850243892787e-05,
+      "loss": 0.9409,
+      "step": 1925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.182820845281538e-05,
+      "loss": 0.9478,
+      "step": 1926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1779605693397265e-05,
+      "loss": 0.957,
+      "step": 1927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.17310420329769e-05,
+      "loss": 1.0039,
+      "step": 1928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.168251753883865e-05,
+      "loss": 0.9551,
+      "step": 1929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1634032278212597e-05,
+      "loss": 0.9651,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.158558631827442e-05,
+      "loss": 0.9819,
+      "step": 1931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1537179726145395e-05,
+      "loss": 0.9697,
+      "step": 1932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1488812568892263e-05,
+      "loss": 0.9773,
+      "step": 1933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1440484913527065e-05,
+      "loss": 0.9373,
+      "step": 1934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1392196827007195e-05,
+      "loss": 0.959,
+      "step": 1935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1343948376235144e-05,
+      "loss": 0.918,
+      "step": 1936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1295739628058566e-05,
+      "loss": 0.9148,
+      "step": 1937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1247570649270027e-05,
+      "loss": 0.8958,
+      "step": 1938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.119944150660706e-05,
+      "loss": 0.9475,
+      "step": 1939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1151352266751994e-05,
+      "loss": 0.9885,
+      "step": 1940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1103302996331832e-05,
+      "loss": 0.9233,
+      "step": 1941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1055293761918237e-05,
+      "loss": 0.9482,
+      "step": 1942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1007324630027413e-05,
+      "loss": 0.9932,
+      "step": 1943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0959395667119945e-05,
+      "loss": 0.9219,
+      "step": 1944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.091150693960083e-05,
+      "loss": 1.001,
+      "step": 1945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0863658513819296e-05,
+      "loss": 1.0056,
+      "step": 1946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0815850456068704e-05,
+      "loss": 0.9805,
+      "step": 1947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0768082832586522e-05,
+      "loss": 0.918,
+      "step": 1948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0720355709554207e-05,
+      "loss": 0.8813,
+      "step": 1949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.067266915309704e-05,
+      "loss": 0.9636,
+      "step": 1950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0625023229284168e-05,
+      "loss": 0.9746,
+      "step": 1951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0577418004128442e-05,
+      "loss": 1.0017,
+      "step": 1952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.052985354358622e-05,
+      "loss": 0.9473,
+      "step": 1953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0482329913557503e-05,
+      "loss": 0.9934,
+      "step": 1954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0434847179885686e-05,
+      "loss": 1.0388,
+      "step": 1955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.038740540835746e-05,
+      "loss": 1.01,
+      "step": 1956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0340004664702826e-05,
+      "loss": 0.9998,
+      "step": 1957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.029264501459492e-05,
+      "loss": 0.9241,
+      "step": 1958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0245326523649895e-05,
+      "loss": 0.9375,
+      "step": 1959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0198049257426943e-05,
+      "loss": 0.9321,
+      "step": 1960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0150813281428132e-05,
+      "loss": 0.9604,
+      "step": 1961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0103618661098273e-05,
+      "loss": 1.061,
+      "step": 1962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.005646546182493e-05,
+      "loss": 0.9214,
+      "step": 1963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0009353748938275e-05,
+      "loss": 0.9441,
+      "step": 1964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.996228358771096e-05,
+      "loss": 0.9238,
+      "step": 1965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9915255043358105e-05,
+      "loss": 0.9055,
+      "step": 1966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9868268181037185e-05,
+      "loss": 0.9155,
+      "step": 1967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9821323065847863e-05,
+      "loss": 0.9475,
+      "step": 1968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9774419762832035e-05,
+      "loss": 0.9045,
+      "step": 1969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9727558336973595e-05,
+      "loss": 1.0103,
+      "step": 1970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9680738853198487e-05,
+      "loss": 0.9314,
+      "step": 1971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.963396137637448e-05,
+      "loss": 0.9487,
+      "step": 1972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.958722597131119e-05,
+      "loss": 0.9695,
+      "step": 1973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9540532702759944e-05,
+      "loss": 0.9297,
+      "step": 1974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9493881635413642e-05,
+      "loss": 0.9114,
+      "step": 1975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9447272833906748e-05,
+      "loss": 1.0278,
+      "step": 1976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9400706362815198e-05,
+      "loss": 0.9749,
+      "step": 1977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9354182286656203e-05,
+      "loss": 0.9275,
+      "step": 1978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93077006698883e-05,
+      "loss": 0.9565,
+      "step": 1979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9261261576911194e-05,
+      "loss": 0.9758,
+      "step": 1980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.921486507206562e-05,
+      "loss": 0.9841,
+      "step": 1981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.916851121963337e-05,
+      "loss": 0.9971,
+      "step": 1982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9122200083837134e-05,
+      "loss": 0.9463,
+      "step": 1983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9075931728840368e-05,
+      "loss": 0.9531,
+      "step": 1984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.90297062187473e-05,
+      "loss": 0.9526,
+      "step": 1985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8983523617602833e-05,
+      "loss": 0.9263,
+      "step": 1986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8937383989392294e-05,
+      "loss": 0.9983,
+      "step": 1987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.889128739804159e-05,
+      "loss": 0.9717,
+      "step": 1988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8845233907416988e-05,
+      "loss": 0.9766,
+      "step": 1989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8799223581324964e-05,
+      "loss": 0.8943,
+      "step": 1990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8753256483512273e-05,
+      "loss": 0.9158,
+      "step": 1991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8707332677665752e-05,
+      "loss": 0.9697,
+      "step": 1992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.866145222741222e-05,
+      "loss": 0.9595,
+      "step": 1993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8615615196318476e-05,
+      "loss": 0.9961,
+      "step": 1994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.856982164789116e-05,
+      "loss": 0.9609,
+      "step": 1995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.852407164557661e-05,
+      "loss": 0.9346,
+      "step": 1996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.847836525276091e-05,
+      "loss": 0.8813,
+      "step": 1997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8432702532769686e-05,
+      "loss": 0.9822,
+      "step": 1998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.838708354886802e-05,
+      "loss": 0.9556,
+      "step": 1999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8341508364260467e-05,
+      "loss": 0.9438,
+      "step": 2000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.829597704209088e-05,
+      "loss": 0.938,
+      "step": 2001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.825048964544228e-05,
+      "loss": 0.9426,
+      "step": 2002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.820504623733692e-05,
+      "loss": 0.9121,
+      "step": 2003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8159646880736036e-05,
+      "loss": 0.9224,
+      "step": 2004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8114291638539883e-05,
+      "loss": 1.0042,
+      "step": 2005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8068980573587545e-05,
+      "loss": 0.9688,
+      "step": 2006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8023713748656945e-05,
+      "loss": 0.8799,
+      "step": 2007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7978491226464707e-05,
+      "loss": 0.9597,
+      "step": 2008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7933313069666026e-05,
+      "loss": 0.9111,
+      "step": 2009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.788817934085469e-05,
+      "loss": 0.9893,
+      "step": 2010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7843090102562908e-05,
+      "loss": 0.9082,
+      "step": 2011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7798045417261235e-05,
+      "loss": 1.0605,
+      "step": 2012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7753045347358505e-05,
+      "loss": 0.9299,
+      "step": 2013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708089955201773e-05,
+      "loss": 0.8633,
+      "step": 2014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7663179303076127e-05,
+      "loss": 1.0186,
+      "step": 2015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.761831345320472e-05,
+      "loss": 0.9834,
+      "step": 2016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7573492467748638e-05,
+      "loss": 0.8726,
+      "step": 2017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.752871640880675e-05,
+      "loss": 0.8989,
+      "step": 2018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.748398533841573e-05,
+      "loss": 0.9294,
+      "step": 2019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7439299318549933e-05,
+      "loss": 0.9172,
+      "step": 2020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.739465841112125e-05,
+      "loss": 0.9182,
+      "step": 2021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7350062677979074e-05,
+      "loss": 0.9998,
+      "step": 2022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7305512180910244e-05,
+      "loss": 0.958,
+      "step": 2023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7261006981638926e-05,
+      "loss": 0.9644,
+      "step": 2024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7216547141826474e-05,
+      "loss": 0.9109,
+      "step": 2025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7172132723071443e-05,
+      "loss": 0.9656,
+      "step": 2026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7127763786909472e-05,
+      "loss": 0.9893,
+      "step": 2027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7083440394813115e-05,
+      "loss": 0.9417,
+      "step": 2028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7039162608191893e-05,
