diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,91920 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 15315,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.347826086956522e-07,
+      "loss": 1.2676,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 1.3317,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3043478260869564e-06,
+      "loss": 1.2587,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 1.2735,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 1.2798,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.608695652173913e-06,
+      "loss": 1.2486,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0434782608695654e-06,
+      "loss": 1.2465,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 1.2379,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.91304347826087e-06,
+      "loss": 1.217,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 1.2083,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 1.1604,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.217391304347826e-06,
+      "loss": 1.1037,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 1.1213,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.086956521739131e-06,
+      "loss": 0.9929,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 0.953,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 0.9224,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.391304347826088e-06,
+      "loss": 0.8577,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 0.7801,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.26086956521739e-06,
+      "loss": 0.7861,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 0.6883,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 0.6404,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 0.5812,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 0.5099,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0434782608695651e-05,
+      "loss": 0.4558,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 0.4223,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 0.3943,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.173913043478261e-05,
+      "loss": 0.3893,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2173913043478261e-05,
+      "loss": 0.3496,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2608695652173914e-05,
+      "loss": 0.3104,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 0.2998,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3478260869565218e-05,
+      "loss": 0.2538,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 0.2686,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4347826086956522e-05,
+      "loss": 0.2387,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4782608695652176e-05,
+      "loss": 0.2158,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 0.2048,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 0.1892,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.608695652173913e-05,
+      "loss": 0.1833,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.652173913043478e-05,
+      "loss": 0.1863,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6956521739130433e-05,
+      "loss": 0.1817,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 0.1758,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.782608695652174e-05,
+      "loss": 0.1804,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 0.172,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8695652173913045e-05,
+      "loss": 0.1651,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 0.1606,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.956521739130435e-05,
+      "loss": 0.1543,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1645,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0434782608695654e-05,
+      "loss": 0.1618,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0869565217391303e-05,
+      "loss": 0.1586,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1304347826086958e-05,
+      "loss": 0.158,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.173913043478261e-05,
+      "loss": 0.1452,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2173913043478262e-05,
+      "loss": 0.1529,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2608695652173914e-05,
+      "loss": 0.1531,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3043478260869567e-05,
+      "loss": 0.149,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.347826086956522e-05,
+      "loss": 0.1461,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.391304347826087e-05,
+      "loss": 0.145,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4347826086956523e-05,
+      "loss": 0.1533,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4782608695652175e-05,
+      "loss": 0.1452,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5217391304347827e-05,
+      "loss": 0.147,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5652173913043483e-05,
+      "loss": 0.1471,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.608695652173913e-05,
+      "loss": 0.1445,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6521739130434787e-05,
+      "loss": 0.1436,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6956521739130436e-05,
+      "loss": 0.1387,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7391304347826085e-05,
+      "loss": 0.1395,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.782608695652174e-05,
+      "loss": 0.146,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.826086956521739e-05,
+      "loss": 0.1425,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8695652173913044e-05,
+      "loss": 0.1412,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9130434782608696e-05,
+      "loss": 0.1412,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9565217391304352e-05,
+      "loss": 0.1369,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-05,
+      "loss": 0.1405,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0434782608695656e-05,
+      "loss": 0.1424,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.086956521739131e-05,
+      "loss": 0.1414,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.130434782608696e-05,
+      "loss": 0.1447,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.173913043478261e-05,
+      "loss": 0.1352,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.217391304347826e-05,
+      "loss": 0.1376,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.260869565217392e-05,
+      "loss": 0.1368,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.304347826086956e-05,
+      "loss": 0.135,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.347826086956522e-05,
+      "loss": 0.1312,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3913043478260867e-05,
+      "loss": 0.1306,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4347826086956526e-05,
+      "loss": 0.1267,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.478260869565218e-05,
+      "loss": 0.1271,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.521739130434783e-05,
+      "loss": 0.1353,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.565217391304348e-05,
+      "loss": 0.1345,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6086956521739134e-05,
+      "loss": 0.1385,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6521739130434786e-05,
+      "loss": 0.1294,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.695652173913043e-05,
+      "loss": 0.1325,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.739130434782609e-05,
+      "loss": 0.1322,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7826086956521736e-05,
+      "loss": 0.1284,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8260869565217395e-05,
+      "loss": 0.1303,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.869565217391305e-05,
+      "loss": 0.1356,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.91304347826087e-05,
+      "loss": 0.1272,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.956521739130435e-05,
+      "loss": 0.1323,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 0.1315,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0434782608695655e-05,
+      "loss": 0.1329,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.086956521739131e-05,
+      "loss": 0.1273,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.130434782608696e-05,
+      "loss": 0.1394,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1739130434782605e-05,
+      "loss": 0.1326,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2173913043478264e-05,
+      "loss": 0.1273,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2608695652173916e-05,
+      "loss": 0.1246,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.304347826086957e-05,
+      "loss": 0.1315,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.347826086956522e-05,
+      "loss": 0.1259,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.391304347826087e-05,
+      "loss": 0.1388,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4347826086956525e-05,
+      "loss": 0.127,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.478260869565218e-05,
+      "loss": 0.1225,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.521739130434783e-05,
+      "loss": 0.1335,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.565217391304348e-05,
+      "loss": 0.1157,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.608695652173913e-05,
+      "loss": 0.1289,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6521739130434785e-05,
+      "loss": 0.1291,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.695652173913044e-05,
+      "loss": 0.1335,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.739130434782609e-05,
+      "loss": 0.1294,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782608695652174e-05,
+      "loss": 0.127,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8260869565217394e-05,
+      "loss": 0.1232,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8695652173913046e-05,
+      "loss": 0.1271,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.91304347826087e-05,
+      "loss": 0.1262,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.956521739130435e-05,
+      "loss": 0.1252,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-05,
+      "loss": 0.1308,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.0434782608695655e-05,
+      "loss": 0.1267,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.08695652173913e-05,
+      "loss": 0.1255,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.1304347826086966e-05,
+      "loss": 0.1232,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.173913043478261e-05,
+      "loss": 0.1283,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.217391304347826e-05,
+      "loss": 0.1265,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.260869565217391e-05,
+      "loss": 0.1293,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.3043478260869574e-05,
+      "loss": 0.1219,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.347826086956522e-05,
+      "loss": 0.1232,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.391304347826087e-05,
+      "loss": 0.1249,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.4347826086956524e-05,
+      "loss": 0.1293,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.478260869565217e-05,
+      "loss": 0.1247,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5217391304347835e-05,
+      "loss": 0.1206,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.565217391304348e-05,
+      "loss": 0.12,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.608695652173913e-05,
+      "loss": 0.1242,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.652173913043478e-05,
+      "loss": 0.1196,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.695652173913044e-05,
+      "loss": 0.1224,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.739130434782609e-05,
+      "loss": 0.116,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.782608695652174e-05,
+      "loss": 0.1261,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.826086956521739e-05,
+      "loss": 0.1249,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.869565217391305e-05,
+      "loss": 0.1258,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9130434782608704e-05,
+      "loss": 0.1256,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.956521739130435e-05,
+      "loss": 0.1285,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6e-05,
+      "loss": 0.1236,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0434782608695654e-05,
+      "loss": 0.1254,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.086956521739131e-05,
+      "loss": 0.119,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.130434782608696e-05,
+      "loss": 0.1236,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.173913043478262e-05,
+      "loss": 0.1259,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.217391304347826e-05,
+      "loss": 0.1247,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.260869565217392e-05,
+      "loss": 0.1268,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.304347826086957e-05,
+      "loss": 0.1164,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.347826086956523e-05,
+      "loss": 0.122,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.391304347826086e-05,
+      "loss": 0.1201,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.434782608695652e-05,
+      "loss": 0.1189,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.478260869565218e-05,
+      "loss": 0.1205,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.521739130434783e-05,
+      "loss": 0.1171,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.565217391304349e-05,
+      "loss": 0.1277,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.608695652173912e-05,
+      "loss": 0.1199,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.652173913043479e-05,
+      "loss": 0.1188,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.695652173913044e-05,
+      "loss": 0.1263,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.73913043478261e-05,
+      "loss": 0.1239,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.782608695652173e-05,
+      "loss": 0.1162,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.826086956521739e-05,
+      "loss": 0.1226,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.869565217391305e-05,
+      "loss": 0.1226,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.91304347826087e-05,
+      "loss": 0.1198,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.956521739130436e-05,
+      "loss": 0.1283,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7e-05,
+      "loss": 0.1122,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.043478260869566e-05,
+      "loss": 0.1232,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.086956521739131e-05,
+      "loss": 0.1206,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.130434782608696e-05,
+      "loss": 0.1227,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.17391304347826e-05,
+      "loss": 0.1189,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.217391304347827e-05,
+      "loss": 0.1181,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.260869565217392e-05,
+      "loss": 0.123,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.304347826086957e-05,
+      "loss": 0.1165,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.347826086956522e-05,
+      "loss": 0.1309,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.391304347826086e-05,
+      "loss": 0.119,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.434782608695653e-05,
+      "loss": 0.1209,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.478260869565218e-05,
+      "loss": 0.1184,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.521739130434783e-05,
+      "loss": 0.1242,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.565217391304347e-05,
+      "loss": 0.1149,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.608695652173914e-05,
+      "loss": 0.1224,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.652173913043479e-05,
+      "loss": 0.1309,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.695652173913044e-05,
+      "loss": 0.1179,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.73913043478261e-05,
+      "loss": 0.1252,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.782608695652173e-05,
+      "loss": 0.1174,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.82608695652174e-05,
+      "loss": 0.1149,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.869565217391305e-05,
+      "loss": 0.1165,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.91304347826087e-05,
+      "loss": 0.1169,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.956521739130434e-05,
+      "loss": 0.1212,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8e-05,
+      "loss": 0.1183,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.043478260869566e-05,
+      "loss": 0.1204,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.086956521739131e-05,
+      "loss": 0.1194,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.130434782608696e-05,
+      "loss": 0.1297,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.173913043478262e-05,
+      "loss": 0.1179,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.217391304347827e-05,
+      "loss": 0.1185,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.260869565217392e-05,
+      "loss": 0.1108,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.304347826086957e-05,
+      "loss": 0.1163,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.347826086956521e-05,
+      "loss": 0.1165,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.391304347826088e-05,
+      "loss": 0.119,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.434782608695653e-05,
+      "loss": 0.1185,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.478260869565218e-05,
+      "loss": 0.1184,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.521739130434783e-05,
+      "loss": 0.1166,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.565217391304348e-05,
+      "loss": 0.1185,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.608695652173914e-05,
+      "loss": 0.1178,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.652173913043479e-05,
+      "loss": 0.1223,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.695652173913044e-05,
+      "loss": 0.1135,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.739130434782609e-05,
+      "loss": 0.1117,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.782608695652174e-05,
+      "loss": 0.1179,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.82608695652174e-05,
+      "loss": 0.1158,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.869565217391305e-05,
+      "loss": 0.1093,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.91304347826087e-05,
+      "loss": 0.1152,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.956521739130435e-05,
+      "loss": 0.1162,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9e-05,
+      "loss": 0.1232,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.043478260869566e-05,
+      "loss": 0.1091,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.086956521739131e-05,
+      "loss": 0.1103,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.130434782608696e-05,
+      "loss": 0.1157,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.173913043478261e-05,
+      "loss": 0.1166,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.217391304347827e-05,
+      "loss": 0.1252,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.260869565217392e-05,
+      "loss": 0.1192,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.304347826086957e-05,
+      "loss": 0.1183,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.347826086956522e-05,
+      "loss": 0.1171,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.391304347826087e-05,
+      "loss": 0.1216,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.434782608695653e-05,
+      "loss": 0.1161,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.478260869565218e-05,
+      "loss": 0.1142,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.521739130434783e-05,
+      "loss": 0.1153,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.565217391304348e-05,
+      "loss": 0.1181,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.608695652173914e-05,
+      "loss": 0.1166,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.652173913043479e-05,
+      "loss": 0.1136,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.695652173913044e-05,
+      "loss": 0.1174,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.739130434782609e-05,
+      "loss": 0.1144,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.782608695652174e-05,
+      "loss": 0.1115,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.82608695652174e-05,
+      "loss": 0.1181,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869565217391305e-05,
+      "loss": 0.1204,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.91304347826087e-05,
+      "loss": 0.117,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956521739130435e-05,
+      "loss": 0.1202,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001,
+      "loss": 0.1227,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010043478260869566,
+      "loss": 0.114,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010086956521739131,
+      "loss": 0.1144,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010130434782608695,
+      "loss": 0.1136,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001017391304347826,
+      "loss": 0.1094,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010217391304347828,
+      "loss": 0.1131,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010260869565217393,
+      "loss": 0.1207,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010304347826086957,
+      "loss": 0.118,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010347826086956522,
+      "loss": 0.1151,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010391304347826087,
+      "loss": 0.1205,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010434782608695653,
+      "loss": 0.1137,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010478260869565218,
+      "loss": 0.1153,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010521739130434782,
+      "loss": 0.1173,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010565217391304347,
+      "loss": 0.1115,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010608695652173915,
+      "loss": 0.1204,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001065217391304348,
+      "loss": 0.1157,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010695652173913044,
+      "loss": 0.1137,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010739130434782609,
+      "loss": 0.1192,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010782608695652174,
+      "loss": 0.1203,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001082608695652174,
+      "loss": 0.1113,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010869565217391305,
+      "loss": 0.1125,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010913043478260869,
+      "loss": 0.1155,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010956521739130434,
+      "loss": 0.1289,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 0.1212,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011043478260869567,
+      "loss": 0.1166,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011086956521739131,
+      "loss": 0.1169,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011130434782608696,
+      "loss": 0.1188,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011173913043478261,
+      "loss": 0.1094,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011217391304347826,
+      "loss": 0.1123,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011260869565217392,
+      "loss": 0.1163,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011304347826086956,
+      "loss": 0.1178,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011347826086956523,
+      "loss": 0.117,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011391304347826089,
+      "loss": 0.1122,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011434782608695654,
+      "loss": 0.1124,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011478260869565218,
+      "loss": 0.1164,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011521739130434783,
+      "loss": 0.1131,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011565217391304348,
+      "loss": 0.1126,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011608695652173913,
+      "loss": 0.1149,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011652173913043479,
+      "loss": 0.1167,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011695652173913042,
+      "loss": 0.1164,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001173913043478261,
+      "loss": 0.1119,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011782608695652176,
+      "loss": 0.1161,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011826086956521741,
+      "loss": 0.1185,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011869565217391305,
+      "loss": 0.1166,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001191304347826087,
+      "loss": 0.1119,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011956521739130435,
+      "loss": 0.1166,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012,
+      "loss": 0.1126,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012043478260869566,
+      "loss": 0.1149,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012086956521739131,
+      "loss": 0.1129,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012130434782608697,
+      "loss": 0.1099,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012173913043478263,
+      "loss": 0.1145,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012217391304347828,
+      "loss": 0.1093,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012260869565217393,
+      "loss": 0.1198,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012304347826086958,
+      "loss": 0.1123,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012347826086956523,
+      "loss": 0.1147,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012391304347826086,
+      "loss": 0.1151,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001243478260869565,
+      "loss": 0.1105,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012478260869565216,
+      "loss": 0.1149,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012521739130434784,
+      "loss": 0.1197,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001256521739130435,
+      "loss": 0.1062,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012608695652173915,
+      "loss": 0.1131,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001265217391304348,
+      "loss": 0.1182,
+      "step": 291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012695652173913045,
+      "loss": 0.1138,
+      "step": 292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001273913043478261,
+      "loss": 0.1254,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012782608695652173,
+      "loss": 0.1065,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012826086956521738,
+      "loss": 0.1174,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012869565217391303,
+      "loss": 0.1154,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001291304347826087,
+      "loss": 0.112,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012956521739130436,
+      "loss": 0.1227,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 0.1134,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013043478260869567,
+      "loss": 0.1128,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013086956521739132,
+      "loss": 0.1084,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013130434782608697,
+      "loss": 0.1071,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001317391304347826,
+      "loss": 0.1104,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013217391304347825,
+      "loss": 0.1197,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001326086956521739,
+      "loss": 0.1153,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013304347826086958,
+      "loss": 0.1161,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013347826086956523,
+      "loss": 0.1127,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013391304347826088,
+      "loss": 0.1122,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013434782608695654,
+      "loss": 0.1083,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001347826086956522,
+      "loss": 0.1105,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013521739130434784,
+      "loss": 0.1124,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013565217391304347,
+      "loss": 0.1106,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013608695652173912,
+      "loss": 0.1148,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013652173913043477,
+      "loss": 0.111,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013695652173913045,
+      "loss": 0.1074,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001373913043478261,
+      "loss": 0.1135,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013782608695652175,
+      "loss": 0.1114,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001382608695652174,
+      "loss": 0.1075,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013869565217391306,
+      "loss": 0.1184,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001391304347826087,
+      "loss": 0.1165,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00013956521739130434,
+      "loss": 0.1173,
+      "step": 321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014,
+      "loss": 0.1176,
+      "step": 322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014043478260869567,
+      "loss": 0.1104,
+      "step": 323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014086956521739132,
+      "loss": 0.1146,
+      "step": 324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014130434782608697,
+      "loss": 0.1178,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014173913043478262,
+      "loss": 0.1083,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014217391304347828,
+      "loss": 0.1144,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014260869565217393,
+      "loss": 0.1129,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014304347826086958,
+      "loss": 0.1142,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001434782608695652,
+      "loss": 0.1139,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014391304347826086,
+      "loss": 0.1086,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014434782608695654,
+      "loss": 0.1203,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001447826086956522,
+      "loss": 0.1089,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014521739130434784,
+      "loss": 0.1187,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001456521739130435,
+      "loss": 0.1153,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014608695652173914,
+      "loss": 0.107,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001465217391304348,
+      "loss": 0.1148,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014695652173913045,
+      "loss": 0.1063,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014739130434782607,
+      "loss": 0.1082,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014782608695652173,
+      "loss": 0.1171,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001482608695652174,
+      "loss": 0.111,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014869565217391306,
+      "loss": 0.1135,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001491304347826087,
+      "loss": 0.1086,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014956521739130436,
+      "loss": 0.1129,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.1056,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015043478260869567,
+      "loss": 0.111,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015086956521739132,
+      "loss": 0.1145,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015130434782608694,
+      "loss": 0.1138,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001517391304347826,
+      "loss": 0.1043,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015217391304347827,
+      "loss": 0.1145,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015260869565217393,
+      "loss": 0.1136,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00015304347826086958,
+      "loss": 0.1151,
+      "step": 352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015347826086956523,
+      "loss": 0.1159,
+      "step": 353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015391304347826088,
+      "loss": 0.1122,
+      "step": 354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015434782608695654,
+      "loss": 0.1148,
+      "step": 355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001547826086956522,
+      "loss": 0.1143,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001552173913043478,
+      "loss": 0.1031,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015565217391304346,
+      "loss": 0.1212,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015608695652173914,
+      "loss": 0.111,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001565217391304348,
+      "loss": 0.1113,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015695652173913045,
+      "loss": 0.1101,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001573913043478261,
+      "loss": 0.1143,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015782608695652175,
+      "loss": 0.11,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001582608695652174,
+      "loss": 0.1128,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015869565217391306,
+      "loss": 0.1182,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015913043478260868,
+      "loss": 0.1142,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015956521739130433,
+      "loss": 0.1084,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016,
+      "loss": 0.1054,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016043478260869567,
+      "loss": 0.109,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016086956521739132,
+      "loss": 0.1065,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016130434782608697,
+      "loss": 0.1163,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016173913043478262,
+      "loss": 0.1115,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016217391304347827,
+      "loss": 0.1159,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016260869565217393,
+      "loss": 0.1129,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016304347826086955,
+      "loss": 0.106,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016347826086956523,
+      "loss": 0.113,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016391304347826088,
+      "loss": 0.1085,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016434782608695653,
+      "loss": 0.1123,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001647826086956522,
+      "loss": 0.108,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016521739130434784,
+      "loss": 0.1142,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001656521739130435,
+      "loss": 0.1089,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00016608695652173914,
+      "loss": 0.1108,
+      "step": 382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001665217391304348,
+      "loss": 0.1165,
+      "step": 383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00016695652173913042,
+      "loss": 0.1055,
+      "step": 384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001673913043478261,
+      "loss": 0.1054,
+      "step": 385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00016782608695652175,
+      "loss": 0.1081,
+      "step": 386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001682608695652174,
+      "loss": 0.1124,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00016869565217391306,
+      "loss": 0.1102,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001691304347826087,
+      "loss": 0.1085,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00016956521739130436,
+      "loss": 0.1092,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017,
+      "loss": 0.1103,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017043478260869566,
+      "loss": 0.1078,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017086956521739132,
+      "loss": 0.1143,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017130434782608697,
+      "loss": 0.1041,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017173913043478262,
+      "loss": 0.1128,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017217391304347827,
+      "loss": 0.1104,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017260869565217393,
+      "loss": 0.11,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017304347826086958,
+      "loss": 0.1068,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017347826086956523,
+      "loss": 0.1106,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017391304347826088,
+      "loss": 0.103,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017434782608695653,
+      "loss": 0.1086,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017478260869565219,
+      "loss": 0.113,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017521739130434784,
+      "loss": 0.1099,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001756521739130435,
+      "loss": 0.1032,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017608695652173914,
+      "loss": 0.1058,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001765217391304348,
+      "loss": 0.1086,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017695652173913045,
+      "loss": 0.1039,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001773913043478261,
+      "loss": 0.1123,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017782608695652175,
+      "loss": 0.1173,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001782608695652174,
+      "loss": 0.1049,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017869565217391305,
+      "loss": 0.1102,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001791304347826087,
+      "loss": 0.1088,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00017956521739130436,
+      "loss": 0.1009,
+      "step": 413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018,
+      "loss": 0.1097,
+      "step": 414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018043478260869566,
+      "loss": 0.115,
+      "step": 415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018086956521739132,
+      "loss": 0.112,
+      "step": 416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018130434782608697,
+      "loss": 0.1004,
+      "step": 417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018173913043478262,
+      "loss": 0.1137,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018217391304347827,
+      "loss": 0.1068,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018260869565217392,
+      "loss": 0.104,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018304347826086958,
+      "loss": 0.1122,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018347826086956523,
+      "loss": 0.1043,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018391304347826088,
+      "loss": 0.1059,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018434782608695653,
+      "loss": 0.1113,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018478260869565218,
+      "loss": 0.1136,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018521739130434784,
+      "loss": 0.1111,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001856521739130435,
+      "loss": 0.1099,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018608695652173914,
+      "loss": 0.1146,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001865217391304348,
+      "loss": 0.1103,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018695652173913045,
+      "loss": 0.1083,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001873913043478261,
+      "loss": 0.1082,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018782608695652175,
+      "loss": 0.1128,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001882608695652174,
+      "loss": 0.1135,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018869565217391305,
+      "loss": 0.1093,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001891304347826087,
+      "loss": 0.1074,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018956521739130436,
+      "loss": 0.1171,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019,
+      "loss": 0.1091,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019043478260869566,
+      "loss": 0.1052,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019086956521739131,
+      "loss": 0.1123,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019130434782608697,
+      "loss": 0.1049,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019173913043478262,
+      "loss": 0.1058,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019217391304347827,
+      "loss": 0.1078,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019260869565217392,
+      "loss": 0.1104,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019304347826086958,
+      "loss": 0.1096,
+      "step": 444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019347826086956523,
+      "loss": 0.1145,
+      "step": 445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019391304347826088,
+      "loss": 0.1114,
+      "step": 446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019434782608695653,
+      "loss": 0.1073,
+      "step": 447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019478260869565218,
+      "loss": 0.1121,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019521739130434784,
+      "loss": 0.108,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001956521739130435,
+      "loss": 0.1116,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019608695652173914,
+      "loss": 0.1129,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001965217391304348,
+      "loss": 0.108,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019695652173913044,
+      "loss": 0.1064,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001973913043478261,
+      "loss": 0.1018,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019782608695652175,
+      "loss": 0.1166,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982608695652174,
+      "loss": 0.1122,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019869565217391305,
+      "loss": 0.1178,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001991304347826087,
+      "loss": 0.1113,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019956521739130436,
+      "loss": 0.0997,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.1029,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999999776372897,
+      "loss": 0.1096,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999999105491602,
+      "loss": 0.1101,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999997987356144,
+      "loss": 0.1105,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999996421966567,
+      "loss": 0.1073,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001999999440932295,
+      "loss": 0.1135,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001999999194942538,
+      "loss": 0.1102,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999989042273963,
+      "loss": 0.1063,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001999998568786883,
+      "loss": 0.1063,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999981886210138,
+      "loss": 0.1101,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001999997763729805,
+      "loss": 0.1012,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999972941132758,
+      "loss": 0.1137,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999967797714474,
+      "loss": 0.1109,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019999962207043422,
+      "loss": 0.1074,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001999995616911986,
+      "loss": 0.1036,
+      "step": 474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999949683944051,
+      "loss": 0.1051,
+      "step": 475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999994275151629,
+      "loss": 0.1073,
+      "step": 476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999935371836888,
+      "loss": 0.1083,
+      "step": 477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999992754490617,
+      "loss": 0.1042,
+      "step": 478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999991927072449,
+      "loss": 0.1057,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999910549292214,
+      "loss": 0.1043,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999901380609737,
+      "loss": 0.1103,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999891764677465,
+      "loss": 0.1081,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999988170149583,
+      "loss": 0.1063,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999871191065286,
+      "loss": 0.1125,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999860233386294,
+      "loss": 0.1078,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999984882845935,
+      "loss": 0.0985,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999836976284966,
+      "loss": 0.0996,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999824676863667,
+      "loss": 0.1013,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999811930196005,
+      "loss": 0.1122,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999979873628255,
+      "loss": 0.1086,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999785095123893,
+      "loss": 0.1071,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999771006720644,
+      "loss": 0.1084,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999756471073435,
+      "loss": 0.1111,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999974148818291,
+      "loss": 0.1067,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999726058049746,
+      "loss": 0.1016,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999971018067463,
+      "loss": 0.1136,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999693856058271,
+      "loss": 0.1131,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999677084201397,
+      "loss": 0.1056,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999659865104768,
+      "loss": 0.1096,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999642198769143,
+      "loss": 0.1045,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001999962408519532,
+      "loss": 0.1106,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999605524384105,
+      "loss": 0.1049,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999586516336327,
+      "loss": 0.1099,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999567061052838,
+      "loss": 0.1061,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019999547158534511,
+      "loss": 0.1103,
+      "step": 505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999526808782233,
+      "loss": 0.1122,
+      "step": 506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999506011796913,
+      "loss": 0.1095,
+      "step": 507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999484767579487,
+      "loss": 0.1079,
+      "step": 508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999463076130896,
+      "loss": 0.1114,
+      "step": 509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999440937452118,
+      "loss": 0.107,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999418351544137,
+      "loss": 0.1054,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999395318407966,
+      "loss": 0.1085,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001999937183804464,
+      "loss": 0.1087,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999347910455205,
+      "loss": 0.1063,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999323535640725,
+      "loss": 0.1131,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000199992987136023,
+      "loss": 0.1094,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999273444341038,
+      "loss": 0.1125,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999247727858066,
+      "loss": 0.1148,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999221564154533,
+      "loss": 0.1032,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999194953231614,
+      "loss": 0.1025,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000199991678950905,
+      "loss": 0.1096,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999140389732393,
+      "loss": 0.1121,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999112437158533,
+      "loss": 0.1074,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001999908403737016,
+      "loss": 0.1071,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999055190368553,
+      "loss": 0.1082,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019999025896155003,
+      "loss": 0.1112,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001999899615473081,
+      "loss": 0.1112,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998965966097312,
+      "loss": 0.1166,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998935330255857,
+      "loss": 0.1037,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998904247207818,
+      "loss": 0.1093,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001999887271695458,
+      "loss": 0.1092,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001999884073949756,
+      "loss": 0.1077,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998808314838183,
+      "loss": 0.1095,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998775442977898,
+      "loss": 0.1126,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998742123918182,
+      "loss": 0.1063,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019998708357660516,
+      "loss": 0.1094,
+      "step": 536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999867414420642,
+      "loss": 0.1049,
+      "step": 537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998639483557415,
+      "loss": 0.1095,
+      "step": 538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998604375715058,
+      "loss": 0.1067,
+      "step": 539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998568820680919,
+      "loss": 0.106,
+      "step": 540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999853281845658,
+      "loss": 0.1065,
+      "step": 541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998496369043662,
+      "loss": 0.1081,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999845947244379,
+      "loss": 0.1096,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998422128658614,
+      "loss": 0.111,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998384337689804,
+      "loss": 0.1037,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998346099539053,