+      "loss": 0.9961,
+      "step": 2029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6994930488392134e-05,
+      "loss": 0.8875,
+      "step": 2030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.695074409669684e-05,
+      "loss": 0.9177,
+      "step": 2031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6906603494325728e-05,
+      "loss": 0.949,
+      "step": 2032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6862508742435012e-05,
+      "loss": 0.9109,
+      "step": 2033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6818459902117427e-05,
+      "loss": 0.9492,
+      "step": 2034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6774457034402097e-05,
+      "loss": 0.9565,
+      "step": 2035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.67305002002544e-05,
+      "loss": 0.9075,
+      "step": 2036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.668658946057599e-05,
+      "loss": 0.9717,
+      "step": 2037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.664272487620466e-05,
+      "loss": 0.9148,
+      "step": 2038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6598906507914214e-05,
+      "loss": 0.9502,
+      "step": 2039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6555134416414424e-05,
+      "loss": 0.9006,
+      "step": 2040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.651140866235099e-05,
+      "loss": 0.9834,
+      "step": 2041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6467729306305407e-05,
+      "loss": 0.9092,
+      "step": 2042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6424096408794824e-05,
+      "loss": 0.8877,
+      "step": 2043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.638051003027209e-05,
+      "loss": 0.9587,
+      "step": 2044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.63369702311256e-05,
+      "loss": 0.9722,
+      "step": 2045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6293477071679146e-05,
+      "loss": 0.9995,
+      "step": 2046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6250030612191976e-05,
+      "loss": 0.9661,
+      "step": 2047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6206630912858616e-05,
+      "loss": 0.9363,
+      "step": 2048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6163278033808776e-05,
+      "loss": 0.9072,
+      "step": 2049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.611997203510733e-05,
+      "loss": 0.9294,
+      "step": 2050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6076712976754198e-05,
+      "loss": 0.9614,
+      "step": 2051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.603350091868423e-05,
+      "loss": 0.9929,
+      "step": 2052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5990335920767203e-05,
+      "loss": 0.9475,
+      "step": 2053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.594721804280768e-05,
+      "loss": 1.0146,
+      "step": 2054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.590414734454493e-05,
+      "loss": 0.9722,
+      "step": 2055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5861123885652828e-05,
+      "loss": 0.9834,
+      "step": 2056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5818147725739857e-05,
+      "loss": 0.9465,
+      "step": 2057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.577521892434895e-05,
+      "loss": 0.9749,
+      "step": 2058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.573233754095738e-05,
+      "loss": 0.9346,
+      "step": 2059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5689503634976786e-05,
+      "loss": 0.9465,
+      "step": 2060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5646717265753013e-05,
+      "loss": 1.0125,
+      "step": 2061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5603978492566e-05,
+      "loss": 0.9817,
+      "step": 2062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5561287374629785e-05,
+      "loss": 1.009,
+      "step": 2063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.551864397109239e-05,
+      "loss": 1.0159,
+      "step": 2064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5476048341035675e-05,
+      "loss": 0.9458,
+      "step": 2065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.543350054347536e-05,
+      "loss": 0.928,
+      "step": 2066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5391000637360897e-05,
+      "loss": 0.9504,
+      "step": 2067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5348548681575332e-05,
+      "loss": 0.9763,
+      "step": 2068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.530614473493533e-05,
+      "loss": 0.9382,
+      "step": 2069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5263788856191025e-05,
+      "loss": 0.8909,
+      "step": 2070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.522148110402593e-05,
+      "loss": 0.9233,
+      "step": 2071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.517922153705692e-05,
+      "loss": 0.9451,
+      "step": 2072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5137010213834086e-05,
+      "loss": 0.9375,
+      "step": 2073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5094847192840644e-05,
+      "loss": 0.9104,
+      "step": 2074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5052732532492958e-05,
+      "loss": 0.949,
+      "step": 2075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5010666291140363e-05,
+      "loss": 0.9617,
+      "step": 2076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4968648527065066e-05,
+      "loss": 0.9116,
+      "step": 2077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.492667929848217e-05,
+      "loss": 0.9773,
+      "step": 2078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4884758663539516e-05,
+      "loss": 0.95,
+      "step": 2079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4842886680317591e-05,
+      "loss": 0.8916,
+      "step": 2080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4801063406829496e-05,
+      "loss": 0.9265,
+      "step": 2081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4759288901020874e-05,
+      "loss": 0.9065,
+      "step": 2082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.471756322076973e-05,
+      "loss": 0.988,
+      "step": 2083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4675886423886486e-05,
+      "loss": 0.9597,
+      "step": 2084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4634258568113834e-05,
+      "loss": 1.0525,
+      "step": 2085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.45926797111266e-05,
+      "loss": 0.95,
+      "step": 2086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4551149910531781e-05,
+      "loss": 0.9128,
+      "step": 2087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.45096692238684e-05,
+      "loss": 0.9548,
+      "step": 2088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4468237708607397e-05,
+      "loss": 0.9048,
+      "step": 2089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4426855422151637e-05,
+      "loss": 0.9106,
+      "step": 2090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4385522421835723e-05,
+      "loss": 1.0374,
+      "step": 2091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4344238764926032e-05,
+      "loss": 1.0369,
+      "step": 2092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4303004508620516e-05,
+      "loss": 0.9282,
+      "step": 2093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4261819710048724e-05,
+      "loss": 0.8962,
+      "step": 2094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4220684426271691e-05,
+      "loss": 1.0154,
+      "step": 2095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.41795987142818e-05,
+      "loss": 0.9265,
+      "step": 2096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4138562631002794e-05,
+      "loss": 0.9194,
+      "step": 2097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4097576233289662e-05,
+      "loss": 0.9353,
+      "step": 2098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.405663957792851e-05,
+      "loss": 0.9058,
+      "step": 2099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4015752721636571e-05,
+      "loss": 0.9336,
+      "step": 2100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.397491572106207e-05,
+      "loss": 0.9265,
+      "step": 2101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3934128632784133e-05,
+      "loss": 0.9548,
+      "step": 2102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3893391513312759e-05,
+      "loss": 0.9314,
+      "step": 2103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.385270441908873e-05,
+      "loss": 0.9597,
+      "step": 2104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.381206740648347e-05,
+      "loss": 0.9883,
+      "step": 2105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3771480531799052e-05,
+      "loss": 0.9404,
+      "step": 2106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3730943851268107e-05,
+      "loss": 0.9736,
+      "step": 2107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3690457421053637e-05,
+      "loss": 0.8779,
+      "step": 2108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.36500212972491e-05,
+      "loss": 0.8821,
+      "step": 2109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3609635535878245e-05,
+      "loss": 0.9836,
+      "step": 2110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3569300192895007e-05,
+      "loss": 0.9187,
+      "step": 2111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3529015324183509e-05,
+      "loss": 0.9763,
+      "step": 2112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.348878098555793e-05,
+      "loss": 0.8674,
+      "step": 2113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3448597232762411e-05,
+      "loss": 1.0186,
+      "step": 2114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3408464121471049e-05,
+      "loss": 0.9626,
+      "step": 2115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3368381707287763e-05,
+      "loss": 0.9558,
+      "step": 2116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3328350045746212e-05,
+      "loss": 1.0598,
+      "step": 2117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3288369192309762e-05,
+      "loss": 0.9753,
+      "step": 2118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3248439202371399e-05,
+      "loss": 0.9563,
+      "step": 2119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3208560131253577e-05,
+      "loss": 0.9011,
+      "step": 2120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3168732034208265e-05,
+      "loss": 0.9502,
+      "step": 2121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3128954966416802e-05,
+      "loss": 0.8665,
+      "step": 2122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.308922898298977e-05,
+      "loss": 0.9624,
+      "step": 2123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3049554138967051e-05,
+      "loss": 0.9204,
+      "step": 2124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3009930489317612e-05,
+      "loss": 0.9031,
+      "step": 2125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2970358088939532e-05,
+      "loss": 0.9299,
+      "step": 2126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2930836992659856e-05,
+      "loss": 0.9382,
+      "step": 2127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2891367255234566e-05,
+      "loss": 0.9536,
+      "step": 2128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2851948931348495e-05,
+      "loss": 0.9143,
+      "step": 2129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.281258207561521e-05,
+      "loss": 0.9512,
+      "step": 2130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2773266742576989e-05,
+      "loss": 0.915,
+      "step": 2131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2734002986704758e-05,
+      "loss": 0.988,
+      "step": 2132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.269479086239791e-05,
+      "loss": 0.9575,
+      "step": 2133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2655630423984366e-05,
+      "loss": 0.9111,
+      "step": 2134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2616521725720425e-05,
+      "loss": 0.9963,
+      "step": 2135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2577464821790674e-05,
+      "loss": 0.98,
+      "step": 2136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2538459766307958e-05,
+      "loss": 1.0122,
+      "step": 2137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2499506613313305e-05,
+      "loss": 0.9822,
+      "step": 2138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.246060541677579e-05,
+      "loss": 0.9351,
+      "step": 2139
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2421756230592534e-05,
+      "loss": 0.9458,
+      "step": 2140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2382959108588627e-05,
+      "loss": 0.9038,
+      "step": 2141
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2344214104516921e-05,
+      "loss": 0.9536,
+      "step": 2142
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2305521272058162e-05,
+      "loss": 0.9519,
+      "step": 2143
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2266880664820797e-05,
+      "loss": 0.9749,
+      "step": 2144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2228292336340857e-05,
+      "loss": 0.9258,
+      "step": 2145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2189756340082003e-05,
+      "loss": 0.9077,
+      "step": 2146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2151272729435375e-05,
+      "loss": 0.9177,
+      "step": 2147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2112841557719507e-05,
+      "loss": 0.9539,
+      "step": 2148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.207446287818031e-05,
+      "loss": 0.9514,
+      "step": 2149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2036136743990967e-05,
+      "loss": 1.0085,
+      "step": 2150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1997863208251825e-05,
+      "loss": 0.9263,
+      "step": 2151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.19596423239904e-05,
+      "loss": 0.9617,
+      "step": 2152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1921474144161248e-05,
+      "loss": 0.8984,
+      "step": 2153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1883358721645876e-05,
+      "loss": 0.9563,
+      "step": 2154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1845296109252724e-05,
+      "loss": 0.9922,
+      "step": 2155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.180728635971708e-05,
+      "loss": 0.9585,
+      "step": 2156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1769329525700935e-05,
+      "loss": 0.9402,
+      "step": 2157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1731425659793027e-05,
+      "loss": 0.989,
+      "step": 2158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1693574814508657e-05,
+      "loss": 0.9438,
+      "step": 2159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1655777042289722e-05,
+      "loss": 0.9023,
+      "step": 2160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1618032395504519e-05,
+      "loss": 0.8923,
+      "step": 2161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1580340926447796e-05,
+      "loss": 0.9531,
+      "step": 2162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1542702687340612e-05,
+      "loss": 0.9453,
+      "step": 2163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.150511773033025e-05,
+      "loss": 0.9033,
+      "step": 2164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1467586107490202e-05,
+      "loss": 0.9604,
+      "step": 2165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1430107870820062e-05,
+      "loss": 0.9517,
+      "step": 2166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1392683072245431e-05,
+      "loss": 0.9573,
+      "step": 2167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1355311763617898e-05,
+      "loss": 0.9299,
+      "step": 2168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.131799399671496e-05,
+      "loss": 0.9175,
+      "step": 2169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1280729823239871e-05,
+      "loss": 0.9844,
+      "step": 2170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1243519294821691e-05,
+      "loss": 0.9104,
+      "step": 2171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1206362463015146e-05,
+      "loss": 0.9827,
+      "step": 2172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1169259379300523e-05,
+      "loss": 1.0112,
+      "step": 2173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1132210095083694e-05,
+      "loss": 0.9233,
+      "step": 2174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1095214661695986e-05,
+      "loss": 0.9417,
+      "step": 2175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1058273130394076e-05,
+      "loss": 0.9946,
+      "step": 2176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1021385552359982e-05,
+      "loss": 0.9333,
+      "step": 2177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0984551978701003e-05,
+      "loss": 0.9326,
+      "step": 2178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0947772460449557e-05,
+      "loss": 0.9248,
+      "step": 2179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.091104704856321e-05,
+      "loss": 0.9617,
+      "step": 2180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0874375793924573e-05,
+      "loss": 0.9304,
+      "step": 2181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0837758747341176e-05,
+      "loss": 0.938,
+      "step": 2182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0801195959545484e-05,
+      "loss": 0.9771,
+      "step": 2183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0764687481194786e-05,
+      "loss": 0.9084,
+      "step": 2184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0728233362871087e-05,
+      "loss": 0.9873,
+      "step": 2185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0691833655081124e-05,
+      "loss": 0.9336,
+      "step": 2186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0655488408256242e-05,
+      "loss": 0.8933,
+      "step": 2187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0619197672752285e-05,
+      "loss": 0.9802,
+      "step": 2188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.058296149884963e-05,
+      "loss": 0.9019,
+      "step": 2189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0546779936753038e-05,
+      "loss": 1.0186,
+      "step": 2190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0510653036591583e-05,
+      "loss": 0.9243,
+      "step": 2191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.047458084841864e-05,
+      "loss": 0.9351,
+      "step": 2192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0438563422211783e-05,
+      "loss": 0.9014,
+      "step": 2193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0402600807872675e-05,
+      "loss": 1.0,
+      "step": 2194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.036669305522706e-05,
+      "loss": 0.926,
+      "step": 2195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0330840214024679e-05,
+      "loss": 0.9253,
+      "step": 2196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0295042333939203e-05,
+      "loss": 0.9387,
+      "step": 2197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0259299464568113e-05,
+      "loss": 0.8274,
+      "step": 2198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0223611655432714e-05,
+      "loss": 0.9363,
+      "step": 2199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0187978955978028e-05,
+      "loss": 0.9431,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0152401415572676e-05,
+      "loss": 0.9568,
+      "step": 2201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0116879083508906e-05,
+      "loss": 0.9446,
+      "step": 2202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0081412009002467e-05,
+      "loss": 0.9814,
+      "step": 2203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0046000241192517e-05,
+      "loss": 0.9287,
+      "step": 2204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0010643829141626e-05,
+      "loss": 0.9153,
+      "step": 2205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.975342821835654e-06,
+      "loss": 0.9258,
+      "step": 2206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.94009726818369e-06,
+      "loss": 0.908,
+      "step": 2207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.904907217018e-06,
+      "loss": 0.8962,
+      "step": 2208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.869772717093973e-06,
+      "loss": 0.9177,
+      "step": 2209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.834693817089996e-06,
+      "loss": 0.9014,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.799670565607428e-06,
+      "loss": 0.9333,
+      "step": 2211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.764703011170567e-06,
+      "loss": 0.9744,
+      "step": 2212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.729791202226485e-06,
+      "loss": 0.9265,
+      "step": 2213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.69493518714506e-06,
+      "loss": 0.99,
+      "step": 2214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.66013501421888e-06,
+      "loss": 0.9475,
+      "step": 2215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.625390731663115e-06,
+      "loss": 0.9585,
+      "step": 2216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.59070238761553e-06,
+      "loss": 0.9536,
+      "step": 2217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.55607003013641e-06,
+      "loss": 0.9351,
+      "step": 2218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521493707208412e-06,
+      "loss": 0.9617,
+      "step": 2219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.48697346673661e-06,
+      "loss": 0.9963,
+      "step": 2220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.452509356548368e-06,
+      "loss": 1.0051,
+      "step": 2221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.41810142439325e-06,
+      "loss": 0.9856,
+      "step": 2222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.383749717943024e-06,
+      "loss": 0.9465,
+      "step": 2223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.349454284791549e-06,
+      "loss": 1.0034,
+      "step": 2224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31521517245469e-06,
+      "loss": 0.9446,
+      "step": 2225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.281032428370317e-06,
+      "loss": 0.9209,
+      "step": 2226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.246906099898194e-06,
+      "loss": 0.9768,
+      "step": 2227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.212836234319911e-06,
+      "loss": 0.947,
+      "step": 2228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.178822878838817e-06,
+      "loss": 1.0027,
+      "step": 2229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.144866080579994e-06,
+      "loss": 0.9558,
+      "step": 2230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.110965886590167e-06,
+      "loss": 0.937,
+      "step": 2231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.077122343837607e-06,
+      "loss": 0.9763,
+      "step": 2232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.04333549921212e-06,
+      "loss": 0.9148,
+      "step": 2233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.009605399524956e-06,
+      "loss": 0.9131,
+      "step": 2234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.975932091508726e-06,
+      "loss": 0.9502,
+      "step": 2235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.942315621817376e-06,
+      "loss": 0.8904,