+      "loss": 0.1077,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998307414208067,
+      "loss": 0.1061,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999826828169858,
+      "loss": 0.1028,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998228702012343,
+      "loss": 0.1063,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999818867515112,
+      "loss": 0.1099,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999814820111671,
+      "loss": 0.1067,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998107279910914,
+      "loss": 0.0996,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998065911535568,
+      "loss": 0.1065,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019998024095992524,
+      "loss": 0.1074,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997981833283642,
+      "loss": 0.109,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997939123410824,
+      "loss": 0.1054,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997895966375974,
+      "loss": 0.1083,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997852362181026,
+      "loss": 0.1116,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997808310827928,
+      "loss": 0.1085,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997763812318645,
+      "loss": 0.1058,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997718866655178,
+      "loss": 0.111,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999767347383953,
+      "loss": 0.111,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997627633873733,
+      "loss": 0.1063,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997581346759837,
+      "loss": 0.1099,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999753461249991,
+      "loss": 0.1084,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019997487431096047,
+      "loss": 0.1147,
+      "step": 566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997439802550354,
+      "loss": 0.1029,
+      "step": 567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997391726864964,
+      "loss": 0.1079,
+      "step": 568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997343204042028,
+      "loss": 0.1048,
+      "step": 569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999729423408371,
+      "loss": 0.1092,
+      "step": 570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997244816992207,
+      "loss": 0.1145,
+      "step": 571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997194952769726,
+      "loss": 0.1034,
+      "step": 572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997144641418498,
+      "loss": 0.1071,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997093882940774,
+      "loss": 0.1192,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019997042677338822,
+      "loss": 0.1075,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996991024614935,
+      "loss": 0.1033,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999693892477142,
+      "loss": 0.1053,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996886377810608,
+      "loss": 0.1018,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996833383734855,
+      "loss": 0.1082,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996779942546523,
+      "loss": 0.1034,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996726054248004,
+      "loss": 0.1063,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999667171884171,
+      "loss": 0.1029,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996616936330072,
+      "loss": 0.1027,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996561706715538,
+      "loss": 0.0998,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996506030000577,
+      "loss": 0.1081,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996449906187685,
+      "loss": 0.1126,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996393335279368,
+      "loss": 0.1085,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996336317278153,
+      "loss": 0.1043,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996278852186597,
+      "loss": 0.1014,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996220940007265,
+      "loss": 0.1073,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999616258074275,
+      "loss": 0.1073,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999610377439566,
+      "loss": 0.1081,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019996044520968629,
+      "loss": 0.1007,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000199959848204643,
+      "loss": 0.1059,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019995924672885354,
+      "loss": 0.1018,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999586407823447,
+      "loss": 0.1071,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999580303651436,
+      "loss": 0.1083,
+      "step": 597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995741547727765,
+      "loss": 0.1087,
+      "step": 598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999567961187742,
+      "loss": 0.1026,
+      "step": 599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995617228966105,
+      "loss": 0.1076,
+      "step": 600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999555439899661,
+      "loss": 0.1051,
+      "step": 601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995491121971737,
+      "loss": 0.1053,
+      "step": 602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995427397894323,
+      "loss": 0.0997,
+      "step": 603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999536322676722,
+      "loss": 0.1086,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995298608593292,
+      "loss": 0.1061,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999523354337543,
+      "loss": 0.1036,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995168031116547,
+      "loss": 0.1028,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999510207181957,
+      "loss": 0.1053,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019995035665487456,
+      "loss": 0.1066,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994968812123165,
+      "loss": 0.1065,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994901511729695,
+      "loss": 0.1116,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999483376431005,
+      "loss": 0.1064,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994765569867267,
+      "loss": 0.1067,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999469692840439,
+      "loss": 0.1079,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999462783992449,
+      "loss": 0.1055,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994558304430658,
+      "loss": 0.1088,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994488321926004,
+      "loss": 0.1072,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999441789241366,
+      "loss": 0.1057,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994347015896773,
+      "loss": 0.1032,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994275692378514,
+      "loss": 0.1094,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994203921862075,
+      "loss": 0.1053,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999413170435066,
+      "loss": 0.1082,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019994059039847505,
+      "loss": 0.1065,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019993985928355858,
+      "loss": 0.1125,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999391236987899,
+      "loss": 0.1066,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019993838364420187,
+      "loss": 0.109,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019993763911982764,
+      "loss": 0.1078,
+      "step": 627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993689012570047,
+      "loss": 0.1071,
+      "step": 628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999361366618539,
+      "loss": 0.1013,
+      "step": 629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999353787283216,
+      "loss": 0.1032,
+      "step": 630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993461632513744,
+      "loss": 0.1055,
+      "step": 631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993384945233555,
+      "loss": 0.1095,
+      "step": 632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993307810995027,
+      "loss": 0.1049,
+      "step": 633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993230229801604,
+      "loss": 0.1047,
+      "step": 634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993152201656757,
+      "loss": 0.1109,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019993073726563978,
+      "loss": 0.1091,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019992994804526775,
+      "loss": 0.1075,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019992915435548678,
+      "loss": 0.1033,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999283561963324,
+      "loss": 0.1032,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019992755356784025,
+      "loss": 0.1011,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019992674647004624,
+      "loss": 0.1099,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999259349029865,
+      "loss": 0.1135,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019992511886669735,
+      "loss": 0.1058,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999242983612152,
+      "loss": 0.1025,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999234733865768,
+      "loss": 0.1071,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019992264394281904,
+      "loss": 0.1025,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019992181002997909,
+      "loss": 0.1035,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999209716480941,
+      "loss": 0.111,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999201287972017,
+      "loss": 0.1075,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999192814773395,
+      "loss": 0.0994,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019991842968854543,
+      "loss": 0.1075,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019991757343085757,
+      "loss": 0.1058,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019991671270431428,
+      "loss": 0.1073,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019991584750895397,
+      "loss": 0.1152,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019991497784481538,
+      "loss": 0.1055,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999141037119374,
+      "loss": 0.1034,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019991322511035914,
+      "loss": 0.1034,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999123420401199,
+      "loss": 0.1021,
+      "step": 658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019991145450125913,
+      "loss": 0.1116,
+      "step": 659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019991056249381656,
+      "loss": 0.1102,
+      "step": 660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990966601783207,
+      "loss": 0.1076,
+      "step": 661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990876507334577,
+      "loss": 0.1,
+      "step": 662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990785966039794,
+      "loss": 0.1044,
+      "step": 663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990694977902913,
+      "loss": 0.1092,
+      "step": 664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990603542927997,
+      "loss": 0.112,
+      "step": 665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990511661119136,
+      "loss": 0.1045,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990419332480442,
+      "loss": 0.11,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990326557016043,
+      "loss": 0.1072,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990233334730086,
+      "loss": 0.1042,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019990139665626745,
+      "loss": 0.1041,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001999004554971021,
+      "loss": 0.107,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989950986984686,
+      "loss": 0.1029,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989855977454405,
+      "loss": 0.1036,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989760521123615,
+      "loss": 0.1106,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989664617996587,
+      "loss": 0.1092,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989568268077607,
+      "loss": 0.1031,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001998947147137099,
+      "loss": 0.1073,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989374227881058,
+      "loss": 0.1045,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989276537612166,
+      "loss": 0.1037,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001998917840056868,
+      "loss": 0.1095,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019989079816754992,
+      "loss": 0.1098,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019988980786175508,
+      "loss": 0.1052,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001998888130883466,
+      "loss": 0.0997,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000199887813847369,
+      "loss": 0.1143,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019988681013886688,
+      "loss": 0.106,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019988580196288517,
+      "loss": 0.1029,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019988478931946902,
+      "loss": 0.1081,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019988377220866367,
+      "loss": 0.1001,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001998827506305146,
+      "loss": 0.0973,
+      "step": 689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019988172458506754,
+      "loss": 0.1042,
+      "step": 690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998806940723683,
+      "loss": 0.1079,
+      "step": 691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998796590924631,
+      "loss": 0.1054,
+      "step": 692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998786196453981,
+      "loss": 0.1065,
+      "step": 693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019987757573121988,
+      "loss": 0.1053,
+      "step": 694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998765273499751,
+      "loss": 0.1056,
+      "step": 695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019987547450171062,
+      "loss": 0.1094,
+      "step": 696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019987441718647358,
+      "loss": 0.1073,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019987335540431122,
+      "loss": 0.0972,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019987228915527107,
+      "loss": 0.1079,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998712184394008,
+      "loss": 0.102,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019987014325674828,
+      "loss": 0.1053,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986906360736163,
+      "loss": 0.1088,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986797949128913,
+      "loss": 0.0995,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986689090857926,
+      "loss": 0.0994,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998657978592807,
+      "loss": 0.1025,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986470034344235,
+      "loss": 0.1034,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986359836111334,
+      "loss": 0.1068,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019986249191234284,
+      "loss": 0.1092,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998613809971804,
+      "loss": 0.1053,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998602656156758,
+      "loss": 0.1096,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985914576787876,
+      "loss": 0.1011,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985802145383946,
+      "loss": 0.1045,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985689267360817,
+      "loss": 0.1029,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001998557594272354,
+      "loss": 0.1037,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985462171477177,
+      "loss": 0.11,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985347953626828,
+      "loss": 0.1034,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985233289177584,
+      "loss": 0.1033,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985118178134588,
+      "loss": 0.1055,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019985002620502985,
+      "loss": 0.1059,
+      "step": 719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001998488661628794,
+      "loss": 0.1103,
+      "step": 720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019984770165494645,
+      "loss": 0.1046,
+      "step": 721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019984653268128306,
+      "loss": 0.1045,
+      "step": 722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019984535924194154,
+      "loss": 0.1046,
+      "step": 723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019984418133697432,
+      "loss": 0.1086,
+      "step": 724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001998429989664341,
+      "loss": 0.1044,
+      "step": 725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019984181213037383,
+      "loss": 0.1064,
+      "step": 726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019984062082884649,
+      "loss": 0.0978,
+      "step": 727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983942506190546,
+      "loss": 0.1074,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983822482960413,
+      "loss": 0.1056,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983702013199624,
+      "loss": 0.1074,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983581096913564,
+      "loss": 0.1022,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983459734107642,
+      "loss": 0.1025,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983337924787286,
+      "loss": 0.1125,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983215668957946,
+      "loss": 0.1021,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019983092966625087,
+      "loss": 0.102,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019982969817794203,
+      "loss": 0.102,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019982846222470793,
+      "loss": 0.1025,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019982722180660387,
+      "loss": 0.1059,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019982597692368536,
+      "loss": 0.1036,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019982472757600807,
+      "loss": 0.107,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019982347376362787,
+      "loss": 0.1042,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019982221548660084,
+      "loss": 0.0994,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019982095274498324,
+      "loss": 0.1058,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001998196855388316,
+      "loss": 0.1029,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019981841386820253,
+      "loss": 0.1031,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019981713773315295,
+      "loss": 0.1065,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019981585713373993,
+      "loss": 0.105,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019981457207002073,
+      "loss": 0.1078,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019981328254205282,
+      "loss": 0.1088,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019981198854989387,
+      "loss": 0.1095,
+      "step": 750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001998106900936018,
+      "loss": 0.1066,
+      "step": 751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019980938717323465,
+      "loss": 0.104,
+      "step": 752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001998080797888507,
+      "loss": 0.0999,
+      "step": 753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001998067679405084,
+      "loss": 0.1093,
+      "step": 754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019980545162826647,
+      "loss": 0.1019,
+      "step": 755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019980413085218376,
+      "loss": 0.1069,
+      "step": 756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019980280561231932,
+      "loss": 0.1069,
+      "step": 757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019980147590873247,
+      "loss": 0.1044,
+      "step": 758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019980014174148262,
+      "loss": 0.1016,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019979880311062949,
+      "loss": 0.1019,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001997974600162329,
+      "loss": 0.1035,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000199796112458353,
+      "loss": 0.1114,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019979476043705002,
+      "loss": 0.107,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019979340395238443,
+      "loss": 0.1073,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019979204300441686,
+      "loss": 0.1032,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019979067759320822,
+      "loss": 0.1027,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019978930771881955,
+      "loss": 0.1059,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001997879333813122,
+      "loss": 0.1059,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019978655458074754,
+      "loss": 0.1059,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019978517131718728,
+      "loss": 0.1107,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019978378359069326,
+      "loss": 0.1089,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019978239140132763,
+      "loss": 0.1014,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019978099474915256,
+      "loss": 0.1084,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019977959363423053,
+      "loss": 0.1066,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019977818805662424,
+      "loss": 0.1051,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019977677801639657,
+      "loss": 0.1075,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019977536351361053,
+      "loss": 0.1052,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001997739445483294,
+      "loss": 0.102,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019977252112061668,
+      "loss": 0.1006,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000199771093230536,
+      "loss": 0.104,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019976966087815123,
+      "loss": 0.1029,
+      "step": 781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019976822406352646,
+      "loss": 0.0978,
+      "step": 782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019976678278672589,
+      "loss": 0.1027,
+      "step": 783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019976533704781402,
+      "loss": 0.0999,
+      "step": 784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019976388684685552,
+      "loss": 0.1008,
+      "step": 785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019976243218391523,
+      "loss": 0.1074,
+      "step": 786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019976097305905824,
+      "loss": 0.103,
+      "step": 787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019975950947234976,
+      "loss": 0.1048,
+      "step": 788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997580414238553,
+      "loss": 0.1017,
+      "step": 789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997565689136405,
+      "loss": 0.0996,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019975509194177123,
+      "loss": 0.11,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997536105083135,
+      "loss": 0.0997,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997521246133336,
+      "loss": 0.0981,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019975063425689804,
+      "loss": 0.0978,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997491394390734,
+      "loss": 0.1031,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019974764015992654,
+      "loss": 0.1092,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019974613641952454,
+      "loss": 0.1032,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019974462821793472,
+      "loss": 0.099,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997431155552244,
+      "loss": 0.1051,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019974159843146134,
+      "loss": 0.1019,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019974007684671335,
+      "loss": 0.1037,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997385508010485,
+      "loss": 0.1048,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000199737020294535,
+      "loss": 0.1064,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019973548532724138,
+      "loss": 0.1077,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019973394589923622,
+      "loss": 0.1073,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997324020105884,
+      "loss": 0.1028,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019973085366136698,
+      "loss": 0.1054,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997293008516412,
+      "loss": 0.1041,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019972774358148054,
+      "loss": 0.1008,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001997261818509546,
+      "loss": 0.1038,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019972461566013325,
+      "loss": 0.1048,
+      "step": 811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019972304500908656,
+      "loss": 0.1062,
+      "step": 812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019972146989788474,
+      "loss": 0.1031,
+      "step": 813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019971989032659826,
+      "loss": 0.1072,
+      "step": 814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019971830629529776,
+      "loss": 0.1019,
+      "step": 815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019971671780405408,
+      "loss": 0.1066,
+      "step": 816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019971512485293829,
+      "loss": 0.1011,
+      "step": 817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001997135274420216,
+      "loss": 0.1078,
+      "step": 818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019971192557137547,
+      "loss": 0.1073,
+      "step": 819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001997103192410716,
+      "loss": 0.1022,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019970870845118172,
+      "loss": 0.1005,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019970709320177798,
+      "loss": 0.1057,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019970547349293258,
+      "loss": 0.1004,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019970384932471795,
+      "loss": 0.1063,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019970222069720675,
+      "loss": 0.1034,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019970058761047178,
+      "loss": 0.111,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019969895006458614,
+      "loss": 0.1033,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019969730805962304,
+      "loss": 0.1067,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019969566159565592,
+      "loss": 0.1044,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019969401067275846,
+      "loss": 0.101,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019969235529100442,
+      "loss": 0.1073,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001996906954504679,
+      "loss": 0.1019,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019968903115122309,
+      "loss": 0.1,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019968736239334446,
+      "loss": 0.1014,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019968568917690663,
+      "loss": 0.1086,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019968401150198447,
+      "loss": 0.1051,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019968232936865297,
+      "loss": 0.1088,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001996806427769874,
+      "loss": 0.1076,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019967895172706316,
+      "loss": 0.1052,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001996772562189559,
+      "loss": 0.1048,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019967555625274145,
+      "loss": 0.0986,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019967385182849585,
+      "loss": 0.106,
+      "step": 842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019967214294629534,
+      "loss": 0.1013,
+      "step": 843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996704296062163,
+      "loss": 0.1018,
+      "step": 844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996687118083354,
+      "loss": 0.1038,
+      "step": 845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019966698955272948,
+      "loss": 0.1027,
+      "step": 846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019966526283947552,
+      "loss": 0.1027,
+      "step": 847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019966353166865083,
+      "loss": 0.1027,
+      "step": 848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019966179604033276,
+      "loss": 0.1059,
+      "step": 849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019966005595459897,
+      "loss": 0.1006,
+      "step": 850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019965831141152725,
+      "loss": 0.1092,
+      "step": 851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019965656241119567,
+      "loss": 0.1022,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019965480895368245,
+      "loss": 0.104,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000199653051039066,
+      "loss": 0.1062,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019965128866742494,
+      "loss": 0.1037,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996495218388381,
+      "loss": 0.0997,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019964775055338453,
+      "loss": 0.1061,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996459748111434,
+      "loss": 0.1033,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019964419461219416,
+      "loss": 0.1074,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019964240995661646,
+      "loss": 0.1112,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019964062084449005,
+      "loss": 0.1105,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019963882727589498,
+      "loss": 0.1084,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019963702925091147,
+      "loss": 0.098,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019963522676961994,
+      "loss": 0.1058,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000199633419832101,
+      "loss": 0.1054,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019963160843843549,
+      "loss": 0.1077,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996297925887044,
+      "loss": 0.1126,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019962797228298895,
+      "loss": 0.0983,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019962614752137056,
+      "loss": 0.1117,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019962431830393085,
+      "loss": 0.1043,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001996224846307516,
+      "loss": 0.1061,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019962064650191484,
+      "loss": 0.1032,
+      "step": 872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019961880391750278,
+      "loss": 0.0962,
+      "step": 873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019961695687759784,
+      "loss": 0.1098,
+      "step": 874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019961510538228264,
+      "loss": 0.1051,
+      "step": 875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019961324943163996,
+      "loss": 0.1043,
+      "step": 876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019961138902575284,
+      "loss": 0.1068,
+      "step": 877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019960952416470446,
+      "loss": 0.1069,
+      "step": 878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019960765484857822,
+      "loss": 0.1037,
+      "step": 879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019960578107745775,
+      "loss": 0.109,
+      "step": 880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019960390285142686,
+      "loss": 0.1082,
+      "step": 881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019960202017056954,
+      "loss": 0.102,
+      "step": 882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019960013303496998,
+      "loss": 0.1068,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001995982414447126,
+      "loss": 0.1027,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000199596345399882,
+      "loss": 0.0984,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000199594444900563,
+      "loss": 0.1007,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019959253994684053,
+      "loss": 0.1028,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019959063053879986,
+      "loss": 0.1077,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019958871667652642,
+      "loss": 0.108,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019958679836010572,
+      "loss": 0.1037,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001995848755896236,
+      "loss": 0.0989,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019958294836516606,
+      "loss": 0.1078,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019958101668681927,
+      "loss": 0.1029,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019957908055466967,
+      "loss": 0.1061,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019957713996880382,
+      "loss": 0.1034,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019957519492930855,
+      "loss": 0.0979,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001995732454362708,
+      "loss": 0.1081,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019957129148977777,
+      "loss": 0.1018,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019956933308991687,
+      "loss": 0.1031,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001995673702367757,
+      "loss": 0.1087,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019956540293044204,
+      "loss": 0.0964,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019956343117100386,
+      "loss": 0.0986,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019956145495854939,
+      "loss": 0.1051,
+      "step": 903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019955947429316697,
+      "loss": 0.1034,
+      "step": 904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995574891749452,
+      "loss": 0.1027,
+      "step": 905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019955549960397288,
+      "loss": 0.0966,
+      "step": 906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019955350558033898,
+      "loss": 0.1051,
+      "step": 907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019955150710413273,
+      "loss": 0.1109,
+      "step": 908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019954950417544343,
+      "loss": 0.1061,
+      "step": 909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019954749679436073,
+      "loss": 0.11,
+      "step": 910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019954548496097435,
+      "loss": 0.1042,
+      "step": 911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019954346867537436,
+      "loss": 0.1031,
+      "step": 912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019954144793765086,
+      "loss": 0.1045,
+      "step": 913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019953942274789427,
+      "loss": 0.1059,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995373931061951,
+      "loss": 0.0997,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019953535901264422,
+      "loss": 0.0934,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019953332046733255,
+      "loss": 0.1022,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019953127747035127,
+      "loss": 0.1028,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019952923002179178,
+      "loss": 0.106,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019952717812174559,
+      "loss": 0.1037,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019952512177030454,
+      "loss": 0.0992,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019952306096756062,
+      "loss": 0.0987,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019952099571360593,
+      "loss": 0.1043,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019951892600853288,
+      "loss": 0.0995,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000199516851852434,
+      "loss": 0.1066,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995147732454021,
+      "loss": 0.1107,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019951269018753012,
+      "loss": 0.1018,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019951060267891124,
+      "loss": 0.1034,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019950851071963882,
+      "loss": 0.0924,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019950641430980644,
+      "loss": 0.1047,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001995043134495078,
+      "loss": 0.1052,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019950220813883695,
+      "loss": 0.1044,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000199500098377888,
+      "loss": 0.1009,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019949798416675534,
+      "loss": 0.1041,
+      "step": 934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019949586550553348,
+      "loss": 0.1055,
+      "step": 935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019949374239431721,
+      "loss": 0.1041,
+      "step": 936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019949161483320149,
+      "loss": 0.1038,
+      "step": 937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994894828222815,
+      "loss": 0.0988,
+      "step": 938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994873463616525,
+      "loss": 0.1062,
+      "step": 939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019948520545141015,
+      "loss": 0.0988,
+      "step": 940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019948306009165014,
+      "loss": 0.1027,
+      "step": 941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994809102824685,
+      "loss": 0.1042,
+      "step": 942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019947875602396127,
+      "loss": 0.11,
+      "step": 943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019947659731622488,
+      "loss": 0.0965,
+      "step": 944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019947443415935584,
+      "loss": 0.1013,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994722665534509,
+      "loss": 0.1027,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019947009449860703,
+      "loss": 0.0978,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994679179949214,
+      "loss": 0.103,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019946573704249126,
+      "loss": 0.1019,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019946355164141428,
+      "loss": 0.1014,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994613617917881,
+      "loss": 0.1028,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994591674937107,
+      "loss": 0.1099,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019945696874728024,
+      "loss": 0.0967,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019945476555259504,
+      "loss": 0.1018,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994525579097536,
+      "loss": 0.1065,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019945034581885472,
+      "loss": 0.0976,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019944812927999732,
+      "loss": 0.1077,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019944590829328057,
+      "loss": 0.102,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019944368285880368,
+      "loss": 0.0979,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019944145297666634,
+      "loss": 0.101,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019943921864696814,
+      "loss": 0.1069,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019943697986980917,
+      "loss": 0.1054,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001994347366452894,
+      "loss": 0.1006,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019943248897350926,
+      "loss": 0.0987,
+      "step": 964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019943023685456925,
+      "loss": 0.1048,
+      "step": 965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019942798028857009,
+      "loss": 0.0991,
+      "step": 966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019942571927561269,
+      "loss": 0.1032,
+      "step": 967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019942345381579823,
+      "loss": 0.1017,
+      "step": 968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019942118390922796,
+      "loss": 0.105,
+      "step": 969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001994189095560035,
+      "loss": 0.1083,
+      "step": 970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019941663075622644,
+      "loss": 0.1054,
+      "step": 971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019941434750999886,
+      "loss": 0.1032,
+      "step": 972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019941205981742274,
+      "loss": 0.1047,
+      "step": 973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019940976767860046,
+      "loss": 0.106,
+      "step": 974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019940747109363453,
+      "loss": 0.1057,
+      "step": 975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019940517006262766,
+      "loss": 0.1087,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019940286458568278,
+      "loss": 0.1006,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019940055466290298,
+      "loss": 0.1009,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019939824029439158,
+      "loss": 0.1053,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001993959214802521,
+      "loss": 0.1015,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019939359822058824,
+      "loss": 0.1018,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019939127051550392,
+      "loss": 0.1024,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019938893836510325,
+      "loss": 0.1029,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019938660176949054,
+      "loss": 0.1022,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019938426072877027,
+      "loss": 0.1029,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019938191524304712,
+      "loss": 0.1056,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019937956531242606,
+      "loss": 0.1048,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019937721093701217,
+      "loss": 0.1044,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019937485211691077,
+      "loss": 0.1015,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019937248885222727,
+      "loss": 0.0993,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019937012114306746,
+      "loss": 0.1043,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001993677489895372,
+      "loss": 0.0997,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001993653723917426,
+      "loss": 0.1095,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019936299134978996,
+      "loss": 0.1081,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019936060586378572,
+      "loss": 0.1001,
+      "step": 995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019935821593383663,
+      "loss": 0.1011,
+      "step": 996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019935582156004958,
+      "loss": 0.102,
+      "step": 997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019935342274253161,
+      "loss": 0.1007,
+      "step": 998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019935101948139006,
+      "loss": 0.1024,
+      "step": 999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001993486117767324,
+      "loss": 0.0987,
+      "step": 1000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019934619962866628,
+      "loss": 0.1053,
+      "step": 1001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019934378303729963,
+      "loss": 0.1084,
+      "step": 1002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019934136200274054,
+      "loss": 0.1065,
+      "step": 1003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019933893652509727,
+      "loss": 0.0996,
+      "step": 1004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001993365066044783,
+      "loss": 0.1075,
+      "step": 1005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019933407224099228,
+      "loss": 0.1052,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019933163343474814,
+      "loss": 0.108,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019932919018585495,
+      "loss": 0.1051,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019932674249442193,
+      "loss": 0.1071,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001993242903605586,
+      "loss": 0.1032,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019932183378437466,
+      "loss": 0.1026,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019931937276597994,
+      "loss": 0.1096,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019931690730548451,
+      "loss": 0.1044,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019931443740299866,
+      "loss": 0.1021,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019931196305863282,
+      "loss": 0.101,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019930948427249773,
+      "loss": 0.1014,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019930700104470414,
+      "loss": 0.1005,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019930451337536322,
+      "loss": 0.1035,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019930202126458617,
+      "loss": 0.1053,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019929952471248448,
+      "loss": 0.1018,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001992970237191698,
+      "loss": 0.0976,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019929451828475402,
+      "loss": 0.1101,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019929200840934912,
+      "loss": 0.1077,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019928949409306744,
+      "loss": 0.1045,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019928697533602135,
+      "loss": 0.098,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001992844521383236,
+      "loss": 0.1048,
+      "step": 1026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019928192450008694,
+      "loss": 0.1006,
+      "step": 1027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992793924214245,
+      "loss": 0.0983,
+      "step": 1028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019927685590244947,
+      "loss": 0.0983,
+      "step": 1029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019927431494327536,
+      "loss": 0.1009,
+      "step": 1030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019927176954401575,
+      "loss": 0.1028,
+      "step": 1031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019926921970478452,
+      "loss": 0.1021,
+      "step": 1032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992666654256957,
+      "loss": 0.1049,
+      "step": 1033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019926410670686355,
+      "loss": 0.1011,
+      "step": 1034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019926154354840247,
+      "loss": 0.1056,
+      "step": 1035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019925897595042717,
+      "loss": 0.1028,
+      "step": 1036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992564039130524,
+      "loss": 0.0989,
+      "step": 1037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019925382743639326,
+      "loss": 0.1079,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019925124652056495,
+      "loss": 0.1033,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019924866116568293,
+      "loss": 0.095,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992460713718628,
+      "loss": 0.0999,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992434771392204,
+      "loss": 0.1074,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992408784678718,
+      "loss": 0.0976,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019923827535793312,
+      "loss": 0.1006,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992356678095209,
+      "loss": 0.1005,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019923305582275172,
+      "loss": 0.1032,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992304393977424,
+      "loss": 0.1026,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019922781853460992,
+      "loss": 0.1041,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019922519323347157,
+      "loss": 0.1035,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019922256349444477,
+      "loss": 0.0992,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019921992931764705,
+      "loss": 0.1069,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992172907031963,
+      "loss": 0.0985,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019921464765121054,
+      "loss": 0.1,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019921200016180792,
+      "loss": 0.1014,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001992093482351069,
+      "loss": 0.0985,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019920669187122606,
+      "loss": 0.1008,
+      "step": 1056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001992040310702842,
+      "loss": 0.1009,
+      "step": 1057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019920136583240038,
+      "loss": 0.1068,
+      "step": 1058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019919869615769376,
+      "loss": 0.1072,
+      "step": 1059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019919602204628375,
+      "loss": 0.1078,
+      "step": 1060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019919334349828994,
+      "loss": 0.0961,
+      "step": 1061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019919066051383216,
+      "loss": 0.1043,
+      "step": 1062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019918797309303036,
+      "loss": 0.1042,
+      "step": 1063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001991852812360048,
+      "loss": 0.1007,