+      "step": 2236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.908756037026112e-06,
+      "loss": 0.873,
+      "step": 2237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.875253383631287e-06,
+      "loss": 0.9248,
+      "step": 2238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.841807708050414e-06,
+      "loss": 0.8958,
+      "step": 2239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.808419056622064e-06,
+      "loss": 0.9009,
+      "step": 2240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.775087475605764e-06,
+      "loss": 0.9646,
+      "step": 2241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.741813011182014e-06,
+      "loss": 0.9243,
+      "step": 2242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.708595709452167e-06,
+      "loss": 0.9138,
+      "step": 2243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.675435616438348e-06,
+      "loss": 0.9707,
+      "step": 2244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.642332778083472e-06,
+      "loss": 0.8865,
+      "step": 2245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.609287240251079e-06,
+      "loss": 0.9412,
+      "step": 2246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.576299048725362e-06,
+      "loss": 1.0039,
+      "step": 2247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.543368249211014e-06,
+      "loss": 0.9441,
+      "step": 2248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.510494887333275e-06,
+      "loss": 0.9072,
+      "step": 2249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.477679008637735e-06,
+      "loss": 0.9709,
+      "step": 2250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.44492065859039e-06,
+      "loss": 1.0039,
+      "step": 2251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.412219882577537e-06,
+      "loss": 1.0205,
+      "step": 2252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.379576725905653e-06,
+      "loss": 0.9795,
+      "step": 2253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.346991233801438e-06,
+      "loss": 0.9214,
+      "step": 2254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.314463451411681e-06,
+      "loss": 0.967,
+      "step": 2255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.281993423803191e-06,
+      "loss": 0.8801,
+      "step": 2256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.249581195962791e-06,
+      "loss": 0.9236,
+      "step": 2257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.217226812797223e-06,
+      "loss": 0.9856,
+      "step": 2258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.18493031913305e-06,
+      "loss": 0.9272,
+      "step": 2259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.152691759716657e-06,
+      "loss": 0.9409,
+      "step": 2260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.120511179214185e-06,
+      "loss": 0.8757,
+      "step": 2261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.088388622211402e-06,
+      "loss": 0.8826,
+      "step": 2262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.056324133213688e-06,
+      "loss": 0.9055,
+      "step": 2263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.024317756645999e-06,
+      "loss": 0.9314,
+      "step": 2264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.992369536852774e-06,
+      "loss": 0.9688,
+      "step": 2265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.960479518097841e-06,
+      "loss": 0.9673,
+      "step": 2266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.928647744564427e-06,
+      "loss": 0.8818,
+      "step": 2267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.896874260355064e-06,
+      "loss": 0.9465,
+      "step": 2268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.865159109491489e-06,
+      "loss": 0.8652,
+      "step": 2269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.83350233591465e-06,
+      "loss": 0.9888,
+      "step": 2270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.801903983484616e-06,
+      "loss": 0.9255,
+      "step": 2271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.77036409598048e-06,
+      "loss": 0.9302,
+      "step": 2272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.738882717100365e-06,
+      "loss": 0.9697,
+      "step": 2273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.707459890461338e-06,
+      "loss": 1.0132,
+      "step": 2274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.676095659599297e-06,
+      "loss": 0.9668,
+      "step": 2275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.644790067969005e-06,
+      "loss": 0.9558,
+      "step": 2276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.613543158943964e-06,
+      "loss": 0.9265,
+      "step": 2277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.582354975816347e-06,
+      "loss": 0.9314,
+      "step": 2278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.55122556179702e-06,
+      "loss": 0.9871,
+      "step": 2279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.520154960015352e-06,
+      "loss": 0.9211,
+      "step": 2280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.489143213519301e-06,
+      "loss": 0.9321,
+      "step": 2281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.458190365275219e-06,
+      "loss": 0.9426,
+      "step": 2282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4272964581678985e-06,
+      "loss": 0.8958,
+      "step": 2283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.39646153500047e-06,
+      "loss": 0.907,
+      "step": 2284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3656856384942965e-06,
+      "loss": 0.9656,
+      "step": 2285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.334968811289006e-06,
+      "loss": 0.9001,
+      "step": 2286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.30431109594239e-06,
+      "loss": 0.9153,
+      "step": 2287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.273712534930294e-06,
+      "loss": 0.8652,
+      "step": 2288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.243173170646645e-06,
+      "loss": 0.8931,
+      "step": 2289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.212693045403363e-06,
+      "loss": 0.9175,
+      "step": 2290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.182272201430245e-06,
+      "loss": 0.9719,
+      "step": 2291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.151910680875001e-06,
+      "loss": 0.937,
+      "step": 2292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1216085258031414e-06,
+      "loss": 0.9709,
+      "step": 2293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0913657781978946e-06,
+      "loss": 0.9912,
+      "step": 2294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.061182479960221e-06,
+      "loss": 0.9417,
+      "step": 2295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.031058672908691e-06,
+      "loss": 0.9626,
+      "step": 2296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.000994398779442e-06,
+      "loss": 0.907,
+      "step": 2297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.97098969922616e-06,
+      "loss": 0.9451,
+      "step": 2298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.94104461581998e-06,
+      "loss": 0.9373,
+      "step": 2299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9111591900494155e-06,
+      "loss": 1.0225,
+      "step": 2300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.881333463320355e-06,
+      "loss": 0.979,
+      "step": 2301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.851567476955978e-06,
+      "loss": 0.9319,
+      "step": 2302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8218612721966505e-06,
+      "loss": 0.9087,
+      "step": 2303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.79221489019996e-06,
+      "loss": 0.9395,
+      "step": 2304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.762628372040603e-06,
+      "loss": 1.0098,
+      "step": 2305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.733101758710297e-06,
+      "loss": 0.9934,
+      "step": 2306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7036350911178045e-06,
+      "loss": 0.938,
+      "step": 2307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.674228410088828e-06,
+      "loss": 0.9155,
+      "step": 2308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6448817563659335e-06,
+      "loss": 0.884,
+      "step": 2309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6155951706085405e-06,
+      "loss": 0.9814,
+      "step": 2310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.586368693392858e-06,
+      "loss": 0.9087,
+      "step": 2311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.557202365211779e-06,
+      "loss": 0.9421,
+      "step": 2312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.528096226474895e-06,
+      "loss": 0.9167,
+      "step": 2313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.499050317508371e-06,
+      "loss": 0.9294,
+      "step": 2314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.470064678554971e-06,
+      "loss": 0.9136,
+      "step": 2315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.441139349773906e-06,
+      "loss": 0.9543,
+      "step": 2316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.412274371240867e-06,
+      "loss": 0.9543,
+      "step": 2317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.383469782947915e-06,
+      "loss": 0.9304,
+      "step": 2318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.354725624803426e-06,
+      "loss": 0.9512,
+      "step": 2319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.326041936632077e-06,
+      "loss": 0.9592,
+      "step": 2320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.297418758174767e-06,
+      "loss": 0.9294,
+      "step": 2321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.268856129088518e-06,
+      "loss": 0.9536,
+      "step": 2322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.240354088946504e-06,
+      "loss": 0.9287,
+      "step": 2323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.211912677237947e-06,
+      "loss": 0.9248,
+      "step": 2324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.1835319333680475e-06,
+      "loss": 0.8948,
+      "step": 2325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.15521189665797e-06,
+      "loss": 0.8928,
+      "step": 2326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1269526063447765e-06,
+      "loss": 0.925,
+      "step": 2327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.098754101581333e-06,
+      "loss": 0.9341,
+      "step": 2328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.070616421436326e-06,
+      "loss": 0.9622,
+      "step": 2329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.04253960489416e-06,
+      "loss": 1.001,
+      "step": 2330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.014523690854895e-06,
+      "loss": 0.8945,
+      "step": 2331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.986568718134222e-06,
+      "loss": 0.8813,
+      "step": 2332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.958674725463409e-06,
+      "loss": 0.9133,
+      "step": 2333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.930841751489219e-06,
+      "loss": 0.9387,
+      "step": 2334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.903069834773883e-06,
+      "loss": 0.9136,
+      "step": 2335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.875359013795062e-06,
+      "loss": 0.9375,
+      "step": 2336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.847709326945716e-06,
+      "loss": 0.8833,
+      "step": 2337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.820120812534147e-06,
+      "loss": 0.9478,
+      "step": 2338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.792593508783905e-06,
+      "loss": 0.9233,
+      "step": 2339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.765127453833696e-06,
+      "loss": 0.9275,
+      "step": 2340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.737722685737401e-06,
+      "loss": 0.9763,
+      "step": 2341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.710379242463992e-06,
+      "loss": 0.9282,
+      "step": 2342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.683097161897432e-06,
+      "loss": 0.9138,
+      "step": 2343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6558764818367195e-06,
+      "loss": 0.9758,
+      "step": 2344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.628717239995762e-06,
+      "loss": 0.9885,
+      "step": 2345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.601619474003328e-06,
+      "loss": 0.9202,
+      "step": 2346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.57458322140304e-06,
+      "loss": 0.8833,
+      "step": 2347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.547608519653285e-06,
+      "loss": 0.884,
+      "step": 2348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.520695406127163e-06,
+      "loss": 0.9094,
+      "step": 2349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.493843918112445e-06,
+      "loss": 0.9963,
+      "step": 2350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.467054092811535e-06,
+      "loss": 0.9106,
+      "step": 2351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4403259673414035e-06,
+      "loss": 0.8901,
+      "step": 2352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.413659578733504e-06,
+      "loss": 0.9221,
+      "step": 2353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.387054963933802e-06,
+      "loss": 0.9241,
+      "step": 2354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.36051215980265e-06,
+      "loss": 0.9407,
+      "step": 2355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.334031203114753e-06,
+      "loss": 1.0149,
+      "step": 2356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.307612130559153e-06,
+      "loss": 0.9404,
+      "step": 2357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.281254978739142e-06,
+      "loss": 0.9871,
+      "step": 2358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.254959784172198e-06,
+      "loss": 0.9395,
+      "step": 2359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.22872658329e-06,
+      "loss": 0.9233,
+      "step": 2360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2025554124383095e-06,
+      "loss": 0.9741,
+      "step": 2361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.176446307876947e-06,
+      "loss": 0.926,
+      "step": 2362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1503993057797465e-06,
+      "loss": 0.9341,
+      "step": 2363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1244144422345046e-06,
+      "loss": 0.9138,
+      "step": 2364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.098491753242918e-06,
+      "loss": 0.9246,
+      "step": 2365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0726312747205295e-06,
+      "loss": 0.8552,
+      "step": 2366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.046833042496718e-06,
+      "loss": 0.9661,
+      "step": 2367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0210970923145975e-06,
+      "loss": 0.9097,
+      "step": 2368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9954234598310055e-06,
+      "loss": 0.9512,
+      "step": 2369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.96981218061644e-06,
+      "loss": 0.925,
+      "step": 2370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.944263290154982e-06,
+      "loss": 0.9719,
+      "step": 2371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.918776823844312e-06,
+      "loss": 0.9473,
+      "step": 2372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.893352816995611e-06,
+      "loss": 0.8975,
+      "step": 2373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.867991304833502e-06,
+      "loss": 0.9607,
+      "step": 2374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8426923224960396e-06,
+      "loss": 0.9036,
+      "step": 2375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.817455905034657e-06,
+      "loss": 0.9124,
+      "step": 2376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7922820874140684e-06,
+      "loss": 0.9355,
+      "step": 2377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.767170904512292e-06,
+      "loss": 0.9141,
+      "step": 2378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742122391120557e-06,
+      "loss": 0.9617,
+      "step": 2379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.717136581943244e-06,
+      "loss": 0.9424,
+      "step": 2380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.692213511597887e-06,
+      "loss": 0.874,
+      "step": 2381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6673532146150885e-06,
+      "loss": 0.8862,
+      "step": 2382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.642555725438464e-06,
+      "loss": 0.9795,
+      "step": 2383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.617821078424611e-06,
+      "loss": 0.9084,
+      "step": 2384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.593149307843081e-06,
+      "loss": 0.9199,
+      "step": 2385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5685404478763075e-06,
+      "loss": 0.9299,
+      "step": 2386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5439945326195336e-06,
+      "loss": 0.9751,
+      "step": 2387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.519511596080817e-06,
+      "loss": 0.9731,
+      "step": 2388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4950916721809735e-06,
+      "loss": 0.8779,
+      "step": 2389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.470734794753473e-06,
+      "loss": 0.9749,
+      "step": 2390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.446440997544471e-06,
+      "loss": 0.9622,
+      "step": 2391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.422210314212727e-06,
+      "loss": 0.8811,
+      "step": 2392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3980427783295196e-06,
+      "loss": 0.9233,
+      "step": 2393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.373938423378682e-06,
+      "loss": 0.9836,
+      "step": 2394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.349897282756487e-06,
+      "loss": 1.0117,
+      "step": 2395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.325919389771627e-06,
+      "loss": 0.9673,
+      "step": 2396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.302004777645163e-06,
+      "loss": 0.9314,
+      "step": 2397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2781534795104994e-06,
+      "loss": 0.9031,
+      "step": 2398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.254365528413295e-06,
+      "loss": 0.9094,
+      "step": 2399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.230640957311471e-06,
+      "loss": 0.8755,
+      "step": 2400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2069797990751e-06,
+      "loss": 0.9407,
+      "step": 2401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.183382086486415e-06,
+      "loss": 0.9436,
+      "step": 2402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.159847852239757e-06,
+      "loss": 0.8962,
+      "step": 2403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.136377128941515e-06,
+      "loss": 0.9878,
+      "step": 2404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.112969949110063e-06,
+      "loss": 0.9314,
+      "step": 2405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.089626345175757e-06,
+      "loss": 0.9766,
+      "step": 2406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.066346349480871e-06,
+      "loss": 0.905,
+      "step": 2407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0431299942795264e-06,
+      "loss": 0.9758,
+      "step": 2408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0199773117377e-06,
+      "loss": 0.8843,
+      "step": 2409
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.996888333933146e-06,
+      "loss": 0.9353,
+      "step": 2410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9738630928553344e-06,
+      "loss": 0.9778,
+      "step": 2411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.95090162040545e-06,
+      "loss": 0.9197,
+      "step": 2412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.928003948396336e-06,
+      "loss": 0.9634,
+      "step": 2413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9051701085523975e-06,
+      "loss": 0.9326,
+      "step": 2414
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.882400132509651e-06,
+      "loss": 0.9934,
+      "step": 2415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.859694051815604e-06,
+      "loss": 0.9319,
+      "step": 2416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.837051897929251e-06,
+      "loss": 0.969,
+      "step": 2417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.814473702220983e-06,
+      "loss": 0.9131,
+      "step": 2418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.791959495972619e-06,
+      "loss": 0.9819,
+      "step": 2419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.769509310377317e-06,
+      "loss": 0.9658,
+      "step": 2420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7471231765395078e-06,
+      "loss": 0.9646,
+      "step": 2421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.724801125474897e-06,
+      "loss": 1.0286,
+      "step": 2422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7025431881104134e-06,
+      "loss": 0.9304,
+      "step": 2423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.680349395284133e-06,
+      "loss": 0.9211,
+      "step": 2424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.658219777745281e-06,
+      "loss": 0.8872,
+      "step": 2425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6361543661541654e-06,
+      "loss": 0.8904,
+      "step": 2426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.614153191082126e-06,
+      "loss": 0.9478,
+      "step": 2427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5922162830115124e-06,
+      "loss": 0.9141,
+      "step": 2428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.570343672335641e-06,
+      "loss": 0.9443,
+      "step": 2429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5485353893587205e-06,
+      "loss": 0.9226,
+      "step": 2430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.526791464295853e-06,
+      "loss": 0.8591,
+      "step": 2431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5051119272729914e-06,
+      "loss": 0.8892,
+      "step": 2432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4834968083268302e-06,
+      "loss": 0.9583,
+      "step": 2433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4619461374048646e-06,
+      "loss": 0.9795,
+      "step": 2434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4404599443652706e-06,
+      "loss": 0.863,
+      "step": 2435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4190382589768753e-06,
+      "loss": 0.8916,
+      "step": 2436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.397681110919171e-06,
+      "loss": 0.9199,
+      "step": 2437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.376388529782215e-06,
+      "loss": 0.9514,
+      "step": 2438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3551605450665992e-06,
+      "loss": 0.918,
+      "step": 2439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.333997186183435e-06,
+      "loss": 0.949,
+      "step": 2440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3128984824542852e-06,
+      "loss": 0.9463,
+      "step": 2441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2918644631111275e-06,
+      "loss": 0.9141,
+      "step": 2442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2708951572963387e-06,
+      "loss": 1.0049,
+      "step": 2443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.24999059406263e-06,