+      "step": 1064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001991825849428758,
+      "loss": 0.0995,
+      "step": 1065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019917988421376404,
+      "loss": 0.1063,
+      "step": 1066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019917717904879023,
+      "loss": 0.104,
+      "step": 1067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001991744694480754,
+      "loss": 0.1083,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019917175541174072,
+      "loss": 0.1081,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019916903693990758,
+      "loss": 0.1046,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019916631403269758,
+      "loss": 0.1044,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001991635866902325,
+      "loss": 0.096,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019916085491263435,
+      "loss": 0.1073,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019915811870002524,
+      "loss": 0.1013,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019915537805252758,
+      "loss": 0.105,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019915263297026395,
+      "loss": 0.098,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019914988345335713,
+      "loss": 0.1038,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019914712950193012,
+      "loss": 0.0991,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019914437111610603,
+      "loss": 0.1085,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019914160829600825,
+      "loss": 0.102,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019913884104176037,
+      "loss": 0.1048,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019913606935348616,
+      "loss": 0.1108,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019913329323130956,
+      "loss": 0.0948,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019913051267535475,
+      "loss": 0.1009,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019912772768574608,
+      "loss": 0.0978,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019912493826260814,
+      "loss": 0.1034,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019912214440606562,
+      "loss": 0.1087,
+      "step": 1087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019911934611624354,
+      "loss": 0.1037,
+      "step": 1088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019911654339326704,
+      "loss": 0.1045,
+      "step": 1089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019911373623726148,
+      "loss": 0.107,
+      "step": 1090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019911092464835236,
+      "loss": 0.1074,
+      "step": 1091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001991081086266655,
+      "loss": 0.1024,
+      "step": 1092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001991052881723268,
+      "loss": 0.1023,
+      "step": 1093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019910246328546245,
+      "loss": 0.1003,
+      "step": 1094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019909963396619876,
+      "loss": 0.0954,
+      "step": 1095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019909680021466225,
+      "loss": 0.1064,
+      "step": 1096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019909396203097969,
+      "loss": 0.0975,
+      "step": 1097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019909111941527805,
+      "loss": 0.0995,
+      "step": 1098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001990882723676844,
+      "loss": 0.1016,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019908542088832617,
+      "loss": 0.1004,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019908256497733076,
+      "loss": 0.1024,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000199079704634826,
+      "loss": 0.109,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019907683986093984,
+      "loss": 0.106,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019907397065580033,
+      "loss": 0.0985,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019907109701953584,
+      "loss": 0.101,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019906821895227488,
+      "loss": 0.1014,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019906533645414615,
+      "loss": 0.1016,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019906244952527864,
+      "loss": 0.1043,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001990595581658014,
+      "loss": 0.1047,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019905666237584382,
+      "loss": 0.1052,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019905376215553533,
+      "loss": 0.1011,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001990508575050057,
+      "loss": 0.106,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019904794842438484,
+      "loss": 0.0976,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019904503491380285,
+      "loss": 0.1115,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019904211697338998,
+      "loss": 0.1103,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019903919460327686,
+      "loss": 0.1016,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001990362678035941,
+      "loss": 0.1041,
+      "step": 1117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019903333657447257,
+      "loss": 0.0989,
+      "step": 1118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001990304009160435,
+      "loss": 0.1048,
+      "step": 1119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001990274608284381,
+      "loss": 0.0982,
+      "step": 1120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019902451631178788,
+      "loss": 0.1039,
+      "step": 1121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019902156736622452,
+      "loss": 0.098,
+      "step": 1122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019901861399187996,
+      "loss": 0.1018,
+      "step": 1123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019901565618888625,
+      "loss": 0.1064,
+      "step": 1124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019901269395737568,
+      "loss": 0.1043,
+      "step": 1125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019900972729748072,
+      "loss": 0.1042,
+      "step": 1126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019900675620933416,
+      "loss": 0.1025,
+      "step": 1127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019900378069306874,
+      "loss": 0.0918,
+      "step": 1128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019900080074881763,
+      "loss": 0.0976,
+      "step": 1129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019899781637671408,
+      "loss": 0.1005,
+      "step": 1130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001989948275768916,
+      "loss": 0.1052,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001989918343494838,
+      "loss": 0.1015,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001989888366946246,
+      "loss": 0.1011,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019898583461244807,
+      "loss": 0.0979,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001989828281030885,
+      "loss": 0.0964,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019897981716668028,
+      "loss": 0.1013,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019897680180335815,
+      "loss": 0.1092,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019897378201325696,
+      "loss": 0.1005,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019897075779651174,
+      "loss": 0.1024,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001989677291532578,
+      "loss": 0.1031,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019896469608363053,
+      "loss": 0.1055,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019896165858776564,
+      "loss": 0.103,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019895861666579897,
+      "loss": 0.1017,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019895557031786656,
+      "loss": 0.0963,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019895251954410467,
+      "loss": 0.104,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019894946434464975,
+      "loss": 0.0988,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019894640471963842,
+      "loss": 0.1011,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019894334066920753,
+      "loss": 0.1085,
+      "step": 1148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001989402721934942,
+      "loss": 0.0981,
+      "step": 1149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019893719929263553,
+      "loss": 0.1005,
+      "step": 1150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019893412196676904,
+      "loss": 0.098,
+      "step": 1151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019893104021603236,
+      "loss": 0.101,
+      "step": 1152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019892795404056332,
+      "loss": 0.1063,
+      "step": 1153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019892486344049996,
+      "loss": 0.1017,
+      "step": 1154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019892176841598046,
+      "loss": 0.099,
+      "step": 1155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001989186689671433,
+      "loss": 0.1033,
+      "step": 1156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001989155650941271,
+      "loss": 0.0985,
+      "step": 1157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019891245679707064,
+      "loss": 0.1,
+      "step": 1158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000198909344076113,
+      "loss": 0.107,
+      "step": 1159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019890622693139332,
+      "loss": 0.1046,
+      "step": 1160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001989031053630511,
+      "loss": 0.0988,
+      "step": 1161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019889997937122588,
+      "loss": 0.1063,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019889684895605753,
+      "loss": 0.107,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000198893714117686,
+      "loss": 0.1035,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019889057485625156,
+      "loss": 0.0981,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001988874311718946,
+      "loss": 0.1031,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001988842830647557,
+      "loss": 0.1051,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019888113053497563,
+      "loss": 0.1054,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019887797358269547,
+      "loss": 0.1009,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019887481220805635,
+      "loss": 0.1018,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019887164641119967,
+      "loss": 0.1064,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019886847619226706,
+      "loss": 0.103,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001988653015514003,
+      "loss": 0.099,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019886212248874135,
+      "loss": 0.0966,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019885893900443242,
+      "loss": 0.1006,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019885575109861587,
+      "loss": 0.1044,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019885255877143433,
+      "loss": 0.1013,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019884936202303047,
+      "loss": 0.1041,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019884616085354736,
+      "loss": 0.1,
+      "step": 1179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019884295526312817,
+      "loss": 0.1013,
+      "step": 1180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019883974525191627,
+      "loss": 0.1059,
+      "step": 1181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019883653082005518,
+      "loss": 0.1007,
+      "step": 1182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001988333119676887,
+      "loss": 0.1028,
+      "step": 1183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001988300886949608,
+      "loss": 0.1032,
+      "step": 1184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019882686100201565,
+      "loss": 0.1068,
+      "step": 1185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019882362888899756,
+      "loss": 0.1042,
+      "step": 1186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019882039235605116,
+      "loss": 0.1059,
+      "step": 1187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019881715140332112,
+      "loss": 0.1092,
+      "step": 1188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019881390603095246,
+      "loss": 0.102,
+      "step": 1189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019881065623909033,
+      "loss": 0.1003,
+      "step": 1190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019880740202788004,
+      "loss": 0.0975,
+      "step": 1191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019880414339746718,
+      "loss": 0.1038,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019880088034799744,
+      "loss": 0.1026,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987976128796168,
+      "loss": 0.1078,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987943409924714,
+      "loss": 0.106,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019879106468670754,
+      "loss": 0.1087,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987877839624718,
+      "loss": 0.1021,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019878449881991089,
+      "loss": 0.0959,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019878120925917174,
+      "loss": 0.1019,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019877791528040146,
+      "loss": 0.0986,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987746168837474,
+      "loss": 0.0966,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987713140693571,
+      "loss": 0.1035,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019876800683737823,
+      "loss": 0.0985,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019876469518795875,
+      "loss": 0.1023,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019876137912124673,
+      "loss": 0.1067,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987580586373905,
+      "loss": 0.1088,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001987547337365386,
+      "loss": 0.1036,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019875140441883975,
+      "loss": 0.1031,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019874807068444276,
+      "loss": 0.0983,
+      "step": 1209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019874473253349683,
+      "loss": 0.0971,
+      "step": 1210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019874138996615125,
+      "loss": 0.0962,
+      "step": 1211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019873804298255546,
+      "loss": 0.0982,
+      "step": 1212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019873469158285918,
+      "loss": 0.099,
+      "step": 1213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001987313357672123,
+      "loss": 0.096,
+      "step": 1214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019872797553576493,
+      "loss": 0.0998,
+      "step": 1215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019872461088866737,
+      "loss": 0.0969,
+      "step": 1216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019872124182607008,
+      "loss": 0.1044,
+      "step": 1217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019871786834812377,
+      "loss": 0.0976,
+      "step": 1218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019871449045497924,
+      "loss": 0.0992,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019871110814678766,
+      "loss": 0.0963,
+      "step": 1220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019870772142370027,
+      "loss": 0.1075,
+      "step": 1221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019870433028586852,
+      "loss": 0.0955,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001987009347334441,
+      "loss": 0.1009,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019869753476657888,
+      "loss": 0.1015,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019869413038542497,
+      "loss": 0.1028,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019869072159013452,
+      "loss": 0.1018,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019868730838086012,
+      "loss": 0.0986,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019868389075775432,
+      "loss": 0.1036,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019868046872097007,
+      "loss": 0.099,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001986770422706603,
+      "loss": 0.1014,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001986736114069784,
+      "loss": 0.1016,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019867017613007767,
+      "loss": 0.1104,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001986667364401119,
+      "loss": 0.1022,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019866329233723481,
+      "loss": 0.0997,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019865984382160055,
+      "loss": 0.1015,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019865639089336324,
+      "loss": 0.0996,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001986529335526774,
+      "loss": 0.0985,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019864947179969763,
+      "loss": 0.0962,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001986460056345788,
+      "loss": 0.0979,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019864253505747582,
+      "loss": 0.1041,
+      "step": 1240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019863906006854402,
+      "loss": 0.098,
+      "step": 1241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019863558066793882,
+      "loss": 0.1015,
+      "step": 1242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019863209685581579,
+      "loss": 0.0986,
+      "step": 1243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019862860863233074,
+      "loss": 0.1011,
+      "step": 1244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019862511599763973,
+      "loss": 0.0994,
+      "step": 1245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019862161895189894,
+      "loss": 0.1002,
+      "step": 1246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019861811749526478,
+      "loss": 0.1074,
+      "step": 1247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019861461162789385,
+      "loss": 0.0936,
+      "step": 1248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000198611101349943,
+      "loss": 0.0996,
+      "step": 1249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019860758666156913,
+      "loss": 0.1023,
+      "step": 1250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001986040675629295,
+      "loss": 0.0995,
+      "step": 1251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001986005440541815,
+      "loss": 0.1058,
+      "step": 1252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019859701613548272,
+      "loss": 0.0967,
+      "step": 1253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019859348380699094,
+      "loss": 0.1022,
+      "step": 1254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019858994706886413,
+      "loss": 0.0996,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019858640592126053,
+      "loss": 0.1034,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019858286036433844,
+      "loss": 0.0989,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001985793103982565,
+      "loss": 0.1,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019857575602317344,
+      "loss": 0.0993,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019857219723924829,
+      "loss": 0.0991,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019856863404664016,
+      "loss": 0.1028,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001985650664455084,
+      "loss": 0.101,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019856149443601263,
+      "loss": 0.0989,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019855791801831258,
+      "loss": 0.0987,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019855433719256823,
+      "loss": 0.0991,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001985507519589397,
+      "loss": 0.1025,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019854716231758736,
+      "loss": 0.0994,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019854356826867178,
+      "loss": 0.1026,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019853996981235365,
+      "loss": 0.0969,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019853636694879394,
+      "loss": 0.1066,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001985327596781538,
+      "loss": 0.1033,
+      "step": 1271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019852914800059458,
+      "loss": 0.108,
+      "step": 1272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019852553191627778,
+      "loss": 0.1055,
+      "step": 1273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019852191142536514,
+      "loss": 0.1009,
+      "step": 1274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019851828652801858,
+      "loss": 0.0965,
+      "step": 1275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019851465722440026,
+      "loss": 0.1006,
+      "step": 1276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019851102351467245,
+      "loss": 0.1014,
+      "step": 1277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019850738539899774,
+      "loss": 0.0987,
+      "step": 1278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001985037428775388,
+      "loss": 0.0985,
+      "step": 1279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019850009595045852,
+      "loss": 0.097,
+      "step": 1280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019849644461792006,
+      "loss": 0.1007,
+      "step": 1281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001984927888800867,
+      "loss": 0.1029,
+      "step": 1282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019848912873712195,
+      "loss": 0.1035,
+      "step": 1283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001984854641891895,
+      "loss": 0.1061,
+      "step": 1284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019848179523645327,
+      "loss": 0.0983,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019847812187907734,
+      "loss": 0.0959,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000198474444117226,
+      "loss": 0.108,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019847076195106378,
+      "loss": 0.0966,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019846707538075532,
+      "loss": 0.105,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019846338440646552,
+      "loss": 0.1,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019845968902835944,
+      "loss": 0.1,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001984559892466024,
+      "loss": 0.0995,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019845228506135986,
+      "loss": 0.0942,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019844857647279744,
+      "loss": 0.1068,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019844486348108106,
+      "loss": 0.093,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001984411460863768,
+      "loss": 0.1057,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001984374242888509,
+      "loss": 0.0999,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019843369808866977,
+      "loss": 0.1018,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019842996748600013,
+      "loss": 0.0989,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001984262324810088,
+      "loss": 0.1042,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019842249307386287,
+      "loss": 0.0959,
+      "step": 1301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019841874926472953,
+      "loss": 0.1001,
+      "step": 1302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019841500105377628,
+      "loss": 0.0983,
+      "step": 1303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019841124844117075,
+      "loss": 0.1026,
+      "step": 1304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019840749142708072,
+      "loss": 0.099,
+      "step": 1305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019840373001167428,
+      "loss": 0.1,
+      "step": 1306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019839996419511967,
+      "loss": 0.1064,
+      "step": 1307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019839619397758526,
+      "loss": 0.1007,
+      "step": 1308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019839241935923972,
+      "loss": 0.0947,
+      "step": 1309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019838864034025185,
+      "loss": 0.098,
+      "step": 1310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019838485692079072,
+      "loss": 0.1007,
+      "step": 1311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019838106910102546,
+      "loss": 0.1015,
+      "step": 1312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019837727688112552,
+      "loss": 0.1003,
+      "step": 1313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019837348026126054,
+      "loss": 0.101,
+      "step": 1314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001983696792416003,
+      "loss": 0.1029,
+      "step": 1315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019836587382231476,
+      "loss": 0.0975,
+      "step": 1316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019836206400357422,
+      "loss": 0.1005,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019835824978554898,
+      "loss": 0.1021,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019835443116840966,
+      "loss": 0.1003,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019835060815232708,
+      "loss": 0.1069,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001983467807374722,
+      "loss": 0.098,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001983429489240162,
+      "loss": 0.098,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019833911271213047,
+      "loss": 0.1003,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019833527210198656,
+      "loss": 0.1049,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019833142709375627,
+      "loss": 0.0997,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001983275776876116,
+      "loss": 0.0958,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019832372388372464,
+      "loss": 0.1023,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001983198656822678,
+      "loss": 0.1007,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019831600308341365,
+      "loss": 0.0983,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019831213608733492,
+      "loss": 0.1034,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019830826469420458,
+      "loss": 0.1007,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019830438890419577,
+      "loss": 0.093,
+      "step": 1332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019830050871748184,
+      "loss": 0.0991,
+      "step": 1333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019829662413423633,
+      "loss": 0.1014,
+      "step": 1334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019829273515463295,
+      "loss": 0.104,
+      "step": 1335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019828884177884573,
+      "loss": 0.1031,
+      "step": 1336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019828494400704868,
+      "loss": 0.1024,
+      "step": 1337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019828104183941623,
+      "loss": 0.0981,
+      "step": 1338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019827713527612284,
+      "loss": 0.1021,
+      "step": 1339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019827322431734327,
+      "loss": 0.1057,
+      "step": 1340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019826930896325244,
+      "loss": 0.0983,
+      "step": 1341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019826538921402542,
+      "loss": 0.1028,
+      "step": 1342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001982614650698376,
+      "loss": 0.0995,
+      "step": 1343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001982575365308644,
+      "loss": 0.1039,
+      "step": 1344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001982536035972816,
+      "loss": 0.105,
+      "step": 1345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019824966626926508,
+      "loss": 0.0917,
+      "step": 1346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001982457245469909,
+      "loss": 0.1008,
+      "step": 1347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001982417784306354,
+      "loss": 0.1032,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019823782792037509,
+      "loss": 0.0999,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019823387301638658,
+      "loss": 0.0958,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019822991371884686,
+      "loss": 0.1036,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019822595002793288,
+      "loss": 0.099,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019822198194382201,
+      "loss": 0.1023,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019821800946669176,
+      "loss": 0.0984,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019821403259671967,
+      "loss": 0.0981,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019821005133408375,
+      "loss": 0.0971,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019820606567896195,
+      "loss": 0.0981,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019820207563153258,
+      "loss": 0.0989,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019819808119197409,
+      "loss": 0.1001,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019819408236046516,
+      "loss": 0.1028,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001981900791371846,
+      "loss": 0.1003,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019818607152231147,
+      "loss": 0.1002,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000198182059516025,
+      "loss": 0.103,
+      "step": 1363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019817804311850467,
+      "loss": 0.0968,
+      "step": 1364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019817402232993008,
+      "loss": 0.1054,
+      "step": 1365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019816999715048107,
+      "loss": 0.1056,
+      "step": 1366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019816596758033764,
+      "loss": 0.1007,
+      "step": 1367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019816193361968008,
+      "loss": 0.0955,
+      "step": 1368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019815789526868877,
+      "loss": 0.0954,
+      "step": 1369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019815385252754429,
+      "loss": 0.0948,
+      "step": 1370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001981498053964275,
+      "loss": 0.0999,
+      "step": 1371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001981457538755194,
+      "loss": 0.1016,
+      "step": 1372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001981416979650012,
+      "loss": 0.0993,
+      "step": 1373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019813763766505434,
+      "loss": 0.0984,
+      "step": 1374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019813357297586036,
+      "loss": 0.1,
+      "step": 1375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019812950389760104,
+      "loss": 0.0974,
+      "step": 1376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019812543043045842,
+      "loss": 0.1009,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019812135257461468,
+      "loss": 0.1018,
+      "step": 1378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001981172703302522,
+      "loss": 0.0986,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019811318369755354,
+      "loss": 0.1073,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001981090926767015,
+      "loss": 0.0982,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019810499726787903,
+      "loss": 0.0994,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019810089747126933,
+      "loss": 0.0959,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019809679328705574,
+      "loss": 0.1018,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019809268471542184,
+      "loss": 0.0984,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019808857175655136,
+      "loss": 0.0983,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001980844544106283,
+      "loss": 0.1051,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019808033267783675,
+      "loss": 0.1016,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001980762065583611,
+      "loss": 0.0991,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001980720760523859,
+      "loss": 0.0999,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019806794116009583,
+      "loss": 0.1045,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001980638018816759,
+      "loss": 0.0992,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001980596582173112,
+      "loss": 0.1024,
+      "step": 1393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019805551016718705,
+      "loss": 0.1057,
+      "step": 1394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019805135773148894,
+      "loss": 0.0957,
+      "step": 1395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001980472009104027,
+      "loss": 0.0969,
+      "step": 1396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019804303970411414,
+      "loss": 0.1006,
+      "step": 1397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019803887411280947,
+      "loss": 0.097,
+      "step": 1398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001980347041366749,
+      "loss": 0.1042,
+      "step": 1399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019803052977589695,
+      "loss": 0.0993,
+      "step": 1400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019802635103066237,
+      "loss": 0.1015,
+      "step": 1401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019802216790115802,
+      "loss": 0.1054,
+      "step": 1402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000198017980387571,
+      "loss": 0.0942,
+      "step": 1403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001980137884900886,
+      "loss": 0.1007,
+      "step": 1404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001980095922088983,
+      "loss": 0.1058,
+      "step": 1405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019800539154418782,
+      "loss": 0.103,
+      "step": 1406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019800118649614497,
+      "loss": 0.1022,
+      "step": 1407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019799697706495786,
+      "loss": 0.1029,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019799276325081476,
+      "loss": 0.1088,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019798854505390413,
+      "loss": 0.1049,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019798432247441458,
+      "loss": 0.1011,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019798009551253508,
+      "loss": 0.1005,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019797586416845457,
+      "loss": 0.0955,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019797162844236238,
+      "loss": 0.104,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001979673883344479,
+      "loss": 0.1036,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019796314384490078,
+      "loss": 0.104,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001979588949739109,
+      "loss": 0.0981,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019795464172166823,
+      "loss": 0.0961,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019795038408836304,
+      "loss": 0.1016,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019794612207418575,
+      "loss": 0.1034,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019794185567932696,
+      "loss": 0.1007,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019793758490397752,
+      "loss": 0.1017,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019793330974832838,
+      "loss": 0.102,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019792903021257084,
+      "loss": 0.1053,
+      "step": 1424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001979247462968962,
+      "loss": 0.1047,
+      "step": 1425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019792045800149616,
+      "loss": 0.1031,
+      "step": 1426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019791616532656247,
+      "loss": 0.0985,
+      "step": 1427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001979118682722871,
+      "loss": 0.0982,
+      "step": 1428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019790756683886227,
+      "loss": 0.106,
+      "step": 1429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019790326102648033,
+      "loss": 0.1098,
+      "step": 1430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978989508353339,
+      "loss": 0.1042,
+      "step": 1431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019789463626561574,
+      "loss": 0.097,
+      "step": 1432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978903173175188,
+      "loss": 0.1027,
+      "step": 1433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978859939912363,
+      "loss": 0.1036,
+      "step": 1434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019788166628696154,
+      "loss": 0.1026,
+      "step": 1435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978773342048881,
+      "loss": 0.1,
+      "step": 1436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019787299774520975,
+      "loss": 0.102,
+      "step": 1437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019786865690812043,
+      "loss": 0.1059,
+      "step": 1438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019786431169381427,
+      "loss": 0.105,
+      "step": 1439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019785996210248564,
+      "loss": 0.0948,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019785560813432906,
+      "loss": 0.0946,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019785124978953929,
+      "loss": 0.1007,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019784688706831122,
+      "loss": 0.0985,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019784251997083998,
+      "loss": 0.0984,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019783814849732092,
+      "loss": 0.101,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019783377264794951,
+      "loss": 0.095,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019782939242292153,
+      "loss": 0.1038,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978250078224328,
+      "loss": 0.1036,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978206188466795,
+      "loss": 0.1069,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001978162254958579,
+      "loss": 0.0996,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019781182777016447,
+      "loss": 0.105,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019780742566979592,
+      "loss": 0.1054,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019780301919494915,
+      "loss": 0.1004,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019779860834582123,
+      "loss": 0.104,
+      "step": 1454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019779419312260945,
+      "loss": 0.0987,
+      "step": 1455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019778977352551125,
+      "loss": 0.098,
+      "step": 1456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019778534955472435,
+      "loss": 0.1025,
+      "step": 1457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019778092121044654,
+      "loss": 0.1037,
+      "step": 1458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019777648849287594,
+      "loss": 0.0927,
+      "step": 1459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019777205140221078,
+      "loss": 0.1064,
+      "step": 1460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019776760993864953,
+      "loss": 0.0998,
+      "step": 1461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001977631641023908,
+      "loss": 0.0929,
+      "step": 1462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019775871389363347,
+      "loss": 0.1094,
+      "step": 1463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019775425931257656,
+      "loss": 0.0954,
+      "step": 1464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019774980035941933,
+      "loss": 0.095,
+      "step": 1465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019774533703436115,
+      "loss": 0.0924,
+      "step": 1466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019774086933760168,
+      "loss": 0.1007,
+      "step": 1467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019773639726934077,
+      "loss": 0.0996,
+      "step": 1468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019773192082977835,
+      "loss": 0.1016,
+      "step": 1469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019772744001911473,
+      "loss": 0.1068,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019772295483755028,
+      "loss": 0.0941,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019771846528528558,
+      "loss": 0.1033,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001977139713625214,
+      "loss": 0.1026,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019770947306945878,
+      "loss": 0.0998,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019770497040629893,
+      "loss": 0.0974,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019770046337324319,
+      "loss": 0.1024,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019769595197049314,
+      "loss": 0.0928,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019769143619825062,
+      "loss": 0.0984,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001976869160567175,
+      "loss": 0.1055,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019768239154609597,
+      "loss": 0.0968,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019767786266658844,
+      "loss": 0.1037,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019767332941839744,
+      "loss": 0.1072,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019766879180172574,
+      "loss": 0.1093,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019766424981677622,
+      "loss": 0.1058,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019765970346375208,
+      "loss": 0.1018,
+      "step": 1485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001976551527428567,
+      "loss": 0.1026,
+      "step": 1486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001976505976542935,
+      "loss": 0.0954,
+      "step": 1487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019764603819826627,
+      "loss": 0.098,
+      "step": 1488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019764147437497894,
+      "loss": 0.1026,
+      "step": 1489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001976369061846356,
+      "loss": 0.0985,
+      "step": 1490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019763233362744064,
+      "loss": 0.0978,
+      "step": 1491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001976277567035985,
+      "loss": 0.0996,
+      "step": 1492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019762317541331385,
+      "loss": 0.0998,
+      "step": 1493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019761858975679165,
+      "loss": 0.1054,
+      "step": 1494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000197613999734237,
+      "loss": 0.0951,
+      "step": 1495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019760940534585523,
+      "loss": 0.0986,
+      "step": 1496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001976048065918517,
+      "loss": 0.0986,
+      "step": 1497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019760020347243216,
+      "loss": 0.1008,
+      "step": 1498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001975955959878025,
+      "loss": 0.1054,
+      "step": 1499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019759098413816878,
+      "loss": 0.1004,
+      "step": 1500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019758636792373727,
+      "loss": 0.1,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019758174734471444,
+      "loss": 0.0999,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001975771224013069,
+      "loss": 0.1033,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001975724930937216,
+      "loss": 0.1061,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001975678594221655,
+      "loss": 0.1041,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019756322138684586,
+      "loss": 0.0983,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001975585789879701,
+      "loss": 0.1044,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019755393222574595,
+      "loss": 0.0935,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019754928110038108,
+      "loss": 0.0988,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001975446256120837,
+      "loss": 0.0982,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019753996576106185,
+      "loss": 0.0992,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019753530154752408,
+      "loss": 0.1024,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001975306329716789,
+      "loss": 0.1029,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019752596003373517,
+      "loss": 0.0969,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019752128273390192,
+      "loss": 0.1063,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019751660107238824,
+      "loss": 0.0988,
+      "step": 1516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019751191504940364,
+      "loss": 0.0961,
+      "step": 1517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019750722466515758,
+      "loss": 0.1011,
+      "step": 1518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019750252991985993,
+      "loss": 0.1035,
+      "step": 1519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019749783081372067,
+      "loss": 0.1001,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001974931273469499,
+      "loss": 0.1034,
+      "step": 1521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019748841951975805,
+      "loss": 0.1018,
+      "step": 1522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019748370733235564,
+      "loss": 0.0991,
+      "step": 1523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001974789907849534,
+      "loss": 0.1026,
+      "step": 1524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001974742698777624,
+      "loss": 0.1033,
+      "step": 1525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019746954461099362,
+      "loss": 0.1005,
+      "step": 1526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019746481498485855,
+      "loss": 0.102,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019746008099956862,
+      "loss": 0.1006,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019745534265533556,
+      "loss": 0.1017,
+      "step": 1529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001974505999523714,
+      "loss": 0.1043,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019744585289088815,
+      "loss": 0.0975,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019744110147109812,
+      "loss": 0.0974,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001974363456932139,
+      "loss": 0.0985,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019743158555744818,
+      "loss": 0.1019,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019742682106401382,
+      "loss": 0.0969,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019742205221312396,
+      "loss": 0.1043,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001974172790049918,
+      "loss": 0.0938,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019741250143983093,
+      "loss": 0.1006,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019740771951785494,
+      "loss": 0.1046,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019740293323927777,
+      "loss": 0.0983,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019739814260431345,
+      "loss": 0.1007,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019739334761317628,
+      "loss": 0.0964,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019738854826608068,
+      "loss": 0.1021,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019738374456324128,
+      "loss": 0.1009,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000197378936504873,
+      "loss": 0.0969,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019737412409119083,
+      "loss": 0.1008,
+      "step": 1546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019736930732241003,
+      "loss": 0.0989,
+      "step": 1547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019736448619874603,
+      "loss": 0.0998,
+      "step": 1548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019735966072041445,
+      "loss": 0.0971,
+      "step": 1549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001973548308876311,
+      "loss": 0.1004,
+      "step": 1550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019734999670061206,
+      "loss": 0.0978,
+      "step": 1551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001973451581595734,
+      "loss": 0.0987,
+      "step": 1552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001973403152647317,
+      "loss": 0.0986,
+      "step": 1553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019733546801630345,
+      "loss": 0.107,
+      "step": 1554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019733061641450547,
+      "loss": 0.1057,
+      "step": 1555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019732576045955477,
+      "loss": 0.1049,
+      "step": 1556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001973209001516685,
+      "loss": 0.1022,
+      "step": 1557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019731603549106406,
+      "loss": 0.102,
+      "step": 1558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000197311166477959,
+      "loss": 0.1001,
+      "step": 1559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019730629311257115,
+      "loss": 0.1022,
+      "step": 1560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019730141539511838,