+      "loss": 0.9644,
+      "step": 2444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2291508023729875e-06,
+      "loss": 0.9424,
+      "step": 2445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2083758111006945e-06,
+      "loss": 0.9153,
+      "step": 2446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1876656490292423e-06,
+      "loss": 1.0378,
+      "step": 2447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1670203448522782e-06,
+      "loss": 0.9319,
+      "step": 2448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1464399271736224e-06,
+      "loss": 0.884,
+      "step": 2449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.125924424507182e-06,
+      "loss": 0.99,
+      "step": 2450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1054738652769257e-06,
+      "loss": 0.9917,
+      "step": 2451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0850882778168333e-06,
+      "loss": 0.96,
+      "step": 2452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0647676903708846e-06,
+      "loss": 0.9341,
+      "step": 2453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.044512131092997e-06,
+      "loss": 0.9043,
+      "step": 2454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0243216280469833e-06,
+      "loss": 0.8799,
+      "step": 2455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.004196209206539e-06,
+      "loss": 1.0107,
+      "step": 2456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.984135902455171e-06,
+      "loss": 0.95,
+      "step": 2457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9641407355861796e-06,
+      "loss": 0.9426,
+      "step": 2458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9442107363026107e-06,
+      "loss": 0.9453,
+      "step": 2459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9243459322172316e-06,
+      "loss": 0.9724,
+      "step": 2460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.904546350852472e-06,
+      "loss": 0.9351,
+      "step": 2461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.884812019640404e-06,
+      "loss": 0.8379,
+      "step": 2462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.86514296592269e-06,
+      "loss": 1.0029,
+      "step": 2463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8455392169505544e-06,
+      "loss": 0.9321,
+      "step": 2464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.826000799884737e-06,
+      "loss": 0.8958,
+      "step": 2465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.806527741795478e-06,
+      "loss": 0.8928,
+      "step": 2466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.787120069662441e-06,
+      "loss": 0.8245,
+      "step": 2467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.767777810374722e-06,
+      "loss": 0.9612,
+      "step": 2468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.748500990730768e-06,
+      "loss": 0.9045,
+      "step": 2469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7292896374383593e-06,
+      "loss": 0.8767,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.710143777114588e-06,
+      "loss": 0.9011,
+      "step": 2471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.691063436285812e-06,
+      "loss": 0.9353,
+      "step": 2472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6720486413875813e-06,
+      "loss": 0.9849,
+      "step": 2473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.653099418764665e-06,
+      "loss": 0.989,
+      "step": 2474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6342157946709746e-06,
+      "loss": 0.9553,
+      "step": 2475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.615397795269514e-06,
+      "loss": 0.9351,
+      "step": 2476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5966454466323954e-06,
+      "loss": 0.895,
+      "step": 2477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.577958774740763e-06,
+      "loss": 0.9131,
+      "step": 2478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5593378054847515e-06,
+      "loss": 0.9336,
+      "step": 2479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.54078256466348e-06,
+      "loss": 0.9167,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5222930779850105e-06,
+      "loss": 0.938,
+      "step": 2481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5038693710662755e-06,
+      "loss": 0.9727,
+      "step": 2482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4855114694330995e-06,
+      "loss": 0.8943,
+      "step": 2483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4672193985201207e-06,
+      "loss": 0.9893,
+      "step": 2484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4489931836707558e-06,
+      "loss": 0.8369,
+      "step": 2485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4308328501372215e-06,
+      "loss": 0.957,
+      "step": 2486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.412738423080396e-06,
+      "loss": 0.9463,
+      "step": 2487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3947099275699103e-06,
+      "loss": 0.9827,
+      "step": 2488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.376747388583994e-06,
+      "loss": 0.9895,
+      "step": 2489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.358850831009518e-06,
+      "loss": 0.9348,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3410202796419532e-06,
+      "loss": 0.9719,
+      "step": 2491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.323255759185278e-06,
+      "loss": 0.9348,
+      "step": 2492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3055572942520252e-06,
+      "loss": 0.9976,
+      "step": 2493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2879249093631927e-06,
+      "loss": 0.8499,
+      "step": 2494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2703586289482214e-06,
+      "loss": 0.9382,
+      "step": 2495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2528584773449657e-06,
+      "loss": 0.9155,
+      "step": 2496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2354244787996748e-06,
+      "loss": 0.9863,
+      "step": 2497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2180566574669216e-06,
+      "loss": 0.9958,
+      "step": 2498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2007550374096074e-06,
+      "loss": 0.9121,
+      "step": 2499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.183519642598908e-06,
+      "loss": 0.9717,
+      "step": 2500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.166350496914238e-06,
+      "loss": 1.0017,
+      "step": 2501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1492476241432304e-06,
+      "loss": 0.9287,
+      "step": 2502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1322110479817138e-06,
+      "loss": 0.9163,
+      "step": 2503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1152407920336347e-06,
+      "loss": 0.9514,
+      "step": 2504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.098336879811058e-06,
+      "loss": 0.873,
+      "step": 2505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.081499334734155e-06,
+      "loss": 0.9497,
+      "step": 2506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.064728180131126e-06,
+      "loss": 0.9014,
+      "step": 2507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.048023439238189e-06,
+      "loss": 0.9141,
+      "step": 2508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.031385135199554e-06,
+      "loss": 0.9751,
+      "step": 2509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0148132910673855e-06,
+      "loss": 0.8914,
+      "step": 2510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9983079298017517e-06,
+      "loss": 0.9431,
+      "step": 2511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.981869074270626e-06,
+      "loss": 0.9219,
+      "step": 2512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9654967472498343e-06,
+      "loss": 0.8984,
+      "step": 2513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9491909714230204e-06,
+      "loss": 1.0503,
+      "step": 2514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9329517693816466e-06,
+      "loss": 0.9045,
+      "step": 2515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9167791636249044e-06,
+      "loss": 0.9709,
+      "step": 2516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9006731765597418e-06,
+      "loss": 0.9688,
+      "step": 2517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8846338305007983e-06,
+      "loss": 0.9973,
+      "step": 2518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8686611476703808e-06,
+      "loss": 0.9397,
+      "step": 2519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.852755150198443e-06,
+      "loss": 0.9019,
+      "step": 2520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.83691586012254e-06,
+      "loss": 0.9026,
+      "step": 2521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8211432993878063e-06,
+      "loss": 0.9795,
+      "step": 2522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8054374898469229e-06,
+      "loss": 0.8772,
+      "step": 2523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7897984532600942e-06,
+      "loss": 0.9756,
+      "step": 2524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7742262112950048e-06,
+      "loss": 0.9355,
+      "step": 2525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7587207855267962e-06,
+      "loss": 0.9795,
+      "step": 2526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7432821974380342e-06,
+      "loss": 0.9883,
+      "step": 2527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.727910468418703e-06,
+      "loss": 0.8914,
+      "step": 2528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7126056197661223e-06,
+      "loss": 0.9172,
+      "step": 2529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.697367672684963e-06,
+      "loss": 0.885,
+      "step": 2530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6821966482872264e-06,
+      "loss": 0.8826,
+      "step": 2531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6670925675921545e-06,
+      "loss": 0.9514,
+      "step": 2532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6520554515262687e-06,
+      "loss": 0.9741,
+      "step": 2533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6370853209233039e-06,
+      "loss": 0.873,
+      "step": 2534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6221821965241746e-06,
+      "loss": 0.8696,
+      "step": 2535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6073460989769807e-06,
+      "loss": 0.8984,
+      "step": 2536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5925770488369517e-06,
+      "loss": 0.939,
+      "step": 2537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5778750665664087e-06,
+      "loss": 0.9297,
+      "step": 2538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.563240172534758e-06,
+      "loss": 0.896,
+      "step": 2539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5486723870184683e-06,
+      "loss": 0.9619,
+      "step": 2540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5341717302010227e-06,
+      "loss": 0.9092,
+      "step": 2541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5197382221728896e-06,
+      "loss": 0.8931,
+      "step": 2542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.505371882931511e-06,
+      "loss": 1.0212,
+      "step": 2543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4910727323812768e-06,
+      "loss": 0.9353,
+      "step": 2544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4768407903334669e-06,
+      "loss": 0.9165,
+      "step": 2545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4626760765062587e-06,
+      "loss": 0.9771,
+      "step": 2546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4485786105246923e-06,
+      "loss": 0.9524,
+      "step": 2547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.434548411920622e-06,
+      "loss": 0.9333,
+      "step": 2548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.420585500132704e-06,
+      "loss": 0.9041,
+      "step": 2549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4066898945063855e-06,
+      "loss": 1.0012,
+      "step": 2550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3928616142938443e-06,
+      "loss": 0.9553,
+      "step": 2551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.379100678653983e-06,
+      "loss": 0.9402,
+      "step": 2552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.365407106652422e-06,
+      "loss": 0.9417,
+      "step": 2553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3517809172614137e-06,
+      "loss": 0.9744,
+      "step": 2554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3382221293598728e-06,
+      "loss": 0.9595,
+      "step": 2555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3247307617333282e-06,
+      "loss": 0.8875,
+      "step": 2556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3113068330739053e-06,
+      "loss": 0.9146,
+      "step": 2557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2979503619802714e-06,
+      "loss": 0.929,
+      "step": 2558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.284661366957668e-06,
+      "loss": 0.9692,
+      "step": 2559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2714398664178172e-06,
+      "loss": 0.9526,
+      "step": 2560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2582858786789387e-06,
+      "loss": 0.9094,
+      "step": 2561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2451994219657203e-06,
+      "loss": 0.946,
+      "step": 2562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2321805144092758e-06,
+      "loss": 0.988,
+      "step": 2563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2192291740471372e-06,
+      "loss": 0.9629,
+      "step": 2564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.206345418823235e-06,
+      "loss": 0.8696,
+      "step": 2565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1935292665878283e-06,
+      "loss": 0.96,
+      "step": 2566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1807807350975476e-06,
+      "loss": 0.9368,
+      "step": 2567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1680998420153132e-06,
+      "loss": 0.8958,
+      "step": 2568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1554866049103496e-06,
+      "loss": 0.9512,
+      "step": 2569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1429410412581276e-06,
+      "loss": 0.9021,
+      "step": 2570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.130463168440371e-06,
+      "loss": 0.9006,
+      "step": 2571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1180530037450176e-06,
+      "loss": 0.9897,
+      "step": 2572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1057105643661802e-06,
+      "loss": 0.906,
+      "step": 2573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0934358674041634e-06,
+      "loss": 0.9868,
+      "step": 2574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0812289298654078e-06,
+      "loss": 0.9258,
+      "step": 2575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0690897686624568e-06,
+      "loss": 0.9595,
+      "step": 2576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0570184006139683e-06,
+      "loss": 0.9414,
+      "step": 2577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.045014842444675e-06,
+      "loss": 0.9612,
+      "step": 2578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.033079110785351e-06,
+      "loss": 0.9375,
+      "step": 2579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0212112221727965e-06,
+      "loss": 0.9482,
+      "step": 2580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0094111930498306e-06,
+      "loss": 0.939,
+      "step": 2581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.976790397652315e-07,
+      "loss": 0.9312,
+      "step": 2582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.860147785737572e-07,
+      "loss": 0.9424,
+      "step": 2583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.744184256360923e-07,
+      "loss": 0.96,
+      "step": 2584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.628899970188344e-07,
+      "loss": 0.9133,
+      "step": 2585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.514295086944735e-07,
+      "loss": 0.9524,
+      "step": 2586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.400369765413753e-07,
+      "loss": 0.8887,
+      "step": 2587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.28712416343741e-07,
+      "loss": 0.9436,
+      "step": 2588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.174558437916148e-07,
+      "loss": 0.9495,
+      "step": 2589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.06267274480832e-07,
+      "loss": 0.8975,
+      "step": 2590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.95146723913004e-07,
+      "loss": 0.8625,
+      "step": 2591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.840942074955172e-07,
+      "loss": 0.9502,
+      "step": 2592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.731097405415056e-07,
+      "loss": 0.9009,
+      "step": 2593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.6219333826979e-07,
+      "loss": 0.9368,
+      "step": 2594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.513450158049108e-07,
+      "loss": 0.8408,
+      "step": 2595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.40564788177095e-07,
+      "loss": 0.9316,
+      "step": 2596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.298526703221899e-07,
+      "loss": 0.9246,
+      "step": 2597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.192086770817176e-07,
+      "loss": 0.9456,
+      "step": 2598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.086328232027873e-07,
+      "loss": 1.0491,
+      "step": 2599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.981251233381059e-07,
+      "loss": 0.8804,
+      "step": 2600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.876855920459614e-07,
+      "loss": 0.9729,
+      "step": 2601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.773142437902004e-07,
+      "loss": 0.9048,
+      "step": 2602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.670110929401785e-07,
+      "loss": 0.9397,
+      "step": 2603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.56776153770794e-07,
+      "loss": 0.9341,
+      "step": 2604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.466094404624202e-07,
+      "loss": 0.877,
+      "step": 2605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.365109671009118e-07,
+      "loss": 0.9119,
+      "step": 2606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.264807476775659e-07,
+      "loss": 0.9211,
+      "step": 2607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.165187960891273e-07,
+      "loss": 0.9336,
+      "step": 2608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.066251261377665e-07,
+      "loss": 0.8958,
+      "step": 2609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.967997515310187e-07,
+      "loss": 0.9373,
+      "step": 2610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.87042685881828e-07,
+      "loss": 0.938,
+      "step": 2611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.773539427084808e-07,
+      "loss": 0.9221,
+      "step": 2612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.677335354346059e-07,
+      "loss": 0.9771,
+      "step": 2613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.581814773891581e-07,
+      "loss": 0.8965,
+      "step": 2614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.486977818063955e-07,
+      "loss": 1.031,
+      "step": 2615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.392824618258519e-07,
+      "loss": 0.9824,
+      "step": 2616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.299355304923371e-07,
+      "loss": 0.9143,
+      "step": 2617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.206570007559031e-07,
+      "loss": 0.9648,
+      "step": 2618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.114468854718336e-07,
+      "loss": 0.9111,
+      "step": 2619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.023051974006322e-07,
+      "loss": 0.9016,
+      "step": 2620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.932319492079896e-07,
+      "loss": 0.9324,
+      "step": 2621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.842271534647726e-07,
+      "loss": 1.0212,
+      "step": 2622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.752908226470177e-07,
+      "loss": 0.9426,
+      "step": 2623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.664229691358935e-07,
+      "loss": 0.9209,
+      "step": 2624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.576236052176942e-07,
+      "loss": 0.9294,
+      "step": 2625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.488927430838286e-07,
+      "loss": 0.9075,
+      "step": 2626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.402303948308041e-07,
+      "loss": 0.9194,
+      "step": 2627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.316365724601813e-07,
+      "loss": 0.8965,
+      "step": 2628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.231112878785971e-07,
+      "loss": 0.8877,
+      "step": 2629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.14654552897731e-07,
+      "loss": 0.9897,
+      "step": 2630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.06266379234277e-07,
+      "loss": 0.96,
+      "step": 2631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.97946778509939e-07,
+      "loss": 0.9546,
+      "step": 2632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.896957622514298e-07,
+      "loss": 0.9404,
+      "step": 2633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.815133418904105e-07,
+      "loss": 0.8948,
+      "step": 2634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.733995287635351e-07,
+      "loss": 0.8633,
+      "step": 2635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.653543341123834e-07,
+      "loss": 0.9404,
+      "step": 2636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.573777690834669e-07,
+      "loss": 1.0063,
+      "step": 2637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4946984472822305e-07,
+      "loss": 0.9048,
+      "step": 2638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4163057200297677e-07,
+      "loss": 0.906,
+      "step": 2639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.338599617689343e-07,
+      "loss": 0.9165,
+      "step": 2640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2615802479218924e-07,
+      "loss": 0.9788,
+      "step": 2641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1852477174367244e-07,
+      "loss": 0.9294,
+      "step": 2642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.109602131991519e-07,
+      "loss": 1.0232,
+      "step": 2643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.034643596392385e-07,
+      "loss": 0.925,
+      "step": 2644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9603722144933575e-07,
+      "loss": 0.9565,
+      "step": 2645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.886788089196514e-07,
+      "loss": 0.9436,
+      "step": 2646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8138913224516906e-07,
+      "loss": 0.8567,
+      "step": 2647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.741682015256487e-07,
+      "loss": 0.9272,
+      "step": 2648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.670160267655931e-07,
+      "loss": 0.9509,
+      "step": 2649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.599326178742535e-07,
+      "loss": 0.9661,
+      "step": 2650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5291798466560167e-07,
+      "loss": 0.8176,
+      "step": 2651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4597213685831907e-07,