+      "loss": 0.1063,
+      "step": 1561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019729653332581894,
+      "loss": 0.0943,
+      "step": 1562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019729164690489114,
+      "loss": 0.0974,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019728675613255354,
+      "loss": 0.1005,
+      "step": 1564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019728186100902484,
+      "loss": 0.0987,
+      "step": 1565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000197276961534524,
+      "loss": 0.0976,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001972720577092702,
+      "loss": 0.0917,
+      "step": 1567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001972671495334827,
+      "loss": 0.1049,
+      "step": 1568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019726223700738106,
+      "loss": 0.1002,
+      "step": 1569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019725732013118501,
+      "loss": 0.0971,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019725239890511437,
+      "loss": 0.1007,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019724747332938933,
+      "loss": 0.1003,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019724254340423018,
+      "loss": 0.1046,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019723760912985738,
+      "loss": 0.1033,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019723267050649162,
+      "loss": 0.0938,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001972277275343538,
+      "loss": 0.0953,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019722278021366499,
+      "loss": 0.1018,
+      "step": 1577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019721782854464649,
+      "loss": 0.0997,
+      "step": 1578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019721287252751969,
+      "loss": 0.0949,
+      "step": 1579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019720791216250633,
+      "loss": 0.097,
+      "step": 1580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001972029474498282,
+      "loss": 0.0994,
+      "step": 1581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019719797838970742,
+      "loss": 0.1004,
+      "step": 1582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019719300498236616,
+      "loss": 0.0993,
+      "step": 1583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971880272280269,
+      "loss": 0.0989,
+      "step": 1584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971830451269123,
+      "loss": 0.1,
+      "step": 1585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971780586792451,
+      "loss": 0.106,
+      "step": 1586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971730678852484,
+      "loss": 0.1007,
+      "step": 1587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971680727451454,
+      "loss": 0.1028,
+      "step": 1588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001971630732591595,
+      "loss": 0.1047,
+      "step": 1589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019715806942751426,
+      "loss": 0.1017,
+      "step": 1590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019715306125043353,
+      "loss": 0.1012,
+      "step": 1591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019714804872814133,
+      "loss": 0.0969,
+      "step": 1592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019714303186086176,
+      "loss": 0.0973,
+      "step": 1593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019713801064881925,
+      "loss": 0.0934,
+      "step": 1594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019713298509223842,
+      "loss": 0.0964,
+      "step": 1595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019712795519134395,
+      "loss": 0.1021,
+      "step": 1596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019712292094636088,
+      "loss": 0.0976,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019711788235751433,
+      "loss": 0.1057,
+      "step": 1598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019711283942502967,
+      "loss": 0.0959,
+      "step": 1599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019710779214913244,
+      "loss": 0.1004,
+      "step": 1600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019710274053004839,
+      "loss": 0.106,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019709768456800344,
+      "loss": 0.0973,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001970926242632237,
+      "loss": 0.0995,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019708755961593557,
+      "loss": 0.1044,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019708249062636547,
+      "loss": 0.0948,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001970774172947402,
+      "loss": 0.0997,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019707233962128662,
+      "loss": 0.093,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019706725760623183,
+      "loss": 0.0938,
+      "step": 1608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019706217124980315,
+      "loss": 0.0973,
+      "step": 1609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019705708055222805,
+      "loss": 0.1007,
+      "step": 1610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019705198551373424,
+      "loss": 0.1008,
+      "step": 1611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019704688613454954,
+      "loss": 0.0972,
+      "step": 1612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019704178241490206,
+      "loss": 0.1009,
+      "step": 1613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019703667435502008,
+      "loss": 0.1028,
+      "step": 1614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019703156195513205,
+      "loss": 0.1012,
+      "step": 1615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001970264452154666,
+      "loss": 0.1018,
+      "step": 1616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019702132413625264,
+      "loss": 0.097,
+      "step": 1617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019701619871771912,
+      "loss": 0.1023,
+      "step": 1618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019701106896009534,
+      "loss": 0.1015,
+      "step": 1619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019700593486361072,
+      "loss": 0.0987,
+      "step": 1620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019700079642849487,
+      "loss": 0.0948,
+      "step": 1621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019699565365497766,
+      "loss": 0.0925,
+      "step": 1622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000196990506543289,
+      "loss": 0.0904,
+      "step": 1623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001969853550936592,
+      "loss": 0.0991,
+      "step": 1624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019698019930631863,
+      "loss": 0.0985,
+      "step": 1625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019697503918149787,
+      "loss": 0.0995,
+      "step": 1626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019696987471942769,
+      "loss": 0.0948,
+      "step": 1627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019696470592033913,
+      "loss": 0.0981,
+      "step": 1628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001969595327844633,
+      "loss": 0.1,
+      "step": 1629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001969543553120316,
+      "loss": 0.0945,
+      "step": 1630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019694917350327566,
+      "loss": 0.0951,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019694398735842713,
+      "loss": 0.0988,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019693879687771802,
+      "loss": 0.1023,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019693360206138045,
+      "loss": 0.0968,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019692840290964678,
+      "loss": 0.0984,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019692319942274955,
+      "loss": 0.1017,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019691799160092148,
+      "loss": 0.1002,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001969127794443955,
+      "loss": 0.1025,
+      "step": 1638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001969075629534047,
+      "loss": 0.0974,
+      "step": 1639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001969023421281824,
+      "loss": 0.1031,
+      "step": 1640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019689711696896213,
+      "loss": 0.0978,
+      "step": 1641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019689188747597756,
+      "loss": 0.0964,
+      "step": 1642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019688665364946258,
+      "loss": 0.0965,
+      "step": 1643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001968814154896513,
+      "loss": 0.0971,
+      "step": 1644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019687617299677797,
+      "loss": 0.0991,
+      "step": 1645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019687092617107706,
+      "loss": 0.0989,
+      "step": 1646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019686567501278328,
+      "loss": 0.0923,
+      "step": 1647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019686041952213145,
+      "loss": 0.1021,
+      "step": 1648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019685515969935664,
+      "loss": 0.105,
+      "step": 1649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019684989554469412,
+      "loss": 0.1074,
+      "step": 1650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019684462705837926,
+      "loss": 0.1021,
+      "step": 1651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019683935424064778,
+      "loss": 0.1064,
+      "step": 1652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019683407709173545,
+      "loss": 0.0997,
+      "step": 1653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001968287956118783,
+      "loss": 0.0985,
+      "step": 1654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019682350980131257,
+      "loss": 0.0977,
+      "step": 1655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019681821966027467,
+      "loss": 0.1,
+      "step": 1656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001968129251890012,
+      "loss": 0.098,
+      "step": 1657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001968076263877289,
+      "loss": 0.101,
+      "step": 1658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019680232325669487,
+      "loss": 0.093,
+      "step": 1659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019679701579613617,
+      "loss": 0.1046,
+      "step": 1660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001967917040062903,
+      "loss": 0.1013,
+      "step": 1661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001967863878873948,
+      "loss": 0.0969,
+      "step": 1662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019678106743968736,
+      "loss": 0.1018,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000196775742663406,
+      "loss": 0.1016,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019677041355878886,
+      "loss": 0.098,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019676508012607432,
+      "loss": 0.1078,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019675974236550085,
+      "loss": 0.0988,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019675440027730725,
+      "loss": 0.1008,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019674905386173243,
+      "loss": 0.1012,
+      "step": 1669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001967437031190155,
+      "loss": 0.0983,
+      "step": 1670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019673834804939574,
+      "loss": 0.0988,
+      "step": 1671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019673298865311275,
+      "loss": 0.0946,
+      "step": 1672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019672762493040612,
+      "loss": 0.0991,
+      "step": 1673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019672225688151582,
+      "loss": 0.0983,
+      "step": 1674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019671688450668192,
+      "loss": 0.0986,
+      "step": 1675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019671150780614473,
+      "loss": 0.0956,
+      "step": 1676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019670612678014467,
+      "loss": 0.095,
+      "step": 1677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019670074142892242,
+      "loss": 0.1015,
+      "step": 1678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001966953517527189,
+      "loss": 0.0958,
+      "step": 1679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019668995775177508,
+      "loss": 0.0981,
+      "step": 1680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019668455942633225,
+      "loss": 0.1019,
+      "step": 1681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019667915677663186,
+      "loss": 0.0942,
+      "step": 1682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019667374980291555,
+      "loss": 0.0977,
+      "step": 1683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019666833850542516,
+      "loss": 0.0977,
+      "step": 1684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019666292288440267,
+      "loss": 0.1022,
+      "step": 1685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019665750294009027,
+      "loss": 0.1082,
+      "step": 1686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019665207867273048,
+      "loss": 0.0993,
+      "step": 1687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001966466500825658,
+      "loss": 0.1014,
+      "step": 1688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019664121716983908,
+      "loss": 0.0971,
+      "step": 1689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001966357799347933,
+      "loss": 0.103,
+      "step": 1690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001966303383776716,
+      "loss": 0.0932,
+      "step": 1691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019662489249871746,
+      "loss": 0.1,
+      "step": 1692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019661944229817433,
+      "loss": 0.0989,
+      "step": 1693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019661398777628603,
+      "loss": 0.1024,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019660852893329652,
+      "loss": 0.0996,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019660306576944995,
+      "loss": 0.0992,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019659759828499063,
+      "loss": 0.1018,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001965921264801631,
+      "loss": 0.0968,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019658665035521215,
+      "loss": 0.0964,
+      "step": 1699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019658116991038265,
+      "loss": 0.1071,
+      "step": 1700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019657568514591967,
+      "loss": 0.1006,
+      "step": 1701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019657019606206862,
+      "loss": 0.1018,
+      "step": 1702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019656470265907495,
+      "loss": 0.099,
+      "step": 1703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019655920493718437,
+      "loss": 0.0963,
+      "step": 1704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001965537028966427,
+      "loss": 0.0936,
+      "step": 1705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019654819653769611,
+      "loss": 0.0982,
+      "step": 1706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019654268586059086,
+      "loss": 0.1042,
+      "step": 1707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019653717086557336,
+      "loss": 0.1024,
+      "step": 1708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019653165155289034,
+      "loss": 0.1022,
+      "step": 1709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001965261279227886,
+      "loss": 0.1,
+      "step": 1710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019652059997551523,
+      "loss": 0.0943,
+      "step": 1711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019651506771131743,
+      "loss": 0.0968,
+      "step": 1712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019650953113044267,
+      "loss": 0.1008,
+      "step": 1713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019650399023313851,
+      "loss": 0.0987,
+      "step": 1714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019649844501965286,
+      "loss": 0.0983,
+      "step": 1715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019649289549023368,
+      "loss": 0.105,
+      "step": 1716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001964873416451292,
+      "loss": 0.0984,
+      "step": 1717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019648178348458776,
+      "loss": 0.1009,
+      "step": 1718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019647622100885803,
+      "loss": 0.1047,
+      "step": 1719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019647065421818873,
+      "loss": 0.1027,
+      "step": 1720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019646508311282887,
+      "loss": 0.0985,
+      "step": 1721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019645950769302763,
+      "loss": 0.1002,
+      "step": 1722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019645392795903435,
+      "loss": 0.0958,
+      "step": 1723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019644834391109857,
+      "loss": 0.0983,
+      "step": 1724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019644275554947006,
+      "loss": 0.0951,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001964371628743988,
+      "loss": 0.0993,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019643156588613488,
+      "loss": 0.0999,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001964259645849286,
+      "loss": 0.0994,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019642035897103055,
+      "loss": 0.1031,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001964147490446914,
+      "loss": 0.1024,
+      "step": 1730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019640913480616207,
+      "loss": 0.0986,
+      "step": 1731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019640351625569365,
+      "loss": 0.0993,
+      "step": 1732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019639789339353744,
+      "loss": 0.1018,
+      "step": 1733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019639226621994492,
+      "loss": 0.1009,
+      "step": 1734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019638663473516777,
+      "loss": 0.1018,
+      "step": 1735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019638099893945786,
+      "loss": 0.0956,
+      "step": 1736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019637535883306723,
+      "loss": 0.097,
+      "step": 1737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001963697144162482,
+      "loss": 0.1007,
+      "step": 1738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019636406568925314,
+      "loss": 0.0988,
+      "step": 1739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019635841265233477,
+      "loss": 0.0994,
+      "step": 1740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019635275530574585,
+      "loss": 0.0912,
+      "step": 1741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019634709364973948,
+      "loss": 0.099,
+      "step": 1742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001963414276845688,
+      "loss": 0.0948,
+      "step": 1743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019633575741048728,
+      "loss": 0.0924,
+      "step": 1744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019633008282774852,
+      "loss": 0.0996,
+      "step": 1745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001963244039366063,
+      "loss": 0.1013,
+      "step": 1746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019631872073731462,
+      "loss": 0.0982,
+      "step": 1747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019631303323012765,
+      "loss": 0.0953,
+      "step": 1748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001963073414152998,
+      "loss": 0.102,
+      "step": 1749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001963016452930856,
+      "loss": 0.1029,
+      "step": 1750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019629594486373982,
+      "loss": 0.0998,
+      "step": 1751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019629024012751746,
+      "loss": 0.1023,
+      "step": 1752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001962845310846736,
+      "loss": 0.0969,
+      "step": 1753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001962788177354636,
+      "loss": 0.0999,
+      "step": 1754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019627310008014305,
+      "loss": 0.0959,
+      "step": 1755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019626737811896758,
+      "loss": 0.0986,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019626165185219316,
+      "loss": 0.0976,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001962559212800759,
+      "loss": 0.0921,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019625018640287206,
+      "loss": 0.0976,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019624444722083823,
+      "loss": 0.0935,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000196238703734231,
+      "loss": 0.0965,
+      "step": 1761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019623295594330727,
+      "loss": 0.0963,
+      "step": 1762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019622720384832417,
+      "loss": 0.1025,
+      "step": 1763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019622144744953893,
+      "loss": 0.0984,
+      "step": 1764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019621568674720895,
+      "loss": 0.1002,
+      "step": 1765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019620992174159194,
+      "loss": 0.1008,
+      "step": 1766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019620415243294578,
+      "loss": 0.1032,
+      "step": 1767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961983788215284,
+      "loss": 0.1008,
+      "step": 1768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019619260090759813,
+      "loss": 0.0982,
+      "step": 1769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019618681869141332,
+      "loss": 0.1034,
+      "step": 1770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961810321732326,
+      "loss": 0.0941,
+      "step": 1771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961752413533148,
+      "loss": 0.1034,
+      "step": 1772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019616944623191887,
+      "loss": 0.0991,
+      "step": 1773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019616364680930407,
+      "loss": 0.0985,
+      "step": 1774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019615784308572973,
+      "loss": 0.0999,
+      "step": 1775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961520350614554,
+      "loss": 0.1043,
+      "step": 1776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961462227367409,
+      "loss": 0.1022,
+      "step": 1777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961404061118462,
+      "loss": 0.1017,
+      "step": 1778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019613458518703136,
+      "loss": 0.0963,
+      "step": 1779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961287599625568,
+      "loss": 0.0972,
+      "step": 1780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019612293043868307,
+      "loss": 0.0959,
+      "step": 1781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961170966156708,
+      "loss": 0.1064,
+      "step": 1782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019611125849378105,
+      "loss": 0.0956,
+      "step": 1783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001961054160732748,
+      "loss": 0.103,
+      "step": 1784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019609956935441348,
+      "loss": 0.1014,
+      "step": 1785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019609371833745844,
+      "loss": 0.0951,
+      "step": 1786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001960878630226715,
+      "loss": 0.0963,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001960820034103145,
+      "loss": 0.1015,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019607613950064945,
+      "loss": 0.0957,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019607027129393871,
+      "loss": 0.0969,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001960643987904447,
+      "loss": 0.0986,
+      "step": 1791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019605852199043006,
+      "loss": 0.0915,
+      "step": 1792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019605264089415762,
+      "loss": 0.1014,
+      "step": 1793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019604675550189047,
+      "loss": 0.0948,
+      "step": 1794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019604086581389177,
+      "loss": 0.1048,
+      "step": 1795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019603497183042499,
+      "loss": 0.1017,
+      "step": 1796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019602907355175373,
+      "loss": 0.1005,
+      "step": 1797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019602317097814175,
+      "loss": 0.0963,
+      "step": 1798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001960172641098531,
+      "loss": 0.1002,
+      "step": 1799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019601135294715195,
+      "loss": 0.0961,
+      "step": 1800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019600543749030268,
+      "loss": 0.1029,
+      "step": 1801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019599951773956984,
+      "loss": 0.0924,
+      "step": 1802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019599359369521822,
+      "loss": 0.0986,
+      "step": 1803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019598766535751277,
+      "loss": 0.0949,
+      "step": 1804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019598173272671865,
+      "loss": 0.0964,
+      "step": 1805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019597579580310116,
+      "loss": 0.1009,
+      "step": 1806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019596985458692582,
+      "loss": 0.1033,
+      "step": 1807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019596390907845846,
+      "loss": 0.1027,
+      "step": 1808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019595795927796486,
+      "loss": 0.0954,
+      "step": 1809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019595200518571123,
+      "loss": 0.0939,
+      "step": 1810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019594604680196382,
+      "loss": 0.1012,
+      "step": 1811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019594008412698912,
+      "loss": 0.095,
+      "step": 1812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019593411716105382,
+      "loss": 0.0983,
+      "step": 1813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019592814590442483,
+      "loss": 0.0999,
+      "step": 1814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019592217035736915,
+      "loss": 0.0948,
+      "step": 1815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001959161905201541,
+      "loss": 0.1023,
+      "step": 1816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001959102063930471,
+      "loss": 0.0984,
+      "step": 1817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019590421797631576,
+      "loss": 0.0976,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000195898225270228,
+      "loss": 0.0991,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019589222827505178,
+      "loss": 0.099,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019588622699105533,
+      "loss": 0.0934,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001958802214185071,
+      "loss": 0.1029,
+      "step": 1822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019587421155767564,
+      "loss": 0.1023,
+      "step": 1823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019586819740882976,
+      "loss": 0.1021,
+      "step": 1824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019586217897223844,
+      "loss": 0.0972,
+      "step": 1825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019585615624817087,
+      "loss": 0.096,
+      "step": 1826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001958501292368964,
+      "loss": 0.0978,
+      "step": 1827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019584409793868465,
+      "loss": 0.1021,
+      "step": 1828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019583806235380527,
+      "loss": 0.1009,
+      "step": 1829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019583202248252827,
+      "loss": 0.0961,
+      "step": 1830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001958259783251238,
+      "loss": 0.1017,
+      "step": 1831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019581992988186215,
+      "loss": 0.095,
+      "step": 1832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019581387715301386,
+      "loss": 0.103,
+      "step": 1833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019580782013884962,
+      "loss": 0.0938,
+      "step": 1834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019580175883964035,
+      "loss": 0.0937,
+      "step": 1835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019579569325565716,
+      "loss": 0.0938,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019578962338717132,
+      "loss": 0.1003,
+      "step": 1837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019578354923445427,
+      "loss": 0.0963,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019577747079777772,
+      "loss": 0.0995,
+      "step": 1839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019577138807741354,
+      "loss": 0.1047,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019576530107363375,
+      "loss": 0.0979,
+      "step": 1841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019575920978671062,
+      "loss": 0.1081,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019575311421691659,
+      "loss": 0.0971,
+      "step": 1843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019574701436452424,
+      "loss": 0.1032,
+      "step": 1844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019574091022980645,
+      "loss": 0.0983,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001957348018130362,
+      "loss": 0.103,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019572868911448663,
+      "loss": 0.0984,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019572257213443126,
+      "loss": 0.0953,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019571645087314358,
+      "loss": 0.0979,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001957103253308974,
+      "loss": 0.099,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001957041955079667,
+      "loss": 0.0959,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001956980614046256,
+      "loss": 0.0949,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019569192302114848,
+      "loss": 0.0962,
+      "step": 1853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019568578035780986,
+      "loss": 0.1036,
+      "step": 1854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019567963341488452,
+      "loss": 0.1015,
+      "step": 1855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019567348219264732,
+      "loss": 0.0953,
+      "step": 1856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019566732669137343,
+      "loss": 0.0964,
+      "step": 1857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001956611669113381,
+      "loss": 0.1038,
+      "step": 1858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001956550028528169,
+      "loss": 0.0993,
+      "step": 1859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019564883451608547,
+      "loss": 0.0994,
+      "step": 1860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001956426619014197,
+      "loss": 0.1007,
+      "step": 1861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001956364850090957,
+      "loss": 0.0986,
+      "step": 1862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019563030383938965,
+      "loss": 0.098,
+      "step": 1863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019562411839257809,
+      "loss": 0.0904,
+      "step": 1864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019561792866893763,
+      "loss": 0.0944,
+      "step": 1865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001956117346687451,
+      "loss": 0.1028,
+      "step": 1866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019560553639227755,
+      "loss": 0.1008,
+      "step": 1867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019559933383981218,
+      "loss": 0.0941,
+      "step": 1868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019559312701162643,
+      "loss": 0.0977,
+      "step": 1869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001955869159079979,
+      "loss": 0.0945,
+      "step": 1870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019558070052920435,
+      "loss": 0.0986,
+      "step": 1871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019557448087552376,
+      "loss": 0.0997,
+      "step": 1872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019556825694723437,
+      "loss": 0.0952,
+      "step": 1873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001955620287446145,
+      "loss": 0.1028,
+      "step": 1874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001955557962679427,
+      "loss": 0.0958,
+      "step": 1875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019554955951749778,
+      "loss": 0.1011,
+      "step": 1876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019554331849355863,
+      "loss": 0.1004,
+      "step": 1877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001955370731964044,
+      "loss": 0.0949,
+      "step": 1878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001955308236263144,
+      "loss": 0.0984,
+      "step": 1879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019552456978356813,
+      "loss": 0.0947,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019551831166844533,
+      "loss": 0.1003,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001955120492812259,
+      "loss": 0.094,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019550578262218988,
+      "loss": 0.0895,
+      "step": 1883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019549951169161762,
+      "loss": 0.0977,
+      "step": 1884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001954932364897895,
+      "loss": 0.0959,
+      "step": 1885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001954869570169863,
+      "loss": 0.0953,
+      "step": 1886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019548067327348876,
+      "loss": 0.0926,
+      "step": 1887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019547438525957796,
+      "loss": 0.1004,
+      "step": 1888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019546809297553517,
+      "loss": 0.1042,
+      "step": 1889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019546179642164176,
+      "loss": 0.0988,
+      "step": 1890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019545549559817936,
+      "loss": 0.1029,
+      "step": 1891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019544919050542984,
+      "loss": 0.1013,
+      "step": 1892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019544288114367513,
+      "loss": 0.1014,
+      "step": 1893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001954365675131974,
+      "loss": 0.0994,
+      "step": 1894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001954302496142791,
+      "loss": 0.1006,
+      "step": 1895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019542392744720277,
+      "loss": 0.0974,
+      "step": 1896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019541760101225117,
+      "loss": 0.0959,
+      "step": 1897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019541127030970722,
+      "loss": 0.1046,
+      "step": 1898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019540493533985411,
+      "loss": 0.0999,
+      "step": 1899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019539859610297517,
+      "loss": 0.0994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019539225259935388,
+      "loss": 0.0952,
+      "step": 1901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019538590482927405,
+      "loss": 0.1012,
+      "step": 1902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019537955279301948,
+      "loss": 0.0987,
+      "step": 1903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001953731964908743,
+      "loss": 0.1034,
+      "step": 1904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019536683592312286,
+      "loss": 0.0978,
+      "step": 1905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001953604710900496,
+      "loss": 0.0954,
+      "step": 1906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001953541019919391,
+      "loss": 0.0991,
+      "step": 1907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019534772862907636,
+      "loss": 0.0996,
+      "step": 1908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019534135100174638,
+      "loss": 0.0973,
+      "step": 1909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019533496911023436,
+      "loss": 0.1012,
+      "step": 1910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019532858295482578,
+      "loss": 0.0979,
+      "step": 1911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019532219253580623,
+      "loss": 0.0967,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019531579785346157,
+      "loss": 0.1001,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019530939890807778,
+      "loss": 0.1056,
+      "step": 1914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019530299569994107,
+      "loss": 0.1059,
+      "step": 1915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952965882293378,
+      "loss": 0.0981,
+      "step": 1916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019529017649655455,
+      "loss": 0.0993,
+      "step": 1917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019528376050187806,
+      "loss": 0.1005,
+      "step": 1918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952773402455954,
+      "loss": 0.1006,
+      "step": 1919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952709157279936,
+      "loss": 0.1004,
+      "step": 1920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019526448694936005,
+      "loss": 0.0963,
+      "step": 1921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019525805390998224,
+      "loss": 0.1009,
+      "step": 1922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019525161661014794,
+      "loss": 0.0978,
+      "step": 1923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952451750501451,
+      "loss": 0.1005,
+      "step": 1924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019523872923026168,
+      "loss": 0.1017,
+      "step": 1925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952322791507861,
+      "loss": 0.0956,
+      "step": 1926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019522582481200677,
+      "loss": 0.0939,
+      "step": 1927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952193662142124,
+      "loss": 0.0974,
+      "step": 1928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952129033576919,
+      "loss": 0.103,
+      "step": 1929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001952064362427342,
+      "loss": 0.0897,
+      "step": 1930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019519996486962865,
+      "loss": 0.1,
+      "step": 1931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019519348923866466,
+      "loss": 0.1035,
+      "step": 1932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019518700935013182,
+      "loss": 0.0965,
+      "step": 1933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019518052520431996,
+      "loss": 0.1053,
+      "step": 1934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019517403680151913,
+      "loss": 0.0986,
+      "step": 1935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019516754414201944,
+      "loss": 0.0991,
+      "step": 1936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001951610472261114,
+      "loss": 0.0968,
+      "step": 1937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019515454605408548,
+      "loss": 0.0982,
+      "step": 1938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019514804062623246,
+      "loss": 0.1087,
+      "step": 1939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019514153094284334,
+      "loss": 0.0968,
+      "step": 1940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019513501700420927,
+      "loss": 0.0952,
+      "step": 1941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019512849881062153,
+      "loss": 0.0931,
+      "step": 1942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019512197636237173,
+      "loss": 0.0971,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019511544965975154,
+      "loss": 0.0973,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019510891870305286,
+      "loss": 0.0908,
+      "step": 1945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001951023834925678,
+      "loss": 0.1055,
+      "step": 1946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001950958440285887,
+      "loss": 0.0997,
+      "step": 1947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019508930031140795,
+      "loss": 0.0998,
+      "step": 1948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001950827523413183,
+      "loss": 0.0979,
+      "step": 1949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019507620011861257,
+      "loss": 0.1088,
+      "step": 1950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001950696436435838,
+      "loss": 0.0998,
+      "step": 1951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019506308291652525,
+      "loss": 0.0991,
+      "step": 1952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019505651793773037,
+      "loss": 0.104,
+      "step": 1953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019504994870749276,
+      "loss": 0.099,
+      "step": 1954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019504337522610623,
+      "loss": 0.1054,
+      "step": 1955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019503679749386478,
+      "loss": 0.1015,
+      "step": 1956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019503021551106264,
+      "loss": 0.0937,
+      "step": 1957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001950236292779941,
+      "loss": 0.1002,
+      "step": 1958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019501703879495384,
+      "loss": 0.0982,
+      "step": 1959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001950104440622366,
+      "loss": 0.0938,
+      "step": 1960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019500384508013726,
+      "loss": 0.1027,
+      "step": 1961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000194997241848951,
+      "loss": 0.097,
+      "step": 1962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019499063436897318,
+      "loss": 0.0985,
+      "step": 1963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001949840226404993,
+      "loss": 0.0974,
+      "step": 1964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019497740666382508,
+      "loss": 0.1024,
+      "step": 1965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019497078643924642,
+      "loss": 0.1,
+      "step": 1966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019496416196705942,
+      "loss": 0.0991,
+      "step": 1967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019495753324756034,
+      "loss": 0.0929,
+      "step": 1968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019495090028104564,
+      "loss": 0.1004,
+      "step": 1969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019494426306781204,
+      "loss": 0.0962,
+      "step": 1970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019493762160815634,
+      "loss": 0.096,
+      "step": 1971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019493097590237563,
+      "loss": 0.0964,
+      "step": 1972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001949243259507671,
+      "loss": 0.0999,
+      "step": 1973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019491767175362816,
+      "loss": 0.1024,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019491101331125648,
+      "loss": 0.0969,
+      "step": 1975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001949043506239498,
+      "loss": 0.0985,
+      "step": 1976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001948976836920062,
+      "loss": 0.0982,
+      "step": 1977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019489101251572372,
+      "loss": 0.1032,
+      "step": 1978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019488433709540087,
+      "loss": 0.1011,
+      "step": 1979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019487765743133613,
+      "loss": 0.0968,
+      "step": 1980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001948709735238283,
+      "loss": 0.098,
+      "step": 1981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019486428537317626,
+      "loss": 0.0944,
+      "step": 1982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019485759297967916,
+      "loss": 0.1,
+      "step": 1983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019485089634363638,
+      "loss": 0.1027,
+      "step": 1984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019484419546534737,
+      "loss": 0.097,
+      "step": 1985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019483749034511182,
+      "loss": 0.0972,
+      "step": 1986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001948307809832297,
+      "loss": 0.0968,
+      "step": 1987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019482406738000098,
+      "loss": 0.0954,
+      "step": 1988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000194817349535726,
+      "loss": 0.0957,
+      "step": 1989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019481062745070523,
+      "loss": 0.0965,
+      "step": 1990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019480390112523925,
+      "loss": 0.097,
+      "step": 1991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019479717055962895,
+      "loss": 0.0991,
+      "step": 1992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019479043575417536,
+      "loss": 0.0931,
+      "step": 1993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019478369670917963,
+      "loss": 0.0994,
+      "step": 1994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019477695342494324,
+      "loss": 0.0992,
+      "step": 1995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001947702059017678,
+      "loss": 0.0989,
+      "step": 1996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019476345413995503,
+      "loss": 0.1004,
+      "step": 1997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019475669813980693,
+      "loss": 0.0967,
+      "step": 1998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019474993790162567,
+      "loss": 0.0982,
+      "step": 1999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001947431734257136,
+      "loss": 0.1006,
+      "step": 2000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001947364047123733,
+      "loss": 0.1003,
+      "step": 2001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019472963176190745,
+      "loss": 0.1035,
+      "step": 2002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019472285457461898,
+      "loss": 0.0991,
+      "step": 2003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019471607315081104,
+      "loss": 0.0961,
+      "step": 2004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001947092874907869,
+      "loss": 0.1011,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019470249759485006,
+      "loss": 0.0947,
+      "step": 2006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001946957034633042,
+      "loss": 0.0953,
+      "step": 2007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019468890509645318,
+      "loss": 0.1032,
+      "step": 2008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001946821024946011,
+      "loss": 0.0977,
+      "step": 2009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019467529565805216,
+      "loss": 0.0983,
+      "step": 2010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019466848458711084,
+      "loss": 0.0924,
+      "step": 2011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019466166928208174,
+      "loss": 0.095,
+      "step": 2012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001946548497432697,
+      "loss": 0.0942,
+      "step": 2013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019464802597097966,
+      "loss": 0.1037,
+      "step": 2014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001946411979655169,
+      "loss": 0.0992,
+      "step": 2015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001946343657271868,
+      "loss": 0.0979,
+      "step": 2016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019462752925629486,
+      "loss": 0.0983,
+      "step": 2017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019462068855314693,
+      "loss": 0.097,
+      "step": 2018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001946138436180489,
+      "loss": 0.0982,
+      "step": 2019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019460699445130695,
+      "loss": 0.0953,
+      "step": 2020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019460014105322738,
+      "loss": 0.0916,
+      "step": 2021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001945932834241168,
+      "loss": 0.1012,
+      "step": 2022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019458642156428178,
+      "loss": 0.0942,
+      "step": 2023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019457955547402928,
+      "loss": 0.1016,
+      "step": 2024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019457268515366645,
+      "loss": 0.1054,
+      "step": 2025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001945658106035005,
+      "loss": 0.0948,
+      "step": 2026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001945589318238389,
+      "loss": 0.0956,
+      "step": 2027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019455204881498934,
+      "loss": 0.1,
+      "step": 2028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019454516157725964,
+      "loss": 0.0983,
+      "step": 2029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019453827011095784,
+      "loss": 0.1018,
+      "step": 2030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001945313744163922,
+      "loss": 0.0994,
+      "step": 2031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019452447449387104,
+      "loss": 0.1042,
+      "step": 2032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019451757034370305,
+      "loss": 0.0976,
+      "step": 2033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019451066196619698,
+      "loss": 0.1014,
+      "step": 2034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001945037493616618,
+      "loss": 0.1002,
+      "step": 2035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019449683253040676,
+      "loss": 0.1018,
+      "step": 2036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019448991147274112,
+      "loss": 0.0959,
+      "step": 2037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019448298618897445,
+      "loss": 0.1024,
+      "step": 2038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019447605667941652,
+      "loss": 0.0985,
+      "step": 2039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019446912294437724,
+      "loss": 0.099,
+      "step": 2040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019446218498416672,
+      "loss": 0.0952,
+      "step": 2041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001944552427990953,
+      "loss": 0.0986,
+      "step": 2042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019444829638947338,
+      "loss": 0.0984,
+      "step": 2043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001944413457556117,
+      "loss": 0.1,
+      "step": 2044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019443439089782114,
+      "loss": 0.1016,
+      "step": 2045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019442743181641276,
+      "loss": 0.0937,
+      "step": 2046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001944204685116978,
+      "loss": 0.0996,
+      "step": 2047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019441350098398766,
+      "loss": 0.1001,
+      "step": 2048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000194406529233594,
+      "loss": 0.0931,
+      "step": 2049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019439955326082866,
+      "loss": 0.1002,
+      "step": 2050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001943925730660036,
+      "loss": 0.1013,
+      "step": 2051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019438558864943103,
+      "loss": 0.0964,
+      "step": 2052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019437860001142332,
+      "loss": 0.0986,
+      "step": 2053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019437160715229305,
+      "loss": 0.106,
+      "step": 2054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019436461007235297,
+      "loss": 0.1003,
+      "step": 2055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019435760877191603,
+      "loss": 0.0955,
+      "step": 2056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001943506032512954,
+      "loss": 0.0969,
+      "step": 2057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019434359351080434,
+      "loss": 0.0929,
+      "step": 2058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001943365795507564,
+      "loss": 0.0943,
+      "step": 2059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001943295613714653,
+      "loss": 0.1019,
+      "step": 2060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019432253897324491,
+      "loss": 0.0972,
+      "step": 2061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019431551235640928,
+      "loss": 0.1045,
+      "step": 2062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019430848152127275,
+      "loss": 0.0997,
+      "step": 2063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019430144646814972,
+      "loss": 0.0981,
+      "step": 2064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019429440719735482,
+      "loss": 0.0986,
+      "step": 2065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019428736370920296,
+      "loss": 0.1011,
+      "step": 2066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001942803160040091,
+      "loss": 0.097,
+      "step": 2067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001942732640820885,
+      "loss": 0.1039,
+      "step": 2068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001942662079437565,
+      "loss": 0.0987,
+      "step": 2069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019425914758932874,
+      "loss": 0.1021,
+      "step": 2070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001942520830191209,
+      "loss": 0.0968,
+      "step": 2071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019424501423344912,
+      "loss": 0.0982,
+      "step": 2072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019423794123262945,
+      "loss": 0.0968,
+      "step": 2073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019423086401697818,
+      "loss": 0.096,
+      "step": 2074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019422378258681194,
+      "loss": 0.0996,
+      "step": 2075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001942166969424474,
+      "loss": 0.0952,
+      "step": 2076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001942096070842015,
+      "loss": 0.0953,
+      "step": 2077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019420251301239128,
+      "loss": 0.1033,
+      "step": 2078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001941954147273341,
+      "loss": 0.1033,
+      "step": 2079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001941883122293474,
+      "loss": 0.0997,
+      "step": 2080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001941812055187488,
+      "loss": 0.0985,
+      "step": 2081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019417409459585621,
+      "loss": 0.0968,
+      "step": 2082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019416697946098764,
+      "loss": 0.0959,
+      "step": 2083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019415986011446135,
+      "loss": 0.0999,
+      "step": 2084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019415273655659572,
+      "loss": 0.0889,
+      "step": 2085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019414560878770935,
+      "loss": 0.0989,
+      "step": 2086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019413847680812104,
+      "loss": 0.0949,
+      "step": 2087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019413134061814982,
+      "loss": 0.1009,
+      "step": 2088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019412420021811477,
+      "loss": 0.0948,
+      "step": 2089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019411705560833532,
+      "loss": 0.1016,
+      "step": 2090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000194109906789131,
+      "loss": 0.0992,
+      "step": 2091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019410275376082153,
+      "loss": 0.1007,
+      "step": 2092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019409559652372682,
+      "loss": 0.0992,
+      "step": 2093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019408843507816702,
+      "loss": 0.1012,
+      "step": 2094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001940812694244624,
+      "loss": 0.1003,
+      "step": 2095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019407409956293344,
+      "loss": 0.1038,
+      "step": 2096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019406692549390082,
+      "loss": 0.099,
+      "step": 2097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019405974721768544,
+      "loss": 0.0949,
+      "step": 2098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001940525647346083,
+      "loss": 0.099,
+      "step": 2099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001940453780449907,
+      "loss": 0.0961,
+      "step": 2100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019403818714915395,
+      "loss": 0.0927,
+      "step": 2101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001940309920474198,
+      "loss": 0.0965,
+      "step": 2102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019402379274011,
+      "loss": 0.0996,
+      "step": 2103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019401658922754656,
+      "loss": 0.1013,
+      "step": 2104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001940093815100516,
+      "loss": 0.094,
+      "step": 2105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019400216958794756,
+      "loss": 0.1028,
+      "step": 2106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019399495346155693,
+      "loss": 0.1037,
+      "step": 2107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001939877331312025,
+      "loss": 0.0969,
+      "step": 2108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019398050859720725,
+      "loss": 0.095,
+      "step": 2109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019397327985989416,
+      "loss": 0.0913,
+      "step": 2110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019396604691958668,
+      "loss": 0.0989,
+      "step": 2111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019395880977660822,
+      "loss": 0.101,
+      "step": 2112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019395156843128253,
+      "loss": 0.0964,
+      "step": 2113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019394432288393342,
+      "loss": 0.088,
+      "step": 2114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019393707313488493,
+      "loss": 0.0997,
+      "step": 2115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019392981918446141,
+      "loss": 0.0966,
+      "step": 2116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019392256103298723,
+      "loss": 0.0995,
+      "step": 2117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019391529868078701,
+      "loss": 0.0964,
+      "step": 2118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019390803212818557,
+      "loss": 0.0997,
+      "step": 2119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001939007613755079,
+      "loss": 0.0954,
+      "step": 2120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019389348642307925,
+      "loss": 0.0975,
+      "step": 2121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019388620727122492,
+      "loss": 0.1002,
+      "step": 2122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001938789239202705,
+      "loss": 0.0942,
+      "step": 2123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019387163637054174,
+      "loss": 0.0925,
+      "step": 2124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019386434462236455,
+      "loss": 0.0946,
+      "step": 2125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019385704867606513,
+      "loss": 0.1007,
+      "step": 2126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019384974853196973,
+      "loss": 0.1002,
+      "step": 2127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001938424441904049,
+      "loss": 0.0992,
+      "step": 2128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019383513565169726,
+      "loss": 0.0976,
+      "step": 2129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019382782291617377,
+      "loss": 0.102,
+      "step": 2130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001938205059841614,
+      "loss": 0.1003,
+      "step": 2131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001938131848559875,
+      "loss": 0.1004,
+      "step": 2132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019380585953197944,
+      "loss": 0.0945,
+      "step": 2133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001937985300124649,
+      "loss": 0.0981,
+      "step": 2134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019379119629777167,
+      "loss": 0.1008,
+      "step": 2135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019378385838822774,
+      "loss": 0.0903,
+      "step": 2136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019377651628416133,
+      "loss": 0.0994,
+      "step": 2137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001937691699859008,
+      "loss": 0.0985,
+      "step": 2138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001937618194937747,
+      "loss": 0.0974,
+      "step": 2139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019375446480811187,
+      "loss": 0.0974,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019374710592924113,
+      "loss": 0.1005,
+      "step": 2141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019373974285749168,
+      "loss": 0.0887,
+      "step": 2142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001937323755931928,
+      "loss": 0.0964,
+      "step": 2143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019372500413667406,
+      "loss": 0.102,
+      "step": 2144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019371762848826508,
+      "loss": 0.0976,
+      "step": 2145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019371024864829576,
+      "loss": 0.096,
+      "step": 2146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001937028646170962,
+      "loss": 0.0964,
+      "step": 2147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001936954763949966,
+      "loss": 0.0969,
+      "step": 2148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019368808398232746,
+      "loss": 0.0991,
+      "step": 2149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019368068737941934,
+      "loss": 0.0971,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001936732865866031,
+      "loss": 0.1006,
+      "step": 2151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019366588160420976,
+      "loss": 0.1049,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019365847243257048,
+      "loss": 0.0983,
+      "step": 2153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019365105907201663,
+      "loss": 0.0999,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001936436415228798,
+      "loss": 0.0983,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019363621978549174,
+      "loss": 0.0965,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019362879386018437,
+      "loss": 0.102,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019362136374728985,
+      "loss": 0.1,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019361392944714049,
+      "loss": 0.1006,
+      "step": 2159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019360649096006874,
+      "loss": 0.0961,
+      "step": 2160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019359904828640737,
+      "loss": 0.0991,
+      "step": 2161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001935916014264892,
+      "loss": 0.101,
+      "step": 2162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001935841503806473,
+      "loss": 0.1011,
+      "step": 2163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019357669514921493,
+      "loss": 0.0906,
+      "step": 2164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019356923573252553,
+      "loss": 0.0975,
+      "step": 2165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019356177213091273,
+      "loss": 0.1014,
+      "step": 2166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019355430434471035,
+      "loss": 0.0971,
+      "step": 2167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019354683237425238,
+      "loss": 0.1018,
+      "step": 2168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019353935621987297,
+      "loss": 0.0951,
+      "step": 2169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019353187588190654,
+      "loss": 0.0954,
+      "step": 2170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019352439136068768,
+      "loss": 0.0954,
+      "step": 2171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019351690265655104,
+      "loss": 0.0887,
+      "step": 2172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019350940976983164,
+      "loss": 0.1008,
+      "step": 2173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001935019127008646,
+      "loss": 0.1007,
+      "step": 2174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019349441144998517,
+      "loss": 0.0933,
+      "step": 2175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001934869060175289,
+      "loss": 0.1005,
+      "step": 2176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019347939640383145,
+      "loss": 0.0989,
+      "step": 2177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001934718826092287,
+      "loss": 0.087,
+      "step": 2178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001934643646340567,
+      "loss": 0.0936,
+      "step": 2179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001934568424786517,
+      "loss": 0.0925,
+      "step": 2180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019344931614335013,
+      "loss": 0.0999,
+      "step": 2181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001934417856284886,
+      "loss": 0.0983,
+      "step": 2182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019343425093440396,
+      "loss": 0.1022,
+      "step": 2183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019342671206143313,
+      "loss": 0.0977,
+      "step": 2184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019341916900991334,
+      "loss": 0.1028,
+      "step": 2185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019341162178018192,
+      "loss": 0.1,
+      "step": 2186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001934040703725765,
+      "loss": 0.0967,
+      "step": 2187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019339651478743472,
+      "loss": 0.1038,
+      "step": 2188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019338895502509456,
+      "loss": 0.0958,
+      "step": 2189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019338139108589412,
+      "loss": 0.1005,
+      "step": 2190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001933738229701717,
+      "loss": 0.0988,
+      "step": 2191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019336625067826582,
+      "loss": 0.1038,
+      "step": 2192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019335867421051514,
+      "loss": 0.0932,
+      "step": 2193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019335109356725843,
+      "loss": 0.0964,
+      "step": 2194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001933435087488349,
+      "loss": 0.0975,
+      "step": 2195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019333591975558364,
+      "loss": 0.1011,
+      "step": 2196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001933283265878442,
+      "loss": 0.0991,
+      "step": 2197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019332072924595605,
+      "loss": 0.1026,
+      "step": 2198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019331312773025907,
+      "loss": 0.0908,
+      "step": 2199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019330552204109322,
+      "loss": 0.0992,
+      "step": 2200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019329791217879868,
+      "loss": 0.0964,
+      "step": 2201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019329029814371578,
+      "loss": 0.0928,
+      "step": 2202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001932826799361851,
+      "loss": 0.0918,
+      "step": 2203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019327505755654733,
+      "loss": 0.0989,
+      "step": 2204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001932674310051434,
+      "loss": 0.101,
+      "step": 2205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019325980028231442,
+      "loss": 0.0963,
+      "step": 2206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019325216538840166,
+      "loss": 0.0964,
+      "step": 2207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019324452632374657,
+      "loss": 0.0995,
+      "step": 2208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019323688308869088,
+      "loss": 0.1019,
+      "step": 2209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019322923568357636,
+      "loss": 0.103,
+      "step": 2210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001932215841087451,
+      "loss": 0.1008,
+      "step": 2211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019321392836453932,
+      "loss": 0.0956,
+      "step": 2212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019320626845130136,
+      "loss": 0.1006,
+      "step": 2213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001931986043693739,
+      "loss": 0.0948,
+      "step": 2214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019319093611909968,
+      "loss": 0.0977,
+      "step": 2215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019318326370082163,
+      "loss": 0.1008,
+      "step": 2216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019317558711488296,
+      "loss": 0.0958,
+      "step": 2217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019316790636162695,
+      "loss": 0.0959,
+      "step": 2218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001931602214413972,
+      "loss": 0.0954,
+      "step": 2219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001931525323545374,
+      "loss": 0.0924,
+      "step": 2220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001931448391013914,
+      "loss": 0.094,
+      "step": 2221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019313714168230328,
+      "loss": 0.0899,
+      "step": 2222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019312944009761737,
+      "loss": 0.0952,
+      "step": 2223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001931217343476781,
+      "loss": 0.0929,
+      "step": 2224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019311402443283014,
+      "loss": 0.1064,
+      "step": 2225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019310631035341828,
+      "loss": 0.0978,
+      "step": 2226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019309859210978752,
+      "loss": 0.1007,
+      "step": 2227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019309086970228307,
+      "loss": 0.0964,
+      "step": 2228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001930831431312504,
+      "loss": 0.0997,
+      "step": 2229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019307541239703497,
+      "loss": 0.0891,
+      "step": 2230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019306767749998263,
+      "loss": 0.1002,
+      "step": 2231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019305993844043925,
+      "loss": 0.0981,
+      "step": 2232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019305219521875103,
+      "loss": 0.1018,
+      "step": 2233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019304444783526425,
+      "loss": 0.0942,
+      "step": 2234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019303669629032542,
+      "loss": 0.1044,
+      "step": 2235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019302894058428126,
+      "loss": 0.1019,
+      "step": 2236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019302118071747856,
+      "loss": 0.0914,
+      "step": 2237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001930134166902645,
+      "loss": 0.0985,
+      "step": 2238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019300564850298626,
+      "loss": 0.1025,
+      "step": 2239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001929978761559913,
+      "loss": 0.1007,
+      "step": 2240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001929900996496272,
+      "loss": 0.1024,
+      "step": 2241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019298231898424184,
+      "loss": 0.1046,
+      "step": 2242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019297453416018315,
+      "loss": 0.0953,
+      "step": 2243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019296674517779936,
+      "loss": 0.097,
+      "step": 2244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019295895203743876,
+      "loss": 0.0961,
+      "step": 2245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019295115473945,
+      "loss": 0.1034,
+      "step": 2246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019294335328418175,
+      "loss": 0.0982,
+      "step": 2247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019293554767198294,
+      "loss": 0.0974,
+      "step": 2248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001929277379032027,
+      "loss": 0.1016,
+      "step": 2249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001929199239781903,
+      "loss": 0.0968,
+      "step": 2250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019291210589729524,
+      "loss": 0.0957,
+      "step": 2251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001929042836608672,
+      "loss": 0.1016,
+      "step": 2252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000192896457269256,
+      "loss": 0.1055,
+      "step": 2253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001928886267228117,
+      "loss": 0.0998,
+      "step": 2254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001928807920218845,
+      "loss": 0.094,
+      "step": 2255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019287295316682487,
+      "loss": 0.1003,
+      "step": 2256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019286511015798334,
+      "loss": 0.0981,
+      "step": 2257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019285726299571074,
+      "loss": 0.0889,
+      "step": 2258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000192849411680358,
+      "loss": 0.1012,
+      "step": 2259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019284155621227626,
+      "loss": 0.0942,
+      "step": 2260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019283369659181692,
+      "loss": 0.0964,
+      "step": 2261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019282583281933147,
+      "loss": 0.1027,
+      "step": 2262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019281796489517162,
+      "loss": 0.0956,
+      "step": 2263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019281009281968924,
+      "loss": 0.0948,
+      "step": 2264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019280221659323648,
+      "loss": 0.1023,
+      "step": 2265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019279433621616557,
+      "loss": 0.1031,
+      "step": 2266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001927864516888289,
+      "loss": 0.1044,
+      "step": 2267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019277856301157926,
+      "loss": 0.09,
+      "step": 2268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019277067018476932,
+      "loss": 0.0971,
+      "step": 2269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001927627732087522,
+      "loss": 0.1001,
+      "step": 2270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000192754872083881,
+      "loss": 0.103,
+      "step": 2271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001927469668105092,
+      "loss": 0.0933,
+      "step": 2272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019273905738899027,
+      "loss": 0.1004,
+      "step": 2273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019273114381967806,
+      "loss": 0.0985,
+      "step": 2274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019272322610292643,
+      "loss": 0.096,
+      "step": 2275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019271530423908956,
+      "loss": 0.0946,
+      "step": 2276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019270737822852166,
+      "loss": 0.0993,
+      "step": 2277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019269944807157735,
+      "loss": 0.1067,
+      "step": 2278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019269151376861125,
+      "loss": 0.0954,
+      "step": 2279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019268357531997823,
+      "loss": 0.0957,
+      "step": 2280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019267563272603333,
+      "loss": 0.0966,
+      "step": 2281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019266768598713178,
+      "loss": 0.1009,
+      "step": 2282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019265973510362904,
+      "loss": 0.102,
+      "step": 2283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001926517800758807,
+      "loss": 0.0972,
+      "step": 2284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019264382090424252,
+      "loss": 0.0935,
+      "step": 2285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001926358575890705,
+      "loss": 0.0951,
+      "step": 2286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019262789013072082,
+      "loss": 0.101,
+      "step": 2287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001926199185295498,
+      "loss": 0.1034,
+      "step": 2288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000192611942785914,
+      "loss": 0.0944,
+      "step": 2289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019260396290017012,
+      "loss": 0.1016,
+      "step": 2290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019259597887267508,
+      "loss": 0.0939,
+      "step": 2291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001925879907037859,
+      "loss": 0.0977,
+      "step": 2292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019257999839386,
+      "loss": 0.098,
+      "step": 2293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001925720019432547,
+      "loss": 0.1039,
+      "step": 2294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001925640013523277,
+      "loss": 0.1031,
+      "step": 2295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019255599662143684,
+      "loss": 0.0971,
+      "step": 2296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001925479877509401,
+      "loss": 0.1044,
+      "step": 2297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019253997474119573,
+      "loss": 0.1046,
+      "step": 2298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019253195759256207,
+      "loss": 0.0922,
+      "step": 2299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019252393630539766,
+      "loss": 0.0953,
+      "step": 2300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019251591088006136,
+      "loss": 0.0961,
+      "step": 2301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019250788131691202,
+      "loss": 0.1024,
+      "step": 2302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019249984761630882,
+      "loss": 0.0996,
+      "step": 2303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019249180977861104,
+      "loss": 0.0964,
+      "step": 2304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019248376780417817,
+      "loss": 0.0922,
+      "step": 2305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019247572169336992,
+      "loss": 0.0983,
+      "step": 2306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019246767144654614,
+      "loss": 0.1009,
+      "step": 2307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019245961706406688,
+      "loss": 0.0972,
+      "step": 2308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019245155854629237,
+      "loss": 0.0964,
+      "step": 2309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019244349589358305,
+      "loss": 0.0945,
+      "step": 2310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019243542910629948,
+      "loss": 0.0993,
+      "step": 2311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019242735818480252,
+      "loss": 0.1008,
+      "step": 2312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019241928312945308,
+      "loss": 0.0995,
+      "step": 2313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001924112039406124,
+      "loss": 0.0933,
+      "step": 2314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001924031206186417,
+      "loss": 0.0951,
+      "step": 2315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001923950331639026,
+      "loss": 0.1045,
+      "step": 2316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019238694157675684,
+      "loss": 0.0908,
+      "step": 2317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019237884585756625,
+      "loss": 0.0974,
+      "step": 2318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019237074600669295,
+      "loss": 0.1012,
+      "step": 2319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001923626420244992,
+      "loss": 0.0981,
+      "step": 2320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019235453391134745,
+      "loss": 0.1,
+      "step": 2321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019234642166760036,
+      "loss": 0.1005,
+      "step": 2322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019233830529362075,
+      "loss": 0.0933,
+      "step": 2323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019233018478977156,
+      "loss": 0.1012,
+      "step": 2324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001923220601564161,
+      "loss": 0.1038,
+      "step": 2325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019231393139391764,
+      "loss": 0.1024,
+      "step": 2326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001923057985026398,
+      "loss": 0.1009,
+      "step": 2327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019229766148294632,
+      "loss": 0.0893,
+      "step": 2328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019228952033520114,
+      "loss": 0.1032,
+      "step": 2329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019228137505976838,
+      "loss": 0.0908,
+      "step": 2330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001922732256570123,
+      "loss": 0.0961,
+      "step": 2331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001922650721272974,
+      "loss": 0.1,
+      "step": 2332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001922569144709884,
+      "loss": 0.0934,
+      "step": 2333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019224875268845007,
+      "loss": 0.1047,
+      "step": 2334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001922405867800475,
+      "loss": 0.1002,
+      "step": 2335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019223241674614592,
+      "loss": 0.0946,
+      "step": 2336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019222424258711073,
+      "loss": 0.1002,
+      "step": 2337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019221606430330752,
+      "loss": 0.0954,
+      "step": 2338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019220788189510207,
+      "loss": 0.0971,
+      "step": 2339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019219969536286033,
+      "loss": 0.0971,
+      "step": 2340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019219150470694844,
+      "loss": 0.0997,
+      "step": 2341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019218330992773277,
+      "loss": 0.0929,
+      "step": 2342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019217511102557978,
+      "loss": 0.094,
+      "step": 2343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019216690800085621,
+      "loss": 0.0967,
+      "step": 2344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019215870085392895,
+      "loss": 0.0965,
+      "step": 2345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019215048958516503,
+      "loss": 0.0975,
+      "step": 2346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019214227419493172,
+      "loss": 0.0986,
+      "step": 2347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001921340546835965,
+      "loss": 0.0965,
+      "step": 2348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001921258310515269,
+      "loss": 0.0943,
+      "step": 2349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019211760329909078,
+      "loss": 0.0992,
+      "step": 2350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019210937142665616,
+      "loss": 0.0935,
+      "step": 2351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019210113543459116,
+      "loss": 0.0932,
+      "step": 2352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019209289532326417,
+      "loss": 0.0986,
+      "step": 2353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001920846510930437,
+      "loss": 0.097,
+      "step": 2354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001920764027442985,
+      "loss": 0.097,
+      "step": 2355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019206815027739748,
+      "loss": 0.0979,
+      "step": 2356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019205989369270975,
+      "loss": 0.0972,
+      "step": 2357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019205163299060456,
+      "loss": 0.098,
+      "step": 2358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001920433681714514,
+      "loss": 0.0994,
+      "step": 2359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019203509923561987,
+      "loss": 0.0993,
+      "step": 2360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019202682618347985,
+      "loss": 0.0957,
+      "step": 2361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019201854901540135,
+      "loss": 0.0973,
+      "step": 2362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019201026773175457,
+      "loss": 0.1001,
+      "step": 2363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019200198233290983,
+      "loss": 0.1003,
+      "step": 2364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001919936928192378,
+      "loss": 0.0974,
+      "step": 2365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001919853991911092,
+      "loss": 0.0969,
+      "step": 2366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001919771014488949,
+      "loss": 0.1021,
+      "step": 2367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001919687995929661,
+      "loss": 0.0974,
+      "step": 2368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019196049362369406,
+      "loss": 0.104,
+      "step": 2369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001919521835414503,
+      "loss": 0.0957,
+      "step": 2370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019194386934660648,
+      "loss": 0.0993,
+      "step": 2371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019193555103953443,
+      "loss": 0.0987,
+      "step": 2372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001919272286206062,
+      "loss": 0.101,
+      "step": 2373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019191890209019406,
+      "loss": 0.098,
+      "step": 2374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019191057144867034,
+      "loss": 0.0978,
+      "step": 2375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019190223669640768,
+      "loss": 0.0992,
+      "step": 2376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019189389783377887,
+      "loss": 0.1017,
+      "step": 2377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019188555486115681,
+      "loss": 0.0964,
+      "step": 2378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019187720777891473,
+      "loss": 0.093,
+      "step": 2379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019186885658742587,
+      "loss": 0.1021,
+      "step": 2380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019186050128706378,
+      "loss": 0.0999,
+      "step": 2381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019185214187820217,
+      "loss": 0.098,
+      "step": 2382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001918437783612149,
+      "loss": 0.1002,
+      "step": 2383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000191835410736476,
+      "loss": 0.1052,
+      "step": 2384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019182703900435976,
+      "loss": 0.1019,
+      "step": 2385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001918186631652406,
+      "loss": 0.103,
+      "step": 2386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019181028321949315,
+      "loss": 0.0967,
+      "step": 2387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019180189916749216,
+      "loss": 0.1022,
+      "step": 2388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019179351100961263,
+      "loss": 0.1035,
+      "step": 2389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019178511874622976,
+      "loss": 0.1043,
+      "step": 2390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019177672237771884,
+      "loss": 0.0958,
+      "step": 2391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019176832190445544,
+      "loss": 0.0981,
+      "step": 2392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019175991732681526,
+      "loss": 0.0969,
+      "step": 2393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019175150864517423,
+      "loss": 0.1004,
+      "step": 2394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019174309585990836,
+      "loss": 0.0964,
+      "step": 2395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019173467897139394,
+      "loss": 0.1002,
+      "step": 2396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001917262579800075,
+      "loss": 0.0951,
+      "step": 2397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019171783288612557,
+      "loss": 0.0953,
+      "step": 2398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019170940369012503,
+      "loss": 0.0991,
+      "step": 2399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019170097039238285,
+      "loss": 0.0932,
+      "step": 2400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019169253299327618,
+      "loss": 0.095,
+      "step": 2401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019168409149318246,
+      "loss": 0.1001,
+      "step": 2402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001916756458924792,
+      "loss": 0.1006,
+      "step": 2403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019166719619154412,
+      "loss": 0.0968,
+      "step": 2404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019165874239075517,
+      "loss": 0.1011,
+      "step": 2405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019165028449049043,
+      "loss": 0.0989,
+      "step": 2406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019164182249112817,
+      "loss": 0.0949,
+      "step": 2407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019163335639304688,
+      "loss": 0.098,
+      "step": 2408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019162488619662522,
+      "loss": 0.1003,
+      "step": 2409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000191616411902242,
+      "loss": 0.1008,
+      "step": 2410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019160793351027624,
+      "loss": 0.0952,
+      "step": 2411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019159945102110713,
+      "loss": 0.0971,
+      "step": 2412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019159096443511407,
+      "loss": 0.0991,
+      "step": 2413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001915824737526766,
+      "loss": 0.0972,
+      "step": 2414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019157397897417452,
+      "loss": 0.1039,
+      "step": 2415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019156548009998775,
+      "loss": 0.1004,
+      "step": 2416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019155697713049637,
+      "loss": 0.0959,
+      "step": 2417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019154847006608067,
+      "loss": 0.098,
+      "step": 2418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019153995890712117,
+      "loss": 0.0949,
+      "step": 2419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019153144365399854,
+      "loss": 0.0954,
+      "step": 2420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001915229243070936,
+      "loss": 0.0992,
+      "step": 2421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001915144008667874,
+      "loss": 0.097,
+      "step": 2422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019150587333346115,
+      "loss": 0.0917,
+      "step": 2423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019149734170749626,
+      "loss": 0.0965,
+      "step": 2424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019148880598927423,
+      "loss": 0.0922,
+      "step": 2425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019148026617917696,
+      "loss": 0.1066,
+      "step": 2426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001914717222775863,
+      "loss": 0.0972,
+      "step": 2427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001914631742848844,
+      "loss": 0.0947,
+      "step": 2428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019145462220145358,
+      "loss": 0.0962,
+      "step": 2429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019144606602767636,
+      "loss": 0.0955,
+      "step": 2430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019143750576393537,
+      "loss": 0.0877,
+      "step": 2431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001914289414106135,
+      "loss": 0.097,
+      "step": 2432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019142037296809378,
+      "loss": 0.1018,
+      "step": 2433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019141180043675946,
+      "loss": 0.0992,
+      "step": 2434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019140322381699392,
+      "loss": 0.0939,
+      "step": 2435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001913946431091808,
+      "loss": 0.0968,
+      "step": 2436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001913860583137038,
+      "loss": 0.0946,
+      "step": 2437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019137746943094695,
+      "loss": 0.0933,
+      "step": 2438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019136887646129436,
+      "loss": 0.1021,
+      "step": 2439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019136027940513034,
+      "loss": 0.0985,
+      "step": 2440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019135167826283945,
+      "loss": 0.1038,
+      "step": 2441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019134307303480633,
+      "loss": 0.0945,
+      "step": 2442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019133446372141585,
+      "loss": 0.0961,
+      "step": 2443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001913258503230531,
+      "loss": 0.1015,
+      "step": 2444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001913172328401033,
+      "loss": 0.095,
+      "step": 2445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019130861127295186,
+      "loss": 0.1006,
+      "step": 2446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019129998562198442,
+      "loss": 0.0976,
+      "step": 2447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019129135588758672,
+      "loss": 0.1007,
+      "step": 2448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019128272207014473,
+      "loss": 0.098,
+      "step": 2449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019127408417004464,
+      "loss": 0.0967,
+      "step": 2450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019126544218767277,
+      "loss": 0.1016,
+      "step": 2451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019125679612341562,
+      "loss": 0.099,
+      "step": 2452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001912481459776599,
+      "loss": 0.1008,
+      "step": 2453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001912394917507925,
+      "loss": 0.0943,
+      "step": 2454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019123083344320046,
+      "loss": 0.0983,
+      "step": 2455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019122217105527103,
+      "loss": 0.0954,
+      "step": 2456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001912135045873917,
+      "loss": 0.0993,
+      "step": 2457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019120483403994997,
+      "loss": 0.0957,
+      "step": 2458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019119615941333372,
+      "loss": 0.094,
+      "step": 2459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001911874807079309,
+      "loss": 0.0972,
+      "step": 2460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019117879792412967,
+      "loss": 0.0942,
+      "step": 2461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001911701110623184,
+      "loss": 0.1059,
+      "step": 2462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001911614201228855,
+      "loss": 0.0982,
+      "step": 2463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019115272510621985,
+      "loss": 0.1025,
+      "step": 2464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019114402601271019,
+      "loss": 0.102,
+      "step": 2465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001911353228427457,
+      "loss": 0.0945,
+      "step": 2466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019112661559671557,
+      "loss": 0.1029,
+      "step": 2467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001911179042750092,
+      "loss": 0.0944,
+      "step": 2468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019110918887801633,
+      "loss": 0.0965,
+      "step": 2469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001911004694061266,
+      "loss": 0.0955,
+      "step": 2470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019109174585973013,
+      "loss": 0.0991,
+      "step": 2471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000191083018239217,
+      "loss": 0.098,
+      "step": 2472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019107428654497764,
+      "loss": 0.0969,
+      "step": 2473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001910655507774025,
+      "loss": 0.097,
+      "step": 2474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001910568109368823,
+      "loss": 0.0972,
+      "step": 2475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019104806702380795,
+      "loss": 0.0947,
+      "step": 2476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019103931903857054,
+      "loss": 0.1009,
+      "step": 2477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019103056698156133,
+      "loss": 0.1014,
+      "step": 2478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001910218108531717,
+      "loss": 0.0955,
+      "step": 2479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019101305065379336,
+      "loss": 0.0947,
+      "step": 2480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019100428638381804,
+      "loss": 0.0995,
+      "step": 2481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019099551804363776,
+      "loss": 0.0961,
+      "step": 2482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019098674563364468,
+      "loss": 0.0961,
+      "step": 2483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019097796915423115,
+      "loss": 0.0952,
+      "step": 2484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019096918860578968,
+      "loss": 0.0995,
+      "step": 2485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000190960403988713,
+      "loss": 0.1005,
+      "step": 2486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019095161530339406,
+      "loss": 0.0995,
+      "step": 2487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019094282255022586,
+      "loss": 0.0933,
+      "step": 2488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001909340257296017,
+      "loss": 0.0946,
+      "step": 2489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000190925224841915,
+      "loss": 0.1036,
+      "step": 2490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001909164198875594,
+      "loss": 0.0976,
+      "step": 2491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001909076108669287,
+      "loss": 0.0979,
+      "step": 2492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019089879778041687,
+      "loss": 0.0949,
+      "step": 2493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019088998062841812,
+      "loss": 0.0954,
+      "step": 2494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019088115941132675,
+      "loss": 0.106,
+      "step": 2495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019087233412953732,
+      "loss": 0.0979,
+      "step": 2496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019086350478344458,
+      "loss": 0.0998,
+      "step": 2497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019085467137344334,
+      "loss": 0.0983,
+      "step": 2498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019084583389992877,
+      "loss": 0.0949,
+      "step": 2499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019083699236329603,
+      "loss": 0.1039,
+      "step": 2500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019082814676394066,
+      "loss": 0.098,
+      "step": 2501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019081929710225825,
+      "loss": 0.0959,
+      "step": 2502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001908104433786446,
+      "loss": 0.0998,
+      "step": 2503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019080158559349567,
+      "loss": 0.0917,
+      "step": 2504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019079272374720766,
+      "loss": 0.099,
+      "step": 2505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019078385784017688,
+      "loss": 0.0965,
+      "step": 2506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019077498787279992,
+      "loss": 0.0969,
+      "step": 2507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001907661138454735,
+      "loss": 0.1013,
+      "step": 2508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001907572357585944,
+      "loss": 0.0962,
+      "step": 2509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019074835361255983,
+      "loss": 0.1002,
+      "step": 2510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019073946740776697,
+      "loss": 0.0967,
+      "step": 2511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019073057714461334,
+      "loss": 0.097,
+      "step": 2512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019072168282349643,
+      "loss": 0.1023,
+      "step": 2513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019071278444481415,
+      "loss": 0.1048,
+      "step": 2514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019070388200896445,
+      "loss": 0.0973,
+      "step": 2515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001906949755163455,
+      "loss": 0.0982,
+      "step": 2516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019068606496735563,
+      "loss": 0.0943,
+      "step": 2517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019067715036239337,
+      "loss": 0.0985,
+      "step": 2518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019066823170185745,
+      "loss": 0.0992,
+      "step": 2519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019065930898614676,
+      "loss": 0.1035,
+      "step": 2520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019065038221566035,
+      "loss": 0.1041,
+      "step": 2521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019064145139079748,
+      "loss": 0.0925,
+      "step": 2522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019063251651195762,
+      "loss": 0.1007,
+      "step": 2523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019062357757954033,
+      "loss": 0.0986,
+      "step": 2524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019061463459394543,
+      "loss": 0.0992,
+      "step": 2525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001906056875555729,
+      "loss": 0.0952,
+      "step": 2526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019059673646482293,
+      "loss": 0.0961,
+      "step": 2527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001905877813220958,
+      "loss": 0.0978,
+      "step": 2528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001905788221277921,
+      "loss": 0.0956,
+      "step": 2529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019056985888231246,
+      "loss": 0.1002,
+      "step": 2530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019056089158605785,
+      "loss": 0.0975,
+      "step": 2531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019055192023942927,
+      "loss": 0.0953,
+      "step": 2532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000190542944842828,
+      "loss": 0.1006,
+      "step": 2533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019053396539665544,
+      "loss": 0.0973,
+      "step": 2534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019052498190131324,
+      "loss": 0.097,
+      "step": 2535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019051599435720315,
+      "loss": 0.093,
+      "step": 2536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019050700276472714,
+      "loss": 0.1007,
+      "step": 2537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019049800712428742,
+      "loss": 0.1011,
+      "step": 2538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019048900743628627,
+      "loss": 0.1,
+      "step": 2539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019048000370112622,
+      "loss": 0.0954,
+      "step": 2540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019047099591920996,
+      "loss": 0.0894,
+      "step": 2541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001904619840909404,
+      "loss": 0.0947,
+      "step": 2542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019045296821672053,
+      "loss": 0.0964,
+      "step": 2543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019044394829695365,
+      "loss": 0.0977,
+      "step": 2544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019043492433204318,
+      "loss": 0.0945,
+      "step": 2545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019042589632239264,
+      "loss": 0.0984,
+      "step": 2546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019041686426840592,
+      "loss": 0.0976,
+      "step": 2547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019040782817048694,
+      "loss": 0.1046,
+      "step": 2548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019039878802903982,
+      "loss": 0.0949,
+      "step": 2549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001903897438444689,
+      "loss": 0.0957,
+      "step": 2550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019038069561717872,
+      "loss": 0.0965,
+      "step": 2551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019037164334757386,
+      "loss": 0.0972,
+      "step": 2552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019036258703605932,
+      "loss": 0.1067,
+      "step": 2553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019035352668304007,
+      "loss": 0.1033,
+      "step": 2554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019034446228892134,
+      "loss": 0.0986,
+      "step": 2555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019033539385410853,
+      "loss": 0.0941,
+      "step": 2556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019032632137900733,
+      "loss": 0.0979,
+      "step": 2557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019031724486402336,
+      "loss": 0.1031,
+      "step": 2558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019030816430956268,
+      "loss": 0.0971,
+      "step": 2559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019029907971603139,
+      "loss": 0.1013,
+      "step": 2560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019028999108383578,
+      "loss": 0.0906,
+      "step": 2561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019028089841338237,
+      "loss": 0.0993,
+      "step": 2562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001902718017050778,
+      "loss": 0.1031,
+      "step": 2563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019026270095932895,
+      "loss": 0.0915,
+      "step": 2564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001902535961765429,
+      "loss": 0.0986,
+      "step": 2565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001902444873571268,
+      "loss": 0.0954,
+      "step": 2566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019023537450148803,
+      "loss": 0.1011,
+      "step": 2567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019022625761003423,
+      "loss": 0.0985,
+      "step": 2568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019021713668317312,
+      "loss": 0.0951,
+      "step": 2569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019020801172131262,
+      "loss": 0.0959,
+      "step": 2570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001901988827248609,
+      "loss": 0.0981,
+      "step": 2571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019018974969422625,
+      "loss": 0.0965,
+      "step": 2572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019018061262981707,
+      "loss": 0.0992,
+      "step": 2573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019017147153204214,
+      "loss": 0.1003,
+      "step": 2574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001901623264013102,
+      "loss": 0.0983,
+      "step": 2575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019015317723803033,
+      "loss": 0.0987,
+      "step": 2576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019014402404261168,
+      "loss": 0.0945,
+      "step": 2577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019013486681546365,
+      "loss": 0.0961,
+      "step": 2578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019012570555699583,
+      "loss": 0.0992,
+      "step": 2579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019011654026761794,
+      "loss": 0.0942,
+      "step": 2580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001901073709477399,
+      "loss": 0.0948,
+      "step": 2581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001900981975977718,
+      "loss": 0.0881,
+      "step": 2582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019008902021812397,
+      "loss": 0.1039,
+      "step": 2583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019007983880920678,
+      "loss": 0.105,
+      "step": 2584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019007065337143096,
+      "loss": 0.0989,
+      "step": 2585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019006146390520732,
+      "loss": 0.1008,
+      "step": 2586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019005227041094685,
+      "loss": 0.1008,
+      "step": 2587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019004307288906068,
+      "loss": 0.0936,
+      "step": 2588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019003387133996027,
+      "loss": 0.1002,
+      "step": 2589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001900246657640571,
+      "loss": 0.0931,
+      "step": 2590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019001545616176285,
+      "loss": 0.1042,
+      "step": 2591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019000624253348954,
+      "loss": 0.096,
+      "step": 2592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018999702487964922,
+      "loss": 0.0975,
+      "step": 2593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018998780320065407,
+      "loss": 0.1024,
+      "step": 2594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001899785774969166,
+      "loss": 0.0973,
+      "step": 2595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018996934776884943,
+      "loss": 0.0958,
+      "step": 2596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018996011401686537,
+      "loss": 0.0903,
+      "step": 2597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001899508762413774,
+      "loss": 0.0966,
+      "step": 2598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018994163444279862,
+      "loss": 0.0937,
+      "step": 2599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018993238862154247,
+      "loss": 0.0988,
+      "step": 2600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018992313877802243,
+      "loss": 0.0988,
+      "step": 2601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001899138849126522,
+      "loss": 0.0922,
+      "step": 2602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018990462702584568,
+      "loss": 0.0989,
+      "step": 2603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018989536511801689,
+      "loss": 0.0923,
+      "step": 2604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001898860991895801,
+      "loss": 0.1016,
+      "step": 2605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018987682924094976,
+      "loss": 0.1063,
+      "step": 2606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018986755527254046,
+      "loss": 0.098,
+      "step": 2607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018985827728476692,
+      "loss": 0.0964,
+      "step": 2608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001898489952780442,
+      "loss": 0.0893,
+      "step": 2609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018983970925278738,
+      "loss": 0.0965,
+      "step": 2610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018983041920941178,
+      "loss": 0.0955,
+      "step": 2611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018982112514833292,
+      "loss": 0.0971,
+      "step": 2612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018981182706996648,
+      "loss": 0.1016,
+      "step": 2613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018980252497472836,
+      "loss": 0.0975,
+      "step": 2614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001897932188630345,
+      "loss": 0.1046,
+      "step": 2615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018978390873530118,
+      "loss": 0.0901,
+      "step": 2616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018977459459194483,
+      "loss": 0.0915,
+      "step": 2617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018976527643338196,
+      "loss": 0.1025,
+      "step": 2618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018975595426002937,
+      "loss": 0.0928,
+      "step": 2619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000189746628072304,
+      "loss": 0.0954,
+      "step": 2620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018973729787062296,
+      "loss": 0.0963,
+      "step": 2621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018972796365540354,
+      "loss": 0.0981,
+      "step": 2622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018971862542706324,
+      "loss": 0.0981,
+      "step": 2623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018970928318601967,
+      "loss": 0.0948,
+      "step": 2624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018969993693269073,
+      "loss": 0.1009,
+      "step": 2625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018969058666749438,
+      "loss": 0.1011,
+      "step": 2626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018968123239084884,
+      "loss": 0.0974,
+      "step": 2627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018967187410317247,
+      "loss": 0.0905,
+      "step": 2628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018966251180488385,
+      "loss": 0.0976,
+      "step": 2629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018965314549640168,
+      "loss": 0.0975,
+      "step": 2630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001896437751781449,
+      "loss": 0.0975,
+      "step": 2631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018963440085053258,
+      "loss": 0.0992,
+      "step": 2632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000189625022513984,
+      "loss": 0.1017,
+      "step": 2633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001896156401689186,
+      "loss": 0.0977,
+      "step": 2634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018960625381575605,
+      "loss": 0.0988,
+      "step": 2635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001895968634549161,
+      "loss": 0.098,
+      "step": 2636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018958746908681877,
+      "loss": 0.099,
+      "step": 2637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018957807071188422,
+      "loss": 0.0982,
+      "step": 2638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001895686683305328,
+      "loss": 0.0963,
+      "step": 2639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018955926194318503,
+      "loss": 0.098,
+      "step": 2640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001895498515502616,
+      "loss": 0.1036,
+      "step": 2641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001895404371521835,
+      "loss": 0.098,
+      "step": 2642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001895310187493716,
+      "loss": 0.098,
+      "step": 2643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001895215963422473,
+      "loss": 0.1013,
+      "step": 2644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018951216993123193,
+      "loss": 0.0974,
+      "step": 2645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018950273951674719,
+      "loss": 0.0946,
+      "step": 2646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018949330509921474,
+      "loss": 0.0937,
+      "step": 2647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018948386667905662,
+      "loss": 0.0981,
+      "step": 2648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018947442425669496,
+      "loss": 0.0991,
+      "step": 2649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018946497783255204,
+      "loss": 0.0999,
+      "step": 2650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018945552740705037,
+      "loss": 0.0948,
+      "step": 2651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018944607298061264,
+      "loss": 0.0985,
+      "step": 2652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001894366145536617,
+      "loss": 0.1029,
+      "step": 2653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018942715212662053,
+      "loss": 0.0958,
+      "step": 2654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018941768569991242,
+      "loss": 0.0946,
+      "step": 2655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018940821527396073,
+      "loss": 0.0973,
+      "step": 2656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018939874084918903,
+      "loss": 0.0911,
+      "step": 2657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018938926242602103,
+      "loss": 0.0997,
+      "step": 2658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001893797800048807,
+      "loss": 0.0988,
+      "step": 2659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018937029358619215,
+      "loss": 0.0982,
+      "step": 2660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018936080317037964,
+      "loss": 0.0913,
+      "step": 2661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018935130875786765,
+      "loss": 0.0941,
+      "step": 2662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001893418103490808,
+      "loss": 0.0997,
+      "step": 2663
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018933230794444393,
+      "loss": 0.094,
+      "step": 2664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018932280154438202,
+      "loss": 0.0931,
+      "step": 2665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018931329114932024,
+      "loss": 0.1007,
+      "step": 2666
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000189303776759684,
+      "loss": 0.0932,
+      "step": 2667
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001892942583758988,
+      "loss": 0.0922,
+      "step": 2668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018928473599839034,
+      "loss": 0.097,
+      "step": 2669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018927520962758453,
+      "loss": 0.0947,
+      "step": 2670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018926567926390745,
+      "loss": 0.1004,
+      "step": 2671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018925614490778532,
+      "loss": 0.0966,
+      "step": 2672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001892466065596446,
+      "loss": 0.1,
+      "step": 2673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018923706421991186,
+      "loss": 0.1018,
+      "step": 2674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001892275178890139,
+      "loss": 0.0901,
+      "step": 2675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018921796756737774,
+      "loss": 0.0962,
+      "step": 2676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018920841325543043,
+      "loss": 0.0943,
+      "step": 2677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018919885495359936,
+      "loss": 0.1001,
+      "step": 2678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000189189292662312,
+      "loss": 0.0948,
+      "step": 2679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018917972638199601,
+      "loss": 0.1008,
+      "step": 2680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018917015611307926,
+      "loss": 0.0966,
+      "step": 2681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018916058185598983,
+      "loss": 0.0972,
+      "step": 2682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018915100361115584,
+      "loss": 0.0992,
+      "step": 2683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018914142137900578,
+      "loss": 0.1003,
+      "step": 2684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018913183515996816,
+      "loss": 0.0992,
+      "step": 2685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018912224495447176,
+      "loss": 0.092,
+      "step": 2686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018911265076294544,
+      "loss": 0.0927,
+      "step": 2687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018910305258581839,
+      "loss": 0.0967,
+      "step": 2688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001890934504235198,
+      "loss": 0.0978,
+      "step": 2689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018908384427647926,
+      "loss": 0.0938,
+      "step": 2690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001890742341451263,
+      "loss": 0.0977,
+      "step": 2691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018906462002989076,
+      "loss": 0.0959,
+      "step": 2692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018905500193120266,
+      "loss": 0.0983,
+      "step": 2693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018904537984949216,
+      "loss": 0.0943,
+      "step": 2694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001890357537851896,
+      "loss": 0.0937,
+      "step": 2695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001890261237387255,
+      "loss": 0.0977,
+      "step": 2696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018901648971053067,
+      "loss": 0.0974,
+      "step": 2697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018900685170103581,
+      "loss": 0.0973,
+      "step": 2698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018899720971067216,
+      "loss": 0.1024,
+      "step": 2699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018898756373987086,
+      "loss": 0.0992,
+      "step": 2700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018897791378906335,
+      "loss": 0.1011,
+      "step": 2701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018896825985868128,
+      "loss": 0.0974,
+      "step": 2702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018895860194915635,
+      "loss": 0.102,
+      "step": 2703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018894894006092055,
+      "loss": 0.0913,
+      "step": 2704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018893927419440606,
+      "loss": 0.1016,
+      "step": 2705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018892960435004509,
+      "loss": 0.1005,
+      "step": 2706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001889199305282702,
+      "loss": 0.093,
+      "step": 2707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018891025272951403,
+      "loss": 0.0951,
+      "step": 2708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018890057095420943,
+      "loss": 0.106,
+      "step": 2709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018889088520278942,
+      "loss": 0.0962,
+      "step": 2710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001888811954756872,
+      "loss": 0.0955,
+      "step": 2711
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018887150177333614,
+      "loss": 0.0974,
+      "step": 2712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001888618040961698,
+      "loss": 0.0981,
+      "step": 2713
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018885210244462192,
+      "loss": 0.0964,
+      "step": 2714
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001888423968191264,
+      "loss": 0.0975,
+      "step": 2715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018883268722011734,
+      "loss": 0.0999,
+      "step": 2716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018882297364802898,
+      "loss": 0.0946,
+      "step": 2717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018881325610329582,
+      "loss": 0.0957,
+      "step": 2718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001888035345863524,
+      "loss": 0.1,
+      "step": 2719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018879380909763357,
+      "loss": 0.0931,
+      "step": 2720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001887840796375743,
+      "loss": 0.0978,
+      "step": 2721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001887743462066098,
+      "loss": 0.0981,
+      "step": 2722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018876460880517528,
+      "loss": 0.0934,
+      "step": 2723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018875486743370634,
+      "loss": 0.0904,
+      "step": 2724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001887451220926386,
+      "loss": 0.0973,
+      "step": 2725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018873537278240804,
+      "loss": 0.1017,
+      "step": 2726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018872561950345057,
+      "loss": 0.0962,
+      "step": 2727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018871586225620244,
+      "loss": 0.0977,
+      "step": 2728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018870610104110015,
+      "loss": 0.0985,
+      "step": 2729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018869633585858018,
+      "loss": 0.0951,
+      "step": 2730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018868656670907927,
+      "loss": 0.1006,
+      "step": 2731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001886767935930344,
+      "loss": 0.0966,
+      "step": 2732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018866701651088264,
+      "loss": 0.0969,
+      "step": 2733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001886572354630613,
+      "loss": 0.0972,
+      "step": 2734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018864745045000784,
+      "loss": 0.1009,
+      "step": 2735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001886376614721599,
+      "loss": 0.0933,
+      "step": 2736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018862786852995525,
+      "loss": 0.089,
+      "step": 2737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018861807162383194,
+      "loss": 0.0944,
+      "step": 2738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018860827075422814,
+      "loss": 0.0989,
+      "step": 2739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018859846592158216,
+      "loss": 0.0981,
+      "step": 2740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018858865712633255,
+      "loss": 0.1009,
+      "step": 2741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000188578844368918,
+      "loss": 0.0956,
+      "step": 2742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001885690276497774,
+      "loss": 0.0929,
+      "step": 2743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018855920696934982,
+      "loss": 0.0949,
+      "step": 2744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018854938232807443,
+      "loss": 0.1006,
+      "step": 2745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001885395537263907,
+      "loss": 0.0963,
+      "step": 2746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018852972116473825,
+      "loss": 0.0964,
+      "step": 2747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001885198846435568,
+      "loss": 0.096,
+      "step": 2748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018851004416328625,
+      "loss": 0.0931,
+      "step": 2749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018850019972436682,
+      "loss": 0.0995,
+      "step": 2750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018849035132723868,
+      "loss": 0.0939,
+      "step": 2751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018848049897234246,
+      "loss": 0.1005,
+      "step": 2752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018847064266011863,
+      "loss": 0.0901,
+      "step": 2753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001884607823910082,
+      "loss": 0.0975,
+      "step": 2754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018845091816545203,
+      "loss": 0.097,
+      "step": 2755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001884410499838914,
+      "loss": 0.0942,
+      "step": 2756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001884311778467676,
+      "loss": 0.0998,
+      "step": 2757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018842130175452223,
+      "loss": 0.096,
+      "step": 2758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018841142170759694,
+      "loss": 0.0935,
+      "step": 2759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018840153770643365,
+      "loss": 0.098,
+      "step": 2760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001883916497514744,
+      "loss": 0.105,
+      "step": 2761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001883817578431615,
+      "loss": 0.0979,
+      "step": 2762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018837186198193727,
+      "loss": 0.0939,
+      "step": 2763
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001883619621682444,
+      "loss": 0.0951,
+      "step": 2764
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001883520584025256,
+      "loss": 0.1002,
+      "step": 2765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018834215068522388,
+      "loss": 0.0946,
+      "step": 2766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001883322390167823,
+      "loss": 0.097,
+      "step": 2767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018832232339764418,
+      "loss": 0.0918,
+      "step": 2768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018831240382825306,
+      "loss": 0.0921,
+      "step": 2769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018830248030905254,
+      "loss": 0.0952,
+      "step": 2770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018829255284048645,
+      "loss": 0.096,
+      "step": 2771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018828262142299882,
+      "loss": 0.0863,
+      "step": 2772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018827268605703384,
+      "loss": 0.0988,
+      "step": 2773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018826274674303586,
+      "loss": 0.0968,
+      "step": 2774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018825280348144942,
+      "loss": 0.1001,
+      "step": 2775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018824285627271924,
+      "loss": 0.0961,
+      "step": 2776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018823290511729024,
+      "loss": 0.0928,
+      "step": 2777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018822295001560743,
+      "loss": 0.0987,
+      "step": 2778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001882129909681161,
+      "loss": 0.0952,
+      "step": 2779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001882030279752617,
+      "loss": 0.097,
+      "step": 2780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018819306103748975,
+      "loss": 0.0945,
+      "step": 2781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018818309015524609,
+      "loss": 0.0956,
+      "step": 2782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018817311532897665,
+      "loss": 0.0962,
+      "step": 2783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018816313655912757,
+      "loss": 0.0955,
+      "step": 2784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001881531538461451,
+      "loss": 0.0977,
+      "step": 2785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018814316719047578,
+      "loss": 0.0948,
+      "step": 2786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018813317659256628,
+      "loss": 0.0901,
+      "step": 2787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018812318205286337,
+      "loss": 0.0982,
+      "step": 2788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018811318357181414,
+      "loss": 0.0951,
+      "step": 2789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018810318114986568,
+      "loss": 0.0993,
+      "step": 2790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018809317478746542,
+      "loss": 0.0981,
+      "step": 2791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001880831644850609,
+      "loss": 0.093,
+      "step": 2792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001880731502430998,
+      "loss": 0.1003,
+      "step": 2793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018806313206203004,
+      "loss": 0.0992,
+      "step": 2794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018805310994229969,
+      "loss": 0.09,
+      "step": 2795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018804308388435697,
+      "loss": 0.0955,
+      "step": 2796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001880330538886503,
+      "loss": 0.0934,
+      "step": 2797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018802301995562827,
+      "loss": 0.0997,
+      "step": 2798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001880129820857397,
+      "loss": 0.0986,
+      "step": 2799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001880029402794335,
+      "loss": 0.0979,
+      "step": 2800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018799289453715878,
+      "loss": 0.0974,
+      "step": 2801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018798284485936487,
+      "loss": 0.0977,
+      "step": 2802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018797279124650124,
+      "loss": 0.0937,
+      "step": 2803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018796273369901752,
+      "loss": 0.0977,
+      "step": 2804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018795267221736358,
+      "loss": 0.099,
+      "step": 2805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018794260680198938,
+      "loss": 0.089,
+      "step": 2806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018793253745334512,
+      "loss": 0.0955,
+      "step": 2807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018792246417188117,
+      "loss": 0.0923,
+      "step": 2808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018791238695804806,
+      "loss": 0.0939,
+      "step": 2809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018790230581229643,
+      "loss": 0.0915,
+      "step": 2810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018789222073507724,
+      "loss": 0.0924,
+      "step": 2811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018788213172684154,
+      "loss": 0.0913,
+      "step": 2812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018787203878804055,
+      "loss": 0.0957,
+      "step": 2813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001878619419191257,
+      "loss": 0.0941,
+      "step": 2814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018785184112054854,
+      "loss": 0.0921,
+      "step": 2815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018784173639276085,
+      "loss": 0.0955,
+      "step": 2816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018783162773621458,
+      "loss": 0.0998,
+      "step": 2817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018782151515136182,
+      "loss": 0.0906,
+      "step": 2818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018781139863865488,
+      "loss": 0.0967,
+      "step": 2819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018780127819854626,
+      "loss": 0.096,
+      "step": 2820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018779115383148854,
+      "loss": 0.1034,
+      "step": 2821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018778102553793452,
+      "loss": 0.0957,
+      "step": 2822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018777089331833727,
+      "loss": 0.0849,
+      "step": 2823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001877607571731499,
+      "loss": 0.0941,
+      "step": 2824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018775061710282575,
+      "loss": 0.1029,
+      "step": 2825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018774047310781837,
+      "loss": 0.0974,
+      "step": 2826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018773032518858146,
+      "loss": 0.0998,
+      "step": 2827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018772017334556885,
+      "loss": 0.0959,
+      "step": 2828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018771001757923465,
+      "loss": 0.0966,
+      "step": 2829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018769985789003299,
+      "loss": 0.0982,
+      "step": 2830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018768969427841836,
+      "loss": 0.1001,
+      "step": 2831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018767952674484528,
+      "loss": 0.0981,
+      "step": 2832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001876693552897685,
+      "loss": 0.0886,
+      "step": 2833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018765917991364294,
+      "loss": 0.0936,
+      "step": 2834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018764900061692369,
+      "loss": 0.0908,
+      "step": 2835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018763881740006607,
+      "loss": 0.0963,
+      "step": 2836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001876286302635255,
+      "loss": 0.1037,
+      "step": 2837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018761843920775756,
+      "loss": 0.097,
+      "step": 2838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018760824423321815,
+      "loss": 0.0965,
+      "step": 2839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018759804534036314,
+      "loss": 0.0995,
+      "step": 2840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018758784252964874,
+      "loss": 0.0964,
+      "step": 2841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018757763580153127,
+      "loss": 0.1034,
+      "step": 2842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001875674251564672,
+      "loss": 0.1009,
+      "step": 2843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018755721059491326,
+      "loss": 0.1051,
+      "step": 2844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018754699211732622,
+      "loss": 0.0982,
+      "step": 2845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018753676972416318,
+      "loss": 0.0959,
+      "step": 2846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001875265434158813,
+      "loss": 0.0998,
+      "step": 2847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000187516313192938,
+      "loss": 0.0961,
+      "step": 2848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018750607905579077,
+      "loss": 0.1002,
+      "step": 2849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018749584100489738,
+      "loss": 0.0953,
+      "step": 2850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001874855990407157,
+      "loss": 0.0956,
+      "step": 2851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018747535316370386,
+      "loss": 0.0981,
+      "step": 2852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018746510337432003,
+      "loss": 0.0997,
+      "step": 2853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001874548496730227,
+      "loss": 0.0945,
+      "step": 2854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018744459206027048,
+      "loss": 0.0898,
+      "step": 2855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018743433053652208,
+      "loss": 0.0957,
+      "step": 2856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001874240651022365,
+      "loss": 0.097,
+      "step": 2857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018741379575787286,
+      "loss": 0.0968,
+      "step": 2858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018740352250389046,
+      "loss": 0.0927,
+      "step": 2859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018739324534074875,
+      "loss": 0.0943,
+      "step": 2860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018738296426890741,
+      "loss": 0.1013,
+      "step": 2861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018737267928882627,
+      "loss": 0.0989,
+      "step": 2862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018736239040096533,
+      "loss": 0.1004,
+      "step": 2863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001873520976057847,
+      "loss": 0.0956,
+      "step": 2864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018734180090374482,
+      "loss": 0.0995,
+      "step": 2865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018733150029530619,
+      "loss": 0.0944,
+      "step": 2866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018732119578092945,
+      "loss": 0.1009,
+      "step": 2867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018731088736107554,
+      "loss": 0.1001,
+      "step": 2868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018730057503620547,
+      "loss": 0.0988,
+      "step": 2869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001872902588067805,
+      "loss": 0.0956,
+      "step": 2870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018727993867326202,
+      "loss": 0.0965,
+      "step": 2871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018726961463611156,
+      "loss": 0.097,
+      "step": 2872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018725928669579095,
+      "loss": 0.1008,
+      "step": 2873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000187248954852762,
+      "loss": 0.0971,
+      "step": 2874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018723861910748687,
+      "loss": 0.0907,
+      "step": 2875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018722827946042784,
+      "loss": 0.0991,
+      "step": 2876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018721793591204735,
+      "loss": 0.1002,
+      "step": 2877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000187207588462808,
+      "loss": 0.0979,
+      "step": 2878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018719723711317258,
+      "loss": 0.0961,
+      "step": 2879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018718688186360408,
+      "loss": 0.0961,
+      "step": 2880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018717652271456564,
+      "loss": 0.0969,
+      "step": 2881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018716615966652057,
+      "loss": 0.0995,
+      "step": 2882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001871557927199324,
+      "loss": 0.0976,
+      "step": 2883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018714542187526473,
+      "loss": 0.0959,
+      "step": 2884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001871350471329814,
+      "loss": 0.0985,
+      "step": 2885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018712466849354645,
+      "loss": 0.0981,
+      "step": 2886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018711428595742413,
+      "loss": 0.0983,
+      "step": 2887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001871038995250787,
+      "loss": 0.0874,
+      "step": 2888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018709350919697478,
+      "loss": 0.096,
+      "step": 2889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018708311497357704,
+      "loss": 0.1043,
+      "step": 2890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018707271685535038,
+      "loss": 0.0921,
+      "step": 2891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018706231484275982,
+      "loss": 0.0932,
+      "step": 2892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018705190893627066,
+      "loss": 0.099,
+      "step": 2893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018704149913634826,
+      "loss": 0.1008,
+      "step": 2894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001870310854434582,
+      "loss": 0.0991,
+      "step": 2895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001870206678580663,
+      "loss": 0.0975,
+      "step": 2896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018701024638063843,
+      "loss": 0.0889,
+      "step": 2897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018699982101164067,
+      "loss": 0.098,
+      "step": 2898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018698939175153937,
+      "loss": 0.0928,
+      "step": 2899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018697895860080098,
+      "loss": 0.102,
+      "step": 2900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018696852155989207,
+      "loss": 0.0956,
+      "step": 2901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018695808062927947,
+      "loss": 0.0941,
+      "step": 2902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018694763580943017,
+      "loss": 0.0932,
+      "step": 2903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001869371871008113,
+      "loss": 0.0937,
+      "step": 2904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018692673450389018,
+      "loss": 0.098,
+      "step": 2905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018691627801913432,
+      "loss": 0.0974,
+      "step": 2906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001869058176470114,
+      "loss": 0.0982,
+      "step": 2907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018689535338798922,
+      "loss": 0.0946,
+      "step": 2908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018688488524253583,
+      "loss": 0.0958,
+      "step": 2909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018687441321111944,
+      "loss": 0.0926,
+      "step": 2910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001868639372942084,
+      "loss": 0.0922,
+      "step": 2911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018685345749227126,
+      "loss": 0.0986,
+      "step": 2912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018684297380577668,
+      "loss": 0.112,
+      "step": 2913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001868324862351936,
+      "loss": 0.6594,
+      "step": 2914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018682199478099108,
+      "loss": 0.476,
+      "step": 2915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018681149944363833,
+      "loss": 0.3005,
+      "step": 2916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018680100022360477,
+      "loss": 5.1748,
+      "step": 2917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018679049712136002,
+      "loss": 1.3053,
+      "step": 2918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018677999013737377,
+      "loss": 1.6842,
+      "step": 2919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018676947927211596,
+      "loss": 2.1709,
+      "step": 2920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018675896452605674,
+      "loss": 1.1054,
+      "step": 2921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018674844589966635,
+      "loss": 5.0822,
+      "step": 2922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018673792339341525,
+      "loss": 1.8414,
+      "step": 2923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018672739700777407,
+      "loss": 1.2124,
+      "step": 2924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018671686674321357,
+      "loss": 1.0425,
+      "step": 2925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018670633260020475,
+      "loss": 0.4248,
+      "step": 2926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001866957945792188,
+      "loss": 3.1703,
+      "step": 2927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018668525268072697,
+      "loss": 7.4973,
+      "step": 2928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018667470690520075,
+      "loss": 10.0221,
+      "step": 2929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001866641572531118,
+      "loss": 8.1721,
+      "step": 2930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018665360372493203,
+      "loss": 8.445,
+      "step": 2931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001866430463211334,
+      "loss": 7.8867,
+      "step": 2932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018663248504218808,
+      "loss": 11.6952,
+      "step": 2933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018662191988856846,
+      "loss": 14.0982,
+      "step": 2934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018661135086074702,
+      "loss": 17.478,
+      "step": 2935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018660077795919656,
+      "loss": 11.3046,
+      "step": 2936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018659020118438985,
+      "loss": 8.7956,
+      "step": 2937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018657962053680004,
+      "loss": 9.9289,
+      "step": 2938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018656903601690025,
+      "loss": 8.7743,
+      "step": 2939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018655844762516396,
+      "loss": 7.7659,
+      "step": 2940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018654785536206472,
+      "loss": 8.0183,
+      "step": 2941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018653725922807623,
+      "loss": 7.2531,
+      "step": 2942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018652665922367249,
+      "loss": 6.6037,
+      "step": 2943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018651605534932749,
+      "loss": 6.1851,
+      "step": 2944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018650544760551556,
+      "loss": 5.6667,
+      "step": 2945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018649483599271109,
+      "loss": 5.4759,
+      "step": 2946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018648422051138877,
+      "loss": 5.1838,
+      "step": 2947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001864736011620233,
+      "loss": 4.8389,
+      "step": 2948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018646297794508966,
+      "loss": 4.9,
+      "step": 2949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018645235086106298,
+      "loss": 4.7752,
+      "step": 2950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018644171991041858,
+      "loss": 4.8028,
+      "step": 2951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018643108509363193,
+      "loss": 4.693,
+      "step": 2952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018642044641117863,
+      "loss": 4.4042,
+      "step": 2953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018640980386353456,
+      "loss": 4.3765,
+      "step": 2954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018639915745117568,
+      "loss": 4.2516,
+      "step": 2955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001863885071745781,
+      "loss": 4.27,
+      "step": 2956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018637785303421832,
+      "loss": 4.2193,
+      "step": 2957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018636719503057266,
+      "loss": 4.1258,
+      "step": 2958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018635653316411793,
+      "loss": 4.1511,
+      "step": 2959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018634586743533098,
+      "loss": 4.1821,
+      "step": 2960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018633519784468878,
+      "loss": 4.0832,
+      "step": 2961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018632452439266856,
+      "loss": 4.0804,
+      "step": 2962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018631384707974767,
+      "loss": 4.1071,
+      "step": 2963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018630316590640374,
+      "loss": 4.0742,
+      "step": 2964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001862924808731144,
+      "loss": 3.9749,
+      "step": 2965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018628179198035757,
+      "loss": 3.9505,
+      "step": 2966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018627109922861134,
+      "loss": 3.9644,
+      "step": 2967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001862604026183539,
+      "loss": 3.905,
+      "step": 2968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018624970215006372,
+      "loss": 3.8946,
+      "step": 2969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018623899782421933,
+      "loss": 3.8898,
+      "step": 2970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001862282896412995,
+      "loss": 3.8961,
+      "step": 2971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001862175776017832,
+      "loss": 3.8894,
+      "step": 2972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018620686170614948,
+      "loss": 3.8999,
+      "step": 2973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001861961419548776,
+      "loss": 3.8782,
+      "step": 2974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018618541834844704,
+      "loss": 3.8327,
+      "step": 2975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018617469088733746,
+      "loss": 3.8628,
+      "step": 2976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018616395957202854,
+      "loss": 3.8366,
+      "step": 2977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018615322440300035,
+      "loss": 3.8166,
+      "step": 2978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018614248538073296,
+      "loss": 3.7967,
+      "step": 2979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001861317425057067,
+      "loss": 3.7968,
+      "step": 2980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018612099577840202,
+      "loss": 3.7907,
+      "step": 2981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018611024519929966,
+      "loss": 3.8036,
+      "step": 2982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018609949076888032,
+      "loss": 3.7574,
+      "step": 2983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018608873248762508,
+      "loss": 3.7836,
+      "step": 2984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018607797035601512,
+      "loss": 3.797,
+      "step": 2985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018606720437453172,
+      "loss": 3.773,
+      "step": 2986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001860564345436564,
+      "loss": 3.7461,
+      "step": 2987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001860456608638709,
+      "loss": 3.7415,
+      "step": 2988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018603488333565702,
+      "loss": 3.7511,
+      "step": 2989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018602410195949683,
+      "loss": 3.7442,
+      "step": 2990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001860133167358725,
+      "loss": 3.7312,
+      "step": 2991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018600252766526643,
+      "loss": 3.7376,
+      "step": 2992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018599173474816117,
+      "loss": 3.7116,
+      "step": 2993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001859809379850394,
+      "loss": 3.7365,
+      "step": 2994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018597013737638401,
+      "loss": 3.7131,
+      "step": 2995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018595933292267814,
+      "loss": 3.6964,
+      "step": 2996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018594852462440494,
+      "loss": 3.7013,
+      "step": 2997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018593771248204784,
+      "loss": 3.7016,
+      "step": 2998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001859268964960904,
+      "loss": 3.67,
+      "step": 2999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018591607666701643,
+      "loss": 3.678,
+      "step": 3000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018590525299530978,
+      "loss": 3.6724,
+      "step": 3001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001858944254814546,
+      "loss": 3.6871,
+      "step": 3002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001858835941259351,
+      "loss": 3.6926,
+      "step": 3003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018587275892923577,
+      "loss": 3.6848,
+      "step": 3004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001858619198918412,
+      "loss": 3.6749,
+      "step": 3005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018585107701423614,
+      "loss": 3.6833,
+      "step": 3006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001858402302969056,
+      "loss": 3.6825,
+      "step": 3007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018582937974033465,
+      "loss": 3.6765,
+      "step": 3008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001858185253450086,
+      "loss": 3.6706,
+      "step": 3009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018580766711141295,
+      "loss": 3.6646,
+      "step": 3010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001857968050400333,
+      "loss": 3.6599,
+      "step": 3011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001857859391313555,
+      "loss": 3.657,
+      "step": 3012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001857750693858655,
+      "loss": 3.6568,
+      "step": 3013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018576419580404943,
+      "loss": 3.6665,
+      "step": 3014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018575331838639367,
+      "loss": 3.6685,
+      "step": 3015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001857424371333847,
+      "loss": 3.663,
+      "step": 3016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018573155204550915,
+      "loss": 3.6604,
+      "step": 3017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001857206631232539,
+      "loss": 3.6706,
+      "step": 3018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018570977036710598,
+      "loss": 3.6873,
+      "step": 3019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018569887377755255,
+      "loss": 3.6621,
+      "step": 3020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018568797335508094,
+      "loss": 3.6778,
+      "step": 3021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001856770691001787,
+      "loss": 3.6469,
+      "step": 3022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018566616101333355,
+      "loss": 3.6555,
+      "step": 3023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018565524909503332,
+      "loss": 3.6695,
+      "step": 3024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018564433334576606,
+      "loss": 3.6679,
+      "step": 3025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018563341376601997,
+      "loss": 3.6684,
+      "step": 3026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018562249035628347,
+      "loss": 3.6531,
+      "step": 3027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001856115631170451,
+      "loss": 3.6649,
+      "step": 3028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018560063204879358,
+      "loss": 3.6707,
+      "step": 3029
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018558969715201778,
+      "loss": 3.6484,
+      "step": 3030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001855787584272068,
+      "loss": 3.663,
+      "step": 3031
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018556781587484986,
+      "loss": 3.6747,
+      "step": 3032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001855568694954364,
+      "loss": 3.6578,
+      "step": 3033
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018554591928945596,
+      "loss": 3.6628,
+      "step": 3034
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018553496525739837,
+      "loss": 3.6522,
+      "step": 3035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018552400739975345,
+      "loss": 3.6825,
+      "step": 3036
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001855130457170114,
+      "loss": 3.6704,
+      "step": 3037
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018550208020966236,
+      "loss": 3.6579,
+      "step": 3038
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018549111087819687,
+      "loss": 3.6485,
+      "step": 3039
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001854801377231055,
+      "loss": 3.6446,
+      "step": 3040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018546916074487903,
+      "loss": 3.6659,
+      "step": 3041
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018545817994400842,
+      "loss": 3.6662,
+      "step": 3042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018544719532098477,
+      "loss": 3.6775,
+      "step": 3043
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001854362068762994,
+      "loss": 3.6444,
+      "step": 3044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018542521461044378,
+      "loss": 3.6612,
+      "step": 3045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001854142185239095,
+      "loss": 3.6465,
+      "step": 3046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001854032186171884,
+      "loss": 3.6736,
+      "step": 3047
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018539221489077243,
+      "loss": 3.6712,
+      "step": 3048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018538120734515378,
+      "loss": 3.6524,
+      "step": 3049
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001853701959808247,
+      "loss": 3.6432,
+      "step": 3050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018535918079827773,
+      "loss": 3.6469,
+      "step": 3051
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018534816179800552,
+      "loss": 3.653,
+      "step": 3052
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018533713898050087,
+      "loss": 3.6531,
+      "step": 3053
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018532611234625682,
+      "loss": 3.6575,
+      "step": 3054
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001853150818957665,
+      "loss": 3.6657,
+      "step": 3055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001853040476295233,
+      "loss": 3.6748,
+      "step": 3056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018529300954802072,
+      "loss": 3.6664,
+      "step": 3057
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001852819676517524,
+      "loss": 3.647,
+      "step": 3058
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018527092194121228,
+      "loss": 3.6812,
+      "step": 3059
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018525987241689427,
+      "loss": 3.6748,
+      "step": 3060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018524881907929262,
+      "loss": 3.663,
+      "step": 3061
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018523776192890171,
+      "loss": 3.6463,
+      "step": 3062
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001852267009662161,
+      "loss": 3.6503,
+      "step": 3063
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018521563619173042,
+      "loss": 3.6436,
+      "step": 3064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001852045676059396,
+      "loss": 3.654,
+      "step": 3065
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018519349520933865,
+      "loss": 3.6785,
+      "step": 3066
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018518241900242284,
+      "loss": 3.6811,
+      "step": 3067
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001851713389856875,
+      "loss": 3.6377,
+      "step": 3068
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018516025515962826,
+      "loss": 3.6637,
+      "step": 3069
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018514916752474076,
+      "loss": 3.6463,
+      "step": 3070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000185138076081521,
+      "loss": 3.6506,
+      "step": 3071
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018512698083046495,
+      "loss": 3.6637,
+      "step": 3072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001851158817720689,
+      "loss": 3.6528,
+      "step": 3073
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018510477890682923,
+      "loss": 3.6419,
+      "step": 3074
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018509367223524262,
+      "loss": 3.6478,
+      "step": 3075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018508256175780568,
+      "loss": 3.6492,
+      "step": 3076
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018507144747501543,
+      "loss": 3.6404,
+      "step": 3077
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001850603293873689,
+      "loss": 3.6376,
+      "step": 3078
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001850492074953634,
+      "loss": 3.6695,
+      "step": 3079
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018503808179949632,
+      "loss": 3.6347,
+      "step": 3080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001850269523002653,
+      "loss": 3.625,
+      "step": 3081
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001850158189981681,
+      "loss": 3.6575,
+      "step": 3082
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018500468189370267,
+      "loss": 3.6361,
+      "step": 3083
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018499354098736704,
+      "loss": 3.6845,
+      "step": 3084
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001849823962796596,
+      "loss": 3.6537,
+      "step": 3085
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018497124777107877,
+      "loss": 3.6789,
+      "step": 3086
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018496009546212314,
+      "loss": 3.6392,
+      "step": 3087
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018494893935329154,
+      "loss": 3.6461,
+      "step": 3088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001849377794450829,
+      "loss": 3.6566,
+      "step": 3089
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018492661573799634,
+      "loss": 3.6466,
+      "step": 3090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001849154482325312,
+      "loss": 3.665,
+      "step": 3091
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018490427692918697,
+      "loss": 3.6433,
+      "step": 3092
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018489310182846324,
+      "loss": 3.6469,
+      "step": 3093
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018488192293085985,
+      "loss": 3.6245,
+      "step": 3094
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018487074023687677,
+      "loss": 3.658,
+      "step": 3095
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018485955374701417,
+      "loss": 3.6281,
+      "step": 3096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018484836346177233,
+      "loss": 3.6409,
+      "step": 3097
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018483716938165178,
+      "loss": 3.6441,
+      "step": 3098
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018482597150715316,
+      "loss": 3.6358,
+      "step": 3099
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018481476983877728,
+      "loss": 3.6247,
+      "step": 3100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018480356437702516,
+      "loss": 3.6647,
+      "step": 3101
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018479235512239802,
+      "loss": 3.6152,
+      "step": 3102
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001847811420753971,
+      "loss": 3.6713,
+      "step": 3103
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000184769925236524,
+      "loss": 3.6511,
+      "step": 3104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018475870460628034,
+      "loss": 3.6389,
+      "step": 3105
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018474748018516794,
+      "loss": 3.6501,
+      "step": 3106
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018473625197368892,
+      "loss": 3.6492,
+      "step": 3107
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018472501997234537,
+      "loss": 3.632,
+      "step": 3108
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001847137841816397,
+      "loss": 3.6422,
+      "step": 3109
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018470254460207442,
+      "loss": 3.653,
+      "step": 3110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018469130123415222,
+      "loss": 3.6391,
+      "step": 3111
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018468005407837597,
+      "loss": 3.6433,
+      "step": 3112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018466880313524868,
+      "loss": 3.62,
+      "step": 3113
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001846575484052736,
+      "loss": 3.6357,
+      "step": 3114
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018464628988895408,
+      "loss": 3.6498,
+      "step": 3115
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018463502758679363,
+      "loss": 3.6529,
+      "step": 3116
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018462376149929602,
+      "loss": 3.6524,
+      "step": 3117
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001846124916269651,
+      "loss": 3.6605,
+      "step": 3118
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001846012179703049,
+      "loss": 3.6431,
+      "step": 3119
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018458994052981967,
+      "loss": 3.6389,
+      "step": 3120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001845786593060138,
+      "loss": 3.6437,
+      "step": 3121
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001845673742993918,
+      "loss": 3.6403,
+      "step": 3122
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018455608551045847,
+      "loss": 3.6391,
+      "step": 3123
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018454479293971866,
+      "loss": 3.6272,
+      "step": 3124
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001845334965876774,
+      "loss": 3.6287,
+      "step": 3125
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018452219645484003,
+      "loss": 3.6526,
+      "step": 3126
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018451089254171185,
+      "loss": 3.6583,
+      "step": 3127
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018449958484879848,
+      "loss": 3.6256,
+      "step": 3128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018448827337660565,
+      "loss": 3.6517,
+      "step": 3129
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001844769581256393,
+      "loss": 3.6451,
+      "step": 3130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018446563909640547,
+      "loss": 3.6266,
+      "step": 3131
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001844543162894104,
+      "loss": 3.6107,
+      "step": 3132
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018444298970516054,
+      "loss": 3.6414,
+      "step": 3133
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018443165934416248,
+      "loss": 3.6577,
+      "step": 3134
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018442032520692294,
+      "loss": 3.6555,
+      "step": 3135
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018440898729394888,
+      "loss": 3.6412,
+      "step": 3136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018439764560574737,
+      "loss": 3.6469,
+      "step": 3137
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001843863001428257,
+      "loss": 3.6525,
+      "step": 3138
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018437495090569124,
+      "loss": 3.6555,
+      "step": 3139
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018436359789485168,
+      "loss": 3.636,
+      "step": 3140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018435224111081468,
+      "loss": 3.6374,
+      "step": 3141
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018434088055408826,
+      "loss": 3.6289,
+      "step": 3142
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018432951622518053,
+      "loss": 3.6427,
+      "step": 3143
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018431814812459968,
+      "loss": 3.6201,
+      "step": 3144
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018430677625285425,
+      "loss": 3.6394,
+      "step": 3145
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018429540061045278,
+      "loss": 3.6576,
+      "step": 3146
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018428402119790408,
+      "loss": 3.6447,
+      "step": 3147
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018427263801571714,
+      "loss": 3.6473,
+      "step": 3148
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000184261251064401,
+      "loss": 3.6265,
+      "step": 3149
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018424986034446499,
+      "loss": 3.6596,
+      "step": 3150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018423846585641855,
+      "loss": 3.6367,
+      "step": 3151
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001842270676007713,
+      "loss": 3.6272,
+      "step": 3152
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018421566557803306,
+      "loss": 3.6342,
+      "step": 3153
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018420425978871378,
+      "loss": 3.6217,
+      "step": 3154
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018419285023332357,
+      "loss": 3.6591,
+      "step": 3155
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018418143691237274,
+      "loss": 3.6402,
+      "step": 3156
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018417001982637177,
+      "loss": 3.6407,
+      "step": 3157
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018415859897583125,
+      "loss": 3.6428,
+      "step": 3158
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018414717436126204,
+      "loss": 3.6585,
+      "step": 3159
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018413574598317504,
+      "loss": 3.6151,
+      "step": 3160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018412431384208147,
+      "loss": 3.6482,
+      "step": 3161
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018411287793849258,
+      "loss": 3.6587,
+      "step": 3162
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018410143827291985,
+      "loss": 3.6508,
+      "step": 3163
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018408999484587497,
+      "loss": 3.6446,
+      "step": 3164
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001840785476578697,
+      "loss": 3.6434,
+      "step": 3165
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018406709670941603,
+      "loss": 3.6664,
+      "step": 3166
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018405564200102612,
+      "loss": 3.6491,
+      "step": 3167
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001840441835332123,
+      "loss": 3.6416,
+      "step": 3168
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018403272130648702,
+      "loss": 3.625,
+      "step": 3169
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018402125532136293,
+      "loss": 3.6391,
+      "step": 3170
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018400978557835292,
+      "loss": 3.6436,
+      "step": 3171
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018399831207796993,
+      "loss": 3.613,
+      "step": 3172
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018398683482072708,
+      "loss": 3.659,
+      "step": 3173
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018397535380713776,
+      "loss": 3.6338,
+      "step": 3174
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018396386903771545,
+      "loss": 3.6277,
+      "step": 3175
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018395238051297378,
+      "loss": 3.606,
+      "step": 3176
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001839408882334266,
+      "loss": 3.6748,
+      "step": 3177
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001839293921995879,
+      "loss": 3.6462,
+      "step": 3178
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018391789241197187,
+      "loss": 3.6354,
+      "step": 3179
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018390638887109278,
+      "loss": 3.6427,
+      "step": 3180
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018389488157746523,
+      "loss": 3.6117,
+      "step": 3181
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018388337053160377,
+      "loss": 3.6532,
+      "step": 3182
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018387185573402336,
+      "loss": 3.6471,
+      "step": 3183
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001838603371852389,
+      "loss": 3.6421,
+      "step": 3184
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001838488148857656,
+      "loss": 3.6587,
+      "step": 3185
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018383728883611885,
+      "loss": 3.6275,
+      "step": 3186
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018382575903681407,
+      "loss": 3.6425,
+      "step": 3187
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018381422548836696,
+      "loss": 3.6418,
+      "step": 3188
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001838026881912934,
+      "loss": 3.6183,
+      "step": 3189
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001837911471461094,
+      "loss": 3.6573,
+      "step": 3190
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001837796023533311,
+      "loss": 3.6579,
+      "step": 3191
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001837680538134749,
+      "loss": 3.6389,
+      "step": 3192
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018375650152705722,
+      "loss": 3.6398,
+      "step": 3193
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018374494549459482,
+      "loss": 3.6218,
+      "step": 3194
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018373338571660451,
+      "loss": 3.6312,
+      "step": 3195
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018372182219360337,
+      "loss": 3.6485,
+      "step": 3196
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001837102549261085,
+      "loss": 3.6471,
+      "step": 3197
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001836986839146373,
+      "loss": 3.6399,
+      "step": 3198
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018368710915970727,
+      "loss": 3.6193,
+      "step": 3199
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018367553066183607,
+      "loss": 3.6237,
+      "step": 3200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001836639484215416,
+      "loss": 3.6646,
+      "step": 3201
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001836523624393419,
+      "loss": 3.635,
+      "step": 3202
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001836407727157551,
+      "loss": 3.6195,
+      "step": 3203
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018362917925129955,
+      "loss": 3.6604,
+      "step": 3204
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001836175820464938,
+      "loss": 3.6595,
+      "step": 3205
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001836059811018566,
+      "loss": 3.6375,
+      "step": 3206
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018359437641790672,
+      "loss": 3.6416,
+      "step": 3207
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018358276799516316,
+      "loss": 3.6593,
+      "step": 3208
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018357115583414522,
+      "loss": 3.6373,
+      "step": 3209
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018355953993537218,
+      "loss": 3.6429,
+      "step": 3210
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001835479202993636,
+      "loss": 3.6316,
+      "step": 3211
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018353629692663916,
+      "loss": 3.6452,
+      "step": 3212
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001835246698177187,
+      "loss": 3.6323,
+      "step": 3213
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018351303897312229,
+      "loss": 3.6517,
+      "step": 3214
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018350140439337009,
+      "loss": 3.6406,
+      "step": 3215
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018348976607898248,
+      "loss": 3.6125,
+      "step": 3216
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018347812403047997,
+      "loss": 3.6161,
+      "step": 3217
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001834664782483833,
+      "loss": 3.6226,
+      "step": 3218
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018345482873321326,
+      "loss": 3.6328,
+      "step": 3219
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018344317548549095,
+      "loss": 3.6387,
+      "step": 3220
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018343151850573752,
+      "loss": 3.6363,
+      "step": 3221
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018341985779447435,
+      "loss": 3.6473,
+      "step": 3222
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000183408193352223,
+      "loss": 3.6554,
+      "step": 3223
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018339652517950509,
+      "loss": 3.6436,
+      "step": 3224
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018338485327684255,
+      "loss": 3.6014,
+      "step": 3225
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001833731776447574,
+      "loss": 3.6379,
+      "step": 3226
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018336149828377182,
+      "loss": 3.6624,
+      "step": 3227
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001833498151944082,
+      "loss": 3.6559,
+      "step": 3228
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018333812837718905,
+      "loss": 3.6678,
+      "step": 3229
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018332643783263705,
+      "loss": 3.6482,
+      "step": 3230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018331474356127512,
+      "loss": 3.6687,
+      "step": 3231
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018330304556362625,
+      "loss": 3.6546,
+      "step": 3232
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018329134384021363,
+      "loss": 3.638,
+      "step": 3233
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018327963839156065,
+      "loss": 3.6484,
+      "step": 3234
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018326792921819082,
+      "loss": 3.6504,
+      "step": 3235
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018325621632062786,
+      "loss": 3.65,
+      "step": 3236
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018324449969939563,
+      "loss": 3.6397,
+      "step": 3237
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018323277935501815,
+      "loss": 3.6365,
+      "step": 3238
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018322105528801962,
+      "loss": 3.6446,
+      "step": 3239
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018320932749892442,
+      "loss": 3.6462,
+      "step": 3240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018319759598825702,
+      "loss": 3.6551,
+      "step": 3241
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001831858607565422,
+      "loss": 3.6469,
+      "step": 3242
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001831741218043048,
+      "loss": 3.639,
+      "step": 3243
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001831623791320698,
+      "loss": 3.6353,
+      "step": 3244
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018315063274036247,
+      "loss": 3.6208,
+      "step": 3245
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001831388826297081,
+      "loss": 3.6611,
+      "step": 3246
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018312712880063227,
+      "loss": 3.6432,
+      "step": 3247
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018311537125366062,
+      "loss": 3.6602,
+      "step": 3248
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001831036099893191,
+      "loss": 3.62,
+      "step": 3249
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018309184500813367,
+      "loss": 3.6407,
+      "step": 3250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018308007631063055,
+      "loss": 3.6577,
+      "step": 3251
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018306830389733608,
+      "loss": 3.6254,
+      "step": 3252
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018305652776877678,
+      "loss": 3.6474,
+      "step": 3253
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018304474792547935,
+      "loss": 3.639,
+      "step": 3254
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018303296436797067,
+      "loss": 3.6402,
+      "step": 3255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018302117709677776,
+      "loss": 3.6532,
+      "step": 3256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018300938611242782,
+      "loss": 3.6426,
+      "step": 3257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018299759141544817,
+      "loss": 3.6246,
+      "step": 3258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018298579300636634,
+      "loss": 3.6647,
+      "step": 3259
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018297399088571003,
+      "loss": 3.6502,
+      "step": 3260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001829621850540071,
+      "loss": 3.6468,
+      "step": 3261
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018295037551178557,
+      "loss": 3.6427,
+      "step": 3262
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018293856225957362,
+      "loss": 3.6407,
+      "step": 3263
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018292674529789963,
+      "loss": 3.6368,
+      "step": 3264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018291492462729205,
+      "loss": 3.6425,
+      "step": 3265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018290310024827964,
+      "loss": 3.6693,
+      "step": 3266
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001828912721613912,
+      "loss": 3.6566,
+      "step": 3267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001828794403671558,
+      "loss": 3.637,
+      "step": 3268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001828676048661025,
+      "loss": 3.6336,
+      "step": 3269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018285576565876084,
+      "loss": 3.6484,
+      "step": 3270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018284392274566015,
+      "loss": 3.6287,
+      "step": 3271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018283207612733023,
+      "loss": 3.6265,
+      "step": 3272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001828202258043009,
+      "loss": 3.6325,
+      "step": 3273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001828083717771021,
+      "loss": 3.6614,
+      "step": 3274
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001827965140462641,
+      "loss": 3.6445,
+      "step": 3275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018278465261231718,
+      "loss": 3.6207,
+      "step": 3276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018277278747579184,
+      "loss": 3.646,
+      "step": 3277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018276091863721883,
+      "loss": 3.6443,
+      "step": 3278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018274904609712892,
+      "loss": 3.6227,
+      "step": 3279
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018273716985605317,
+      "loss": 3.6349,
+      "step": 3280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018272528991452266,
+      "loss": 3.6343,
+      "step": 3281
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018271340627306883,
+      "loss": 3.6248,
+      "step": 3282
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001827015189322231,
+      "loss": 3.6413,
+      "step": 3283
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018268962789251718,
+      "loss": 3.6433,
+      "step": 3284
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018267773315448288,
+      "loss": 3.6472,
+      "step": 3285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001826658347186522,
+      "loss": 3.6439,
+      "step": 3286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018265393258555732,
+      "loss": 3.6342,
+      "step": 3287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018264202675573056,
+      "loss": 3.6304,
+      "step": 3288
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001826301172297044,
+      "loss": 3.6361,
+      "step": 3289
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018261820400801153,
+      "loss": 3.6151,
+      "step": 3290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018260628709118473,
+      "loss": 3.6453,
+      "step": 3291
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018259436647975706,
+      "loss": 3.6608,
+      "step": 3292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018258244217426154,
+      "loss": 3.6358,
+      "step": 3293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018257051417523164,
+      "loss": 3.6466,
+      "step": 3294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018255858248320076,
+      "loss": 3.6512,
+      "step": 3295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001825466470987026,
+      "loss": 3.6284,
+      "step": 3296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001825347080222709,
+      "loss": 3.6512,
+      "step": 3297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018252276525443972,
+      "loss": 3.6261,
+      "step": 3298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018251081879574313,
+      "loss": 3.6464,
+      "step": 3299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001824988686467155,
+      "loss": 3.6569,
+      "step": 3300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018248691480789132,
+      "loss": 3.6386,
+      "step": 3301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018247495727980516,
+      "loss": 3.6364,
+      "step": 3302
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018246299606299187,
+      "loss": 3.6498,
+      "step": 3303
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018245103115798643,
+      "loss": 3.6434,
+      "step": 3304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001824390625653239,
+      "loss": 3.6579,
+      "step": 3305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001824270902855397,
+      "loss": 3.6451,
+      "step": 3306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018241511431916924,
+      "loss": 3.6426,
+      "step": 3307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001824031346667481,
+      "loss": 3.6315,
+      "step": 3308
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018239115132881218,
+      "loss": 3.6507,
+      "step": 3309
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018237916430589732,
+      "loss": 3.635,
+      "step": 3310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018236717359853974,
+      "loss": 3.6384,
+      "step": 3311
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018235517920727569,
+      "loss": 3.6508,
+      "step": 3312
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001823431811326416,
+      "loss": 3.6395,
+      "step": 3313
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018233117937517415,
+      "loss": 3.6494,
+      "step": 3314
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018231917393541006,
+      "loss": 3.6456,
+      "step": 3315
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018230716481388632,
+      "loss": 3.6304,
+      "step": 3316
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018229515201114005,
+      "loss": 3.6468,
+      "step": 3317
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018228313552770848,
+      "loss": 3.6155,
+      "step": 3318
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018227111536412907,
+      "loss": 3.648,
+      "step": 3319
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001822590915209395,
+      "loss": 3.639,
+      "step": 3320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018224706399867744,
+      "loss": 3.6439,
+      "step": 3321
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018223503279788086,
+      "loss": 3.6416,
+      "step": 3322
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001822229979190879,
+      "loss": 3.6571,
+      "step": 3323
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001822109593628367,
+      "loss": 3.6423,
+      "step": 3324
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018219891712966586,
+      "loss": 3.6134,
+      "step": 3325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018218687122011388,
+      "loss": 3.6415,
+      "step": 3326
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018217482163471955,
+      "loss": 3.6436,
+      "step": 3327
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018216276837402175,
+      "loss": 3.6465,
+      "step": 3328
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018215071143855963,
+      "loss": 3.6199,
+      "step": 3329
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018213865082887236,
+      "loss": 3.6317,
+      "step": 3330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018212658654549938,
+      "loss": 3.6512,
+      "step": 3331
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018211451858898036,
+      "loss": 3.6572,
+      "step": 3332
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018210244695985494,
+      "loss": 3.6489,
+      "step": 3333
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018209037165866306,
+      "loss": 3.6137,
+      "step": 3334
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018207829268594478,
+      "loss": 3.6237,
+      "step": 3335
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018206621004224037,
+      "loss": 3.6471,
+      "step": 3336
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018205412372809023,
+      "loss": 3.6764,
+      "step": 3337
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001820420337440349,
+      "loss": 3.6437,
+      "step": 3338
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001820299400906151,
+      "loss": 3.6438,
+      "step": 3339
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018201784276837177,
+      "loss": 3.6647,
+      "step": 3340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018200574177784594,
+      "loss": 3.6287,
+      "step": 3341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018199363711957883,
+      "loss": 3.618,
+      "step": 3342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018198152879411183,
+      "loss": 3.616,
+      "step": 3343
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018196941680198648,
+      "loss": 3.6466,
+      "step": 3344
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001819573011437445,
+      "loss": 3.639,
+      "step": 3345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001819451818199278,
+      "loss": 3.651,
+      "step": 3346
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018193305883107836,
+      "loss": 3.6257,
+      "step": 3347
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018192093217773842,
+      "loss": 3.6378,
+      "step": 3348
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018190880186045037,
+      "loss": 3.6363,
+      "step": 3349
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018189666787975673,
+      "loss": 3.6281,
+      "step": 3350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018188453023620016,
+      "loss": 3.6587,
+      "step": 3351
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018187238893032354,
+      "loss": 3.6339,
+      "step": 3352
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018186024396266994,
+      "loss": 3.6657,
+      "step": 3353
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001818480953337825,
+      "loss": 3.6215,
+      "step": 3354
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018183594304420459,
+      "loss": 3.6319,
+      "step": 3355
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018182378709447972,
+      "loss": 3.6366,
+      "step": 3356
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001818116274851516,
+      "loss": 3.6329,
+      "step": 3357
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018179946421676403,
+      "loss": 3.6352,
+      "step": 3358
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000181787297289861,
+      "loss": 3.6267,
+      "step": 3359
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018177512670498674,
+      "loss": 3.6348,
+      "step": 3360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018176295246268554,
+      "loss": 3.6396,
+      "step": 3361
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018175077456350196,
+      "loss": 3.6126,
+      "step": 3362
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018173859300798058,
+      "loss": 3.6226,
+      "step": 3363
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001817264077966663,
+      "loss": 3.6473,
+      "step": 3364
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018171421893010403,
+      "loss": 3.6261,
+      "step": 3365
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018170202640883899,
+      "loss": 3.6522,
+      "step": 3366
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001816898302334165,
+      "loss": 3.6459,
+      "step": 3367
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018167763040438198,
+      "loss": 3.6391,
+      "step": 3368
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001816654269222811,
+      "loss": 3.6226,
+      "step": 3369
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018165321978765966,
+      "loss": 3.6292,
+      "step": 3370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018164100900106364,
+      "loss": 3.6534,
+      "step": 3371
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018162879456303917,
+      "loss": 3.6409,
+      "step": 3372
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018161657647413257,
+      "loss": 3.6245,
+      "step": 3373
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018160435473489026,
+      "loss": 3.6226,
+      "step": 3374
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018159212934585887,
+      "loss": 3.6575,
+      "step": 3375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001815799003075852,
+      "loss": 3.6295,
+      "step": 3376
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001815676676206162,
+      "loss": 3.6497,
+      "step": 3377
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018155543128549898,
+      "loss": 3.6674,
+      "step": 3378
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001815431913027808,
+      "loss": 3.6387,
+      "step": 3379
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018153094767300908,
+      "loss": 3.6448,
+      "step": 3380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018151870039673148,
+      "loss": 3.6099,
+      "step": 3381
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018150644947449574,
+      "loss": 3.616,
+      "step": 3382
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018149419490684975,
+      "loss": 3.6279,
+      "step": 3383
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018148193669434166,
+      "loss": 3.6475,
+      "step": 3384
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018146967483751968,
+      "loss": 3.6106,
+      "step": 3385
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018145740933693224,
+      "loss": 3.6288,
+      "step": 3386
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018144514019312796,
+      "loss": 3.6181,
+      "step": 3387
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018143286740665552,
+      "loss": 3.637,
+      "step": 3388
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018142059097806383,
+      "loss": 3.6477,
+      "step": 3389
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000181408310907902,
+      "loss": 3.6175,
+      "step": 3390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018139602719671925,
+      "loss": 3.6436,
+      "step": 3391
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018138373984506493,
+      "loss": 3.6443,
+      "step": 3392
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018137144885348866,
+      "loss": 3.6381,
+      "step": 3393
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018135915422254012,
+      "loss": 3.6351,
+      "step": 3394
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001813468559527692,
+      "loss": 3.6544,
+      "step": 3395
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018133455404472596,
+      "loss": 3.6405,
+      "step": 3396
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001813222484989606,
+      "loss": 3.639,
+      "step": 3397
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001813099393160235,
+      "loss": 3.6205,
+      "step": 3398
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018129762649646516,
+      "loss": 3.632,
+      "step": 3399
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001812853100408363,
+      "loss": 3.6335,
+      "step": 3400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018127298994968774,
+      "loss": 3.6272,
+      "step": 3401
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001812606662235706,
+      "loss": 3.6525,
+      "step": 3402
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018124833886303595,
+      "loss": 3.6251,
+      "step": 3403
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018123600786863522,
+      "loss": 3.6233,
+      "step": 3404
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018122367324091988,
+      "loss": 3.6497,
+      "step": 3405
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018121133498044158,
+      "loss": 3.6363,
+      "step": 3406
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018119899308775223,
+      "loss": 3.6347,
+      "step": 3407
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018118664756340373,
+      "loss": 3.6218,
+      "step": 3408
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018117429840794832,
+      "loss": 3.6157,
+      "step": 3409
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018116194562193828,
+      "loss": 3.6282,
+      "step": 3410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001811495892059261,
+      "loss": 3.6399,
+      "step": 3411
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018113722916046442,
+      "loss": 3.6619,
+      "step": 3412
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018112486548610605,
+      "loss": 3.6438,
+      "step": 3413
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000181112498183404,
+      "loss": 3.6461,
+      "step": 3414
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018110012725291132,
+      "loss": 3.6429,
+      "step": 3415
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018108775269518138,
+      "loss": 3.6534,
+      "step": 3416
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001810753745107676,
+      "loss": 3.6387,
+      "step": 3417
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018106299270022365,
+      "loss": 3.6417,
+      "step": 3418
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018105060726410322,
+      "loss": 3.6448,
+      "step": 3419
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018103821820296032,
+      "loss": 3.6291,
+      "step": 3420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018102582551734905,
+      "loss": 3.6361,
+      "step": 3421
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018101342920782368,
+      "loss": 3.6451,
+      "step": 3422
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018100102927493862,
+      "loss": 3.6425,
+      "step": 3423
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018098862571924848,
+      "loss": 3.6306,
+      "step": 3424
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018097621854130797,
+      "loss": 3.6471,
+      "step": 3425
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001809638077416721,
+      "loss": 3.6254,
+      "step": 3426
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018095139332089587,
+      "loss": 3.6639,
+      "step": 3427
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018093897527953452,
+      "loss": 3.6556,
+      "step": 3428
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018092655361814347,
+      "loss": 3.6322,
+      "step": 3429
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018091412833727831,
+      "loss": 3.6494,
+      "step": 3430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 3.6691,
+      "step": 3431
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018088926691934865,
+      "loss": 3.6468,
+      "step": 3432
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018087683078339613,
+      "loss": 3.6453,
+      "step": 3433
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018086439103019331,
+      "loss": 3.6254,
+      "step": 3434
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018085194766029658,
+      "loss": 3.6411,
+      "step": 3435
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018083950067426251,
+      "loss": 3.6365,
+      "step": 3436
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018082705007264785,
+      "loss": 3.6563,
+      "step": 3437
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018081459585600932,
+      "loss": 3.6327,
+      "step": 3438
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018080213802490404,
+      "loss": 3.632,
+      "step": 3439
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001807896765798892,
+      "loss": 3.612,
+      "step": 3440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018077721152152208,
+      "loss": 3.6454,
+      "step": 3441
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018076474285036022,
+      "loss": 3.6669,
+      "step": 3442
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018075227056696126,
+      "loss": 3.6301,
+      "step": 3443
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018073979467188309,
+      "loss": 3.639,
+      "step": 3444
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018072731516568364,
+      "loss": 3.6221,
+      "step": 3445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001807148320489211,
+      "loss": 3.643,
+      "step": 3446
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018070234532215374,
+      "loss": 3.6317,
+      "step": 3447
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018068985498594006,
+      "loss": 3.6477,
+      "step": 3448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018067736104083868,
+      "loss": 3.6421,
+      "step": 3449
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018066486348740845,
+      "loss": 3.6376,
+      "step": 3450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018065236232620827,
+      "loss": 3.64,
+      "step": 3451
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018063985755779728,
+      "loss": 3.6624,
+      "step": 3452
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018062734918273475,
+      "loss": 3.642,
+      "step": 3453
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018061483720158012,
+      "loss": 3.6713,
+      "step": 3454
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000180602321614893,
+      "loss": 3.6426,
+      "step": 3455
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001805898024232332,
+      "loss": 3.6427,
+      "step": 3456
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001805772796271606,
+      "loss": 3.6324,
+      "step": 3457
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018056475322723528,
+      "loss": 3.6438,
+      "step": 3458
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001805522232240175,
+      "loss": 3.6391,
+      "step": 3459
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001805396896180677,
+      "loss": 3.6236,
+      "step": 3460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018052715240994639,
+      "loss": 3.6389,
+      "step": 3461
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018051461160021432,
+      "loss": 3.66,
+      "step": 3462
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001805020671894324,
+      "loss": 3.6162,
+      "step": 3463
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018048951917816174,
+      "loss": 3.6217,
+      "step": 3464
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018047696756696345,
+      "loss": 3.6349,
+      "step": 3465
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018046441235639894,
+      "loss": 3.6204,
+      "step": 3466
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018045185354702978,
+      "loss": 3.6381,
+      "step": 3467
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018043929113941765,
+      "loss": 3.6497,
+      "step": 3468
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018042672513412443,
+      "loss": 3.6421,
+      "step": 3469
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018041415553171207,
+      "loss": 3.6708,
+      "step": 3470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018040158233274283,
+      "loss": 3.6532,
+      "step": 3471
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018038900553777899,
+      "loss": 3.6336,
+      "step": 3472
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018037642514738314,
+      "loss": 3.625,
+      "step": 3473
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018036384116211783,
+      "loss": 3.6487,
+      "step": 3474
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018035125358254595,
+      "loss": 3.619,
+      "step": 3475
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018033866240923046,
+      "loss": 3.6374,
+      "step": 3476
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018032606764273456,
+      "loss": 3.6588,
+      "step": 3477
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018031346928362146,
+      "loss": 3.6471,
+      "step": 3478
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018030086733245474,
+      "loss": 3.6425,
+      "step": 3479
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018028826178979795,
+      "loss": 3.6612,
+      "step": 3480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018027565265621486,
+      "loss": 3.6348,
+      "step": 3481
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001802630399322695,
+      "loss": 3.6265,
+      "step": 3482
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018025042361852594,
+      "loss": 3.6363,
+      "step": 3483
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001802378037155484,
+      "loss": 3.6127,
+      "step": 3484
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001802251802239014,
+      "loss": 3.6571,
+      "step": 3485
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018021255314414948,
+      "loss": 3.6526,
+      "step": 3486
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018019992247685738,
+      "loss": 3.6418,
+      "step": 3487
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018018728822259003,
+      "loss": 3.6582,
+      "step": 3488
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018017465038191254,
+      "loss": 3.6504,
+      "step": 3489
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018016200895539007,
+      "loss": 3.6518,
+      "step": 3490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018014936394358805,
+      "loss": 3.6321,
+      "step": 3491
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018013671534707206,
+      "loss": 3.6252,
+      "step": 3492
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018012406316640774,
+      "loss": 3.6445,
+      "step": 3493
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018011140740216105,
+      "loss": 3.6315,
+      "step": 3494
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018009874805489798,
+      "loss": 3.6496,
+      "step": 3495
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001800860851251847,
+      "loss": 3.6314,
+      "step": 3496
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018007341861358764,
+      "loss": 3.6124,
+      "step": 3497
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018006074852067324,
+      "loss": 3.6461,
+      "step": 3498
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001800480748470082,
+      "loss": 3.6363,
+      "step": 3499
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018003539759315937,
+      "loss": 3.6228,
+      "step": 3500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018002271675969376,
+      "loss": 3.6206,
+      "step": 3501
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018001003234717846,
+      "loss": 3.6419,
+      "step": 3502
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017999734435618085,
+      "loss": 3.6325,
+      "step": 3503
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001799846527872684,
+      "loss": 3.6347,
+      "step": 3504
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001799719576410087,
+      "loss": 3.6217,
+      "step": 3505
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017995925891796957,
+      "loss": 3.643,
+      "step": 3506
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017994655661871897,
+      "loss": 3.6321,
+      "step": 3507
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017993385074382504,
+      "loss": 3.6545,
+      "step": 3508
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000179921141293856,
+      "loss": 3.6238,
+      "step": 3509
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017990842826938033,
+      "loss": 3.6729,
+      "step": 3510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001798957116709666,
+      "loss": 3.6179,
+      "step": 3511
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017988299149918361,
+      "loss": 3.6519,
+      "step": 3512
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001798702677546002,
+      "loss": 3.6442,
+      "step": 3513
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017985754043778554,
+      "loss": 3.6464,
+      "step": 3514
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017984480954930877,
+      "loss": 3.6524,
+      "step": 3515
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017983207508973933,
+      "loss": 3.6232,
+      "step": 3516
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017981933705964678,
+      "loss": 3.6498,
+      "step": 3517
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001798065954596008,
+      "loss": 3.6159,
+      "step": 3518
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001797938502901713,
+      "loss": 3.6671,
+      "step": 3519
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001797811015519283,
+      "loss": 3.6301,
+      "step": 3520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017976834924544202,
+      "loss": 3.6319,
+      "step": 3521
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017975559337128275,
+      "loss": 3.6545,
+      "step": 3522
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017974283393002108,
+      "loss": 3.6159,
+      "step": 3523
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001797300709222276,
+      "loss": 3.6517,
+      "step": 3524
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017971730434847318,
+      "loss": 3.6384,
+      "step": 3525
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017970453420932883,
+      "loss": 3.6573,
+      "step": 3526
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017969176050536566,
+      "loss": 3.6412,
+      "step": 3527
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000179678983237155,
+      "loss": 3.6401,
+      "step": 3528
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001796662024052683,
+      "loss": 3.6131,
+      "step": 3529
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017965341801027724,
+      "loss": 3.6667,
+      "step": 3530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017964063005275355,
+      "loss": 3.6417,
+      "step": 3531
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001796278385332692,
+      "loss": 3.6437,
+      "step": 3532
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017961504345239632,
+      "loss": 3.6443,
+      "step": 3533
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017960224481070714,
+      "loss": 3.6392,
+      "step": 3534
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017958944260877409,
+      "loss": 3.6404,
+      "step": 3535
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017957663684716976,
+      "loss": 3.6128,
+      "step": 3536
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001795638275264669,
+      "loss": 3.6224,
+      "step": 3537
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001795510146472384,
+      "loss": 3.631,
+      "step": 3538
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017953819821005734,
+      "loss": 3.6567,
+      "step": 3539
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001795253782154969,
+      "loss": 3.6292,
+      "step": 3540
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001795125546641305,
+      "loss": 3.6167,
+      "step": 3541
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017949972755653166,
+      "loss": 3.6396,
+      "step": 3542
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017948689689327408,
+      "loss": 3.644,
+      "step": 3543
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017947406267493165,
+      "loss": 3.6405,
+      "step": 3544
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001794612249020783,
+      "loss": 3.6415,
+      "step": 3545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001794483835752883,
+      "loss": 3.654,
+      "step": 3546
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017943553869513595,
+      "loss": 3.6487,
+      "step": 3547
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017942269026219573,
+      "loss": 3.6189,
+      "step": 3548
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001794098382770423,
+      "loss": 3.6559,
+      "step": 3549
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017939698274025044,
+      "loss": 3.6376,
+      "step": 3550
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017938412365239517,
+      "loss": 3.6351,
+      "step": 3551
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017937126101405163,
+      "loss": 3.6402,
+      "step": 3552
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017935839482579498,
+      "loss": 3.6467,
+      "step": 3553
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017934552508820084,
+      "loss": 3.6312,
+      "step": 3554
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001793326518018447,
+      "loss": 3.6385,
+      "step": 3555
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001793197749673024,
+      "loss": 3.6108,
+      "step": 3556
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017930689458514975,
+      "loss": 3.6406,
+      "step": 3557
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017929401065596294,
+      "loss": 3.6548,
+      "step": 3558
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017928112318031815,
+      "loss": 3.647,
+      "step": 3559
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017926823215879182,
+      "loss": 3.6405,
+      "step": 3560
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017925533759196048,
+      "loss": 3.6349,
+      "step": 3561
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017924243948040083,
+      "loss": 3.6406,
+      "step": 3562
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017922953782468977,
+      "loss": 3.6457,
+      "step": 3563
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017921663262540432,
+      "loss": 3.6494,
+      "step": 3564
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017920372388312164,
+      "loss": 3.6373,
+      "step": 3565
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017919081159841914,
+      "loss": 3.6496,
+      "step": 3566
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017917789577187433,
+      "loss": 3.6596,
+      "step": 3567
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017916497640406483,
+      "loss": 3.6258,
+      "step": 3568
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017915205349556846,
+      "loss": 3.6254,
+      "step": 3569
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001791391270469632,
+      "loss": 3.6238,
+      "step": 3570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017912619705882726,
+      "loss": 3.6334,
+      "step": 3571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017911326353173886,
+      "loss": 3.6559,
+      "step": 3572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001791003264662765,
+      "loss": 3.6205,
+      "step": 3573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017908738586301878,
+      "loss": 3.6611,
+      "step": 3574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017907444172254451,
+      "loss": 3.6379,
+      "step": 3575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017906149404543252,
+      "loss": 3.6231,
+      "step": 3576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017904854283226203,
+      "loss": 3.6431,
+      "step": 3577
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017903558808361224,
+      "loss": 3.6323,
+      "step": 3578
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001790226298000625,
+      "loss": 3.628,
+      "step": 3579
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017900966798219242,
+      "loss": 3.6367,
+      "step": 3580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017899670263058176,
+      "loss": 3.626,
+      "step": 3581
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017898373374581036,
+      "loss": 3.6413,
+      "step": 3582
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017897076132845826,
+      "loss": 3.6604,
+      "step": 3583
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017895778537910562,
+      "loss": 3.6267,
+      "step": 3584
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017894480589833288,
+      "loss": 3.6394,
+      "step": 3585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001789318228867205,
+      "loss": 3.6559,
+      "step": 3586
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017891883634484917,
+      "loss": 3.6338,
+      "step": 3587
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017890584627329968,
+      "loss": 3.6358,
+      "step": 3588
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017889285267265305,
+      "loss": 3.6421,
+      "step": 3589
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017887985554349042,
+      "loss": 3.64,
+      "step": 3590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017886685488639309,
+      "loss": 3.6625,
+      "step": 3591
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017885385070194251,
+      "loss": 3.6412,
+      "step": 3592
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001788408429907203,
+      "loss": 3.6425,
+      "step": 3593
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017882783175330826,
+      "loss": 3.6251,
+      "step": 3594
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017881481699028831,
+      "loss": 3.656,
+      "step": 3595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017880179870224253,
+      "loss": 3.6214,
+      "step": 3596
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017878877688975314,
+      "loss": 3.6375,
+      "step": 3597
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017877575155340262,
+      "loss": 3.6332,
+      "step": 3598
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001787627226937735,
+      "loss": 3.6376,
+      "step": 3599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017874969031144846,
+      "loss": 3.6138,
+      "step": 3600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017873665440701044,
+      "loss": 3.6464,
+      "step": 3601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017872361498104245,
+      "loss": 3.6423,
+      "step": 3602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001787105720341277,
+      "loss": 3.6463,
+      "step": 3603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001786975255668495,
+      "loss": 3.6174,
+      "step": 3604
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017868447557979137,
+      "loss": 3.6542,
+      "step": 3605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000178671422073537,
+      "loss": 3.6338,
+      "step": 3606
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017865836504867022,
+      "loss": 3.6446,
+      "step": 3607
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000178645304505775,
+      "loss": 3.652,
+      "step": 3608
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017863224044543547,
+      "loss": 3.6308,
+      "step": 3609
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001786191728682359,
+      "loss": 3.663,
+      "step": 3610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001786061017747608,
+      "loss": 3.6298,
+      "step": 3611
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017859302716559475,
+      "loss": 3.6382,
+      "step": 3612
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017857994904132252,
+      "loss": 3.6262,
+      "step": 3613
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017856686740252902,
+      "loss": 3.6543,
+      "step": 3614
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017855378224979936,
+      "loss": 3.6229,
+      "step": 3615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017854069358371873,
+      "loss": 3.64,
+      "step": 3616
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001785276014048726,
+      "loss": 3.6327,
+      "step": 3617
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017851450571384646,
+      "loss": 3.6355,
+      "step": 3618
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017850140651122607,
+      "loss": 3.6364,
+      "step": 3619
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017848830379759724,
+      "loss": 3.6216,
+      "step": 3620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017847519757354605,
+      "loss": 3.633,
+      "step": 3621
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017846208783965864,
+      "loss": 3.6377,
+      "step": 3622
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001784489745965214,
+      "loss": 3.6406,
+      "step": 3623
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017843585784472078,
+      "loss": 3.6253,
+      "step": 3624
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001784227375848434,
+      "loss": 3.6327,
+      "step": 3625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017840961381747613,
+      "loss": 3.6461,
+      "step": 3626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017839648654320594,
+      "loss": 3.6264,
+      "step": 3627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017838335576261994,
+      "loss": 3.6403,
+      "step": 3628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017837022147630536,
+      "loss": 3.6485,
+      "step": 3629
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017835708368484969,
+      "loss": 3.6596,
+      "step": 3630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017834394238884054,
+      "loss": 3.637,
+      "step": 3631
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001783307975888656,
+      "loss": 3.6311,
+      "step": 3632
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017831764928551283,
+      "loss": 3.6382,
+      "step": 3633
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017830449747937026,
+      "loss": 3.6329,
+      "step": 3634
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001782913421710261,
+      "loss": 3.6374,
+      "step": 3635
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017827818336106876,
+      "loss": 3.6363,
+      "step": 3636
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017826502105008676,
+      "loss": 3.6449,
+      "step": 3637
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001782518552386688,
+      "loss": 3.6362,
+      "step": 3638
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001782386859274037,
+      "loss": 3.6467,
+      "step": 3639
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001782255131168805,
+      "loss": 3.6509,
+      "step": 3640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017821233680768834,
+      "loss": 3.6356,
+      "step": 3641
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017819915700041653,
+      "loss": 3.6759,
+      "step": 3642
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017818597369565454,
+      "loss": 3.6469,
+      "step": 3643
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000178172786893992,
+      "loss": 3.6175,
+      "step": 3644
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017815959659601873,
+      "loss": 3.6491,
+      "step": 3645
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017814640280232463,
+      "loss": 3.654,
+      "step": 3646
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017813320551349982,
+      "loss": 3.644,
+      "step": 3647
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017812000473013453,
+      "loss": 3.6219,
+      "step": 3648
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001781068004528192,
+      "loss": 3.6294,
+      "step": 3649
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017809359268214437,
+      "loss": 3.6235,
+      "step": 3650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017808038141870077,
+      "loss": 3.6146,
+      "step": 3651
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017806716666307932,
+      "loss": 3.6503,
+      "step": 3652
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017805394841587097,
+      "loss": 3.632,
+      "step": 3653
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000178040726677667,
+      "loss": 3.6281,
+      "step": 3654
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017802750144905872,
+      "loss": 3.6241,
+      "step": 3655
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017801427273063763,
+      "loss": 3.6491,
+      "step": 3656
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001780010405229954,
+      "loss": 3.6411,
+      "step": 3657
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017798780482672382,
+      "loss": 3.6306,
+      "step": 3658
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001779745656424149,
+      "loss": 3.6284,
+      "step": 3659
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017796132297066074,
+      "loss": 3.6348,
+      "step": 3660
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017794807681205364,
+      "loss": 3.6441,
+      "step": 3661
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017793482716718605,
+      "loss": 3.6405,
+      "step": 3662
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017792157403665054,
+      "loss": 3.6511,
+      "step": 3663
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017790831742103984,
+      "loss": 3.6398,
+      "step": 3664
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017789505732094694,
+      "loss": 3.643,
+      "step": 3665
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001778817937369648,
+      "loss": 3.6396,
+      "step": 3666
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017786852666968673,
+      "loss": 3.6499,
+      "step": 3667
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017785525611970607,
+      "loss": 3.6423,
+      "step": 3668
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017784198208761634,
+      "loss": 3.6365,
+      "step": 3669
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017782870457401124,
+      "loss": 3.6233,
+      "step": 3670
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017781542357948462,
+      "loss": 3.6361,
+      "step": 3671
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017780213910463047,
+      "loss": 3.6323,
+      "step": 3672
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001777888511500429,
+      "loss": 3.6221,
+      "step": 3673
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001777755597163163,
+      "loss": 3.6203,
+      "step": 3674
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017776226480404505,
+      "loss": 3.6368,
+      "step": 3675
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017774896641382384,
+      "loss": 3.6293,
+      "step": 3676
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017773566454624745,
+      "loss": 3.6386,
+      "step": 3677
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017772235920191073,
+      "loss": 3.6224,
+      "step": 3678
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017770905038140886,
+      "loss": 3.6301,
+      "step": 3679
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017769573808533703,
+      "loss": 3.6568,
+      "step": 3680
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017768242231429063,
+      "loss": 3.6413,
+      "step": 3681
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017766910306886527,
+      "loss": 3.6155,
+      "step": 3682
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017765578034965662,
+      "loss": 3.6332,
+      "step": 3683
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001776424541572605,
+      "loss": 3.6648,
+      "step": 3684
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017762912449227302,
+      "loss": 3.6293,
+      "step": 3685
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017761579135529032,
+      "loss": 3.6395,
+      "step": 3686
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017760245474690868,
+      "loss": 3.6438,
+      "step": 3687
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017758911466772467,
+      "loss": 3.6567,
+      "step": 3688
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017757577111833487,
+      "loss": 3.6336,
+      "step": 3689
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017756242409933609,
+      "loss": 3.653,
+      "step": 3690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017754907361132528,
+      "loss": 3.6241,
+      "step": 3691
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017753571965489954,
+      "loss": 3.6301,
+      "step": 3692
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017752236223065617,
+      "loss": 3.6282,
+      "step": 3693
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017750900133919255,
+      "loss": 3.6388,
+      "step": 3694
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017749563698110624,
+      "loss": 3.638,
+      "step": 3695
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017748226915699502,
+      "loss": 3.6339,
+      "step": 3696
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001774688978674567,
+      "loss": 3.6146,
+      "step": 3697
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017745552311308939,
+      "loss": 3.6199,
+      "step": 3698
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017744214489449122,
+      "loss": 3.6294,
+      "step": 3699
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017742876321226057,
+      "loss": 3.6539,
+      "step": 3700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017741537806699594,
+      "loss": 3.6383,
+      "step": 3701
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017740198945929598,
+      "loss": 3.6574,
+      "step": 3702
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017738859738975952,
+      "loss": 3.6404,
+      "step": 3703
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017737520185898546,
+      "loss": 3.6396,
+      "step": 3704
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017736180286757303,
+      "loss": 3.6244,
+      "step": 3705
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001773484004161214,
+      "loss": 3.6238,
+      "step": 3706
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017733499450523006,
+      "loss": 3.652,
+      "step": 3707
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017732158513549857,
+      "loss": 3.6432,
+      "step": 3708
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001773081723075267,
+      "loss": 3.6254,
+      "step": 3709
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017729475602191425,
+      "loss": 3.6197,
+      "step": 3710
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001772813362792614,
+      "loss": 3.6275,
+      "step": 3711
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001772679130801683,
+      "loss": 3.6505,
+      "step": 3712
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017725448642523528,
+      "loss": 3.6446,
+      "step": 3713
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001772410563150629,
+      "loss": 3.6263,
+      "step": 3714
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001772276227502518,
+      "loss": 3.6308,
+      "step": 3715
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017721418573140278,
+      "loss": 3.6208,
+      "step": 3716
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017720074525911687,
+      "loss": 3.6267,
+      "step": 3717
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017718730133399517,
+      "loss": 3.6184,
+      "step": 3718
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017717385395663894,
+      "loss": 3.6244,
+      "step": 3719
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001771604031276497,
+      "loss": 3.6401,
+      "step": 3720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017714694884762895,
+      "loss": 3.6188,
+      "step": 3721
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017713349111717847,
+      "loss": 3.6569,
+      "step": 3722
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001771200299369002,
+      "loss": 3.653,
+      "step": 3723
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017710656530739617,
+      "loss": 3.6229,
+      "step": 3724
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001770930972292686,
+      "loss": 3.6744,
+      "step": 3725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001770796257031198,
+      "loss": 3.6217,
+      "step": 3726
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017706615072955238,
+      "loss": 3.6364,
+      "step": 3727
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017705267230916895,
+      "loss": 3.6198,
+      "step": 3728
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017703919044257237,
+      "loss": 3.6382,
+      "step": 3729
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017702570513036562,
+      "loss": 3.6398,
+      "step": 3730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017701221637315178,
+      "loss": 3.6299,
+      "step": 3731
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001769987241715342,
+      "loss": 3.6177,
+      "step": 3732
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017698522852611635,
+      "loss": 3.6379,
+      "step": 3733
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017697172943750174,
+      "loss": 3.6204,
+      "step": 3734
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001769582269062942,
+      "loss": 3.6273,
+      "step": 3735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017694472093309762,
+      "loss": 3.6456,
+      "step": 3736
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017693121151851604,
+      "loss": 3.6233,
+      "step": 3737
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017691769866315366,
+      "loss": 3.645,
+      "step": 3738
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017690418236761488,
+      "loss": 3.6395,
+      "step": 3739
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017689066263250423,
+      "loss": 3.6374,
+      "step": 3740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017687713945842634,
+      "loss": 3.6483,
+      "step": 3741
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001768636128459861,
+      "loss": 3.6175,
+      "step": 3742
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017685008279578844,
+      "loss": 3.6399,
+      "step": 3743
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001768365493084385,
+      "loss": 3.6268,
+      "step": 3744
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017682301238454163,
+      "loss": 3.6295,
+      "step": 3745
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017680947202470323,
+      "loss": 3.6418,
+      "step": 3746
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017679592822952888,
+      "loss": 3.6387,
+      "step": 3747
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017678238099962438,
+      "loss": 3.6469,
+      "step": 3748
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017676883033559558,
+      "loss": 3.6284,
+      "step": 3749
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001767552762380486,
+      "loss": 3.6164,
+      "step": 3750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001767417187075896,
+      "loss": 3.6308,
+      "step": 3751
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017672815774482498,
+      "loss": 3.6308,
+      "step": 3752
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017671459335036124,
+      "loss": 3.6278,
+      "step": 3753
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017670102552480506,
+      "loss": 3.6286,
+      "step": 3754
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017668745426876328,
+      "loss": 3.6322,
+      "step": 3755
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017667387958284286,
+      "loss": 3.6382,
+      "step": 3756
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017666030146765092,
+      "loss": 3.6482,
+      "step": 3757
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001766467199237948,
+      "loss": 3.6409,
+      "step": 3758
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001766331349518819,
+      "loss": 3.6311,
+      "step": 3759
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017661954655251983,
+      "loss": 3.6548,
+      "step": 3760
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017660595472631635,
+      "loss": 3.6411,
+      "step": 3761
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017659235947387929,
+      "loss": 3.639,
+      "step": 3762
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001765787607958168,
+      "loss": 3.632,
+      "step": 3763
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017656515869273699,
+      "loss": 3.6376,
+      "step": 3764
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001765515531652483,
+      "loss": 3.6373,
+      "step": 3765
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017653794421395918,
+      "loss": 3.638,
+      "step": 3766
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017652433183947837,
+      "loss": 3.6534,
+      "step": 3767
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017651071604241463,
+      "loss": 3.6368,
+      "step": 3768
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017649709682337697,
+      "loss": 3.6241,
+      "step": 3769
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017648347418297445,
+      "loss": 3.6526,
+      "step": 3770
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001764698481218164,
+      "loss": 3.6207,
+      "step": 3771
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017645621864051226,
+      "loss": 3.6354,
+      "step": 3772
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001764425857396716,
+      "loss": 3.6276,
+      "step": 3773
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017642894941990417,
+      "loss": 3.6248,
+      "step": 3774
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017641530968181982,
+      "loss": 3.6295,
+      "step": 3775
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017640166652602862,
+      "loss": 3.6542,
+      "step": 3776
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017638801995314082,
+      "loss": 3.6146,
+      "step": 3777
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017637436996376663,
+      "loss": 3.6343,
+      "step": 3778
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001763607165585167,
+      "loss": 3.6407,
+      "step": 3779
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017634705973800158,
+      "loss": 3.6422,
+      "step": 3780
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017633339950283214,
+      "loss": 3.6268,
+      "step": 3781
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001763197358536193,
+      "loss": 3.6283,
+      "step": 3782
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001763060687909742,
+      "loss": 3.6171,
+      "step": 3783
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001762923983155081,
+      "loss": 3.6287,
+      "step": 3784
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001762787244278324,
+      "loss": 3.6201,
+      "step": 3785
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001762650471285587,
+      "loss": 3.6207,
+      "step": 3786
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017625136641829866,
+      "loss": 3.619,
+      "step": 3787
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017623768229766424,
+      "loss": 3.6409,
+      "step": 3788
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017622399476726743,
+      "loss": 3.6595,
+      "step": 3789
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017621030382772038,
+      "loss": 3.6373,
+      "step": 3790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017619660947963545,
+      "loss": 3.6447,
+      "step": 3791
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017618291172362516,
+      "loss": 3.6466,
+      "step": 3792
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001761692105603021,
+      "loss": 3.647,
+      "step": 3793
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017615550599027908,
+      "loss": 3.6552,
+      "step": 3794
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017614179801416903,
+      "loss": 3.6488,
+      "step": 3795
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017612808663258506,
+      "loss": 3.6211,
+      "step": 3796
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017611437184614044,
+      "loss": 3.6511,
+      "step": 3797
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001761006536554485,
+      "loss": 3.6147,
+      "step": 3798
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017608693206112283,
+      "loss": 3.6285,
+      "step": 3799
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017607320706377716,
+      "loss": 3.6383,
+      "step": 3800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017605947866402533,
+      "loss": 3.6309,
+      "step": 3801
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001760457468624813,
+      "loss": 3.6276,
+      "step": 3802
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017603201165975932,
+      "loss": 3.6364,
+      "step": 3803
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001760182730564736,
+      "loss": 3.6448,
+      "step": 3804
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001760045310532387,
+      "loss": 3.6517,
+      "step": 3805
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017599078565066917,
+      "loss": 3.6297,
+      "step": 3806
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017597703684937982,
+      "loss": 3.6117,
+      "step": 3807
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017596328464998556,
+      "loss": 3.6119,
+      "step": 3808
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017594952905310144,
+      "loss": 3.6243,
+      "step": 3809
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017593577005934274,
+      "loss": 3.6262,
+      "step": 3810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017592200766932477,
+      "loss": 3.6376,
+      "step": 3811
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001759082418836631,
+      "loss": 3.6561,
+      "step": 3812
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001758944727029734,
+      "loss": 3.6179,
+      "step": 3813
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017588070012787148,
+      "loss": 3.6148,
+      "step": 3814
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017586692415897336,
+      "loss": 3.6125,
+      "step": 3815
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017585314479689516,
+      "loss": 3.6408,
+      "step": 3816
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001758393620422532,
+      "loss": 3.6272,
+      "step": 3817
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017582557589566387,
+      "loss": 3.6471,
+      "step": 3818
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017581178635774376,
+      "loss": 3.6416,
+      "step": 3819
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017579799342910964,
+      "loss": 3.6472,
+      "step": 3820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001757841971103784,
+      "loss": 3.6252,
+      "step": 