+      "loss": 1.0159,
+      "step": 2652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.390950840757967e-07,
+      "loss": 0.8901,
+      "step": 2653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.322868358460962e-07,
+      "loss": 0.9143,
+      "step": 2654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.255474016019666e-07,
+      "loss": 0.9607,
+      "step": 2655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.188767906807999e-07,
+      "loss": 0.97,
+      "step": 2656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1227501232465317e-07,
+      "loss": 0.9834,
+      "step": 2657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0574207568019874e-07,
+      "loss": 0.9065,
+      "step": 2658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9927798979874077e-07,
+      "loss": 0.9089,
+      "step": 2659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9288276363618197e-07,
+      "loss": 0.8965,
+      "step": 2660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8655640605303456e-07,
+      "loss": 0.9431,
+      "step": 2661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.802989258143818e-07,
+      "loss": 0.9492,
+      "step": 2662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7411033158988295e-07,
+      "loss": 0.8989,
+      "step": 2663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.679906319537628e-07,
+      "loss": 0.8506,
+      "step": 2664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.619398353847835e-07,
+      "loss": 0.887,
+      "step": 2665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.559579502662446e-07,
+      "loss": 0.947,
+      "step": 2666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.500449848859776e-07,
+      "loss": 0.9517,
+      "step": 2667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4420094743631273e-07,
+      "loss": 0.907,
+      "step": 2668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3842584601409533e-07,
+      "loss": 0.9414,
+      "step": 2669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3271968862065285e-07,
+      "loss": 0.9939,
+      "step": 2670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.270824831617946e-07,
+      "loss": 0.9456,
+      "step": 2671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2151423744780074e-07,
+      "loss": 0.9292,
+      "step": 2672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.160149591934002e-07,
+      "loss": 0.9377,
+      "step": 2673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.10584656017776e-07,
+      "loss": 0.9187,
+      "step": 2674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0522333544453764e-07,
+      "loss": 0.9338,
+      "step": 2675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9993100490173778e-07,
+      "loss": 0.9661,
+      "step": 2676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9470767172182212e-07,
+      "loss": 0.9033,
+      "step": 2677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8955334314166296e-07,
+      "loss": 0.9417,
+      "step": 2678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8446802630250892e-07,
+      "loss": 1.0098,
+      "step": 2679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7945172825000188e-07,
+      "loss": 0.9058,
+      "step": 2680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7450445593416577e-07,
+      "loss": 0.967,
+      "step": 2681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6962621620937313e-07,
+      "loss": 1.012,
+      "step": 2682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6481701583437869e-07,
+      "loss": 0.9773,
+      "step": 2683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6007686147225254e-07,
+      "loss": 0.9167,
+      "step": 2684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5540575969042458e-07,
+      "loss": 0.9243,
+      "step": 2685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.508037169606513e-07,
+      "loss": 0.8955,
+      "step": 2686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4627073965899908e-07,
+      "loss": 0.8989,
+      "step": 2687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.418068340658496e-07,
+      "loss": 0.9497,
+      "step": 2688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3741200636589458e-07,
+      "loss": 0.8994,
+      "step": 2689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.330862626481022e-07,
+      "loss": 0.9529,
+      "step": 2690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2882960890573948e-07,
+      "loss": 0.9421,
+      "step": 2691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2464205103634995e-07,
+      "loss": 0.9492,
+      "step": 2692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2052359484173715e-07,
+      "loss": 0.9651,
+      "step": 2693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1647424602797552e-07,
+      "loss": 0.9536,
+      "step": 2694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1249401020537731e-07,
+      "loss": 0.9746,
+      "step": 2695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0858289288851464e-07,
+      "loss": 0.9192,
+      "step": 2696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0474089949619181e-07,
+      "loss": 0.9534,
+      "step": 2697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0096803535143973e-07,
+      "loss": 0.9258,
+      "step": 2698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.726430568151035e-08,
+      "loss": 0.9744,
+      "step": 2699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.36297156178767e-08,
+      "loss": 0.9307,
+      "step": 2700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.006427019622176e-08,
+      "loss": 0.9197,
+      "step": 2701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.656797435642184e-08,
+      "loss": 0.9478,
+      "step": 2702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.314083294254649e-08,
+      "loss": 0.874,
+      "step": 2703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.978285070286417e-08,
+      "loss": 0.9587,
+      "step": 2704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.649403228980889e-08,
+      "loss": 0.9785,
+      "step": 2705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.327438226000238e-08,
+      "loss": 0.915,
+      "step": 2706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.01239050742264e-08,
+      "loss": 0.917,
+      "step": 2707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.704260509742266e-08,
+      "loss": 0.9968,
+      "step": 2708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.403048659870404e-08,
+      "loss": 0.9045,
+      "step": 2709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.108755375131004e-08,
+      "loss": 0.9939,
+      "step": 2710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.821381063264575e-08,
+      "loss": 0.8879,
+      "step": 2711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.540926122424295e-08,
+      "loss": 0.873,
+      "step": 2712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.26739094117712e-08,
+      "loss": 0.9219,
+      "step": 2713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0007758985021195e-08,
+      "loss": 0.8657,
+      "step": 2714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.741081363791589e-08,
+      "loss": 0.9644,
+      "step": 2715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4883076968482705e-08,
+      "loss": 0.8621,
+      "step": 2716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.242455247887578e-08,
+      "loss": 0.9265,
+      "step": 2717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0035243575342605e-08,
+      "loss": 0.9128,
+      "step": 2718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7715153568251835e-08,
+      "loss": 1.0061,
+      "step": 2719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.546428567204329e-08,
+      "loss": 0.8977,
+      "step": 2720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3282643005277945e-08,
+      "loss": 0.8989,
+      "step": 2721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1170228590593485e-08,
+      "loss": 0.9089,
+      "step": 2722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9127045354704342e-08,
+      "loss": 0.8484,
+      "step": 2723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7153096128423872e-08,
+      "loss": 0.9028,
+      "step": 2724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5248383646625516e-08,
+      "loss": 0.9348,
+      "step": 2725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.341291054827055e-08,
+      "loss": 0.9399,
+      "step": 2726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.164667937638587e-08,
+      "loss": 0.8936,
+      "step": 2727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9949692578058453e-08,
+      "loss": 0.8794,
+      "step": 2728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.832195250443536e-08,
+      "loss": 0.8887,
+      "step": 2729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6763461410740365e-08,
+      "loss": 0.897,
+      "step": 2730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.527422145624069e-08,
+      "loss": 0.9226,
+      "step": 2731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.385423470425251e-08,
+      "loss": 0.9341,
+      "step": 2732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2503503122157645e-08,
+      "loss": 0.9407,
+      "step": 2733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1222028581375776e-08,
+      "loss": 0.9495,
+      "step": 2734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0009812857370016e-08,
+      "loss": 0.9473,
+      "step": 2735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.866857629652448e-09,
+      "loss": 0.9727,
+      "step": 2736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.793164481773029e-09,
+      "loss": 0.9465,
+      "step": 2737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.788734901319593e-09,
+      "loss": 0.9753,
+      "step": 2738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.853570279917842e-09,
+      "loss": 0.916,
+      "step": 2739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9876719132258e-09,
+      "loss": 0.9324,
+      "step": 2740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.191041000933815e-09,
+      "loss": 0.9233,
+      "step": 2741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4636786467590056e-09,
+      "loss": 0.9585,
+      "step": 2742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.805585858461912e-09,
+      "loss": 0.8806,
+      "step": 2743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2167635478187454e-09,
+      "loss": 0.9377,
+      "step": 2744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.697212530632486e-09,
+      "loss": 0.97,
+      "step": 2745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2469335267384363e-09,
+      "loss": 0.8511,
+      "step": 2746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.659271599875674e-10,
+      "loss": 0.8904,
+      "step": 2747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.541939582631716e-10,
+      "loss": 0.9053,
+      "step": 2748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1173435346976144e-10,
+      "loss": 0.9753,
+      "step": 2749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.385486815219661e-10,
+      "loss": 0.9487,
+      "step": 2750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4637182377839086e-11,
+      "loss": 0.9016,
+      "step": 2751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.8884,
+      "step": 2752
+    },
+    {
+      "epoch": 1.0,
+      "step": 2752,
+      "total_flos": 1.7310806196197786e+17,
+      "train_loss": 1.0010608850523484,
+      "train_runtime": 21926.9796,
+      "train_samples_per_second": 16.068,
+      "train_steps_per_second": 0.126
+    }
+  ],
+  "max_steps": 2752,
+  "num_train_epochs": 1,
+  "total_flos": 1.7310806196197786e+17,
+  "trial_name": null,
+  "trial_params": null
+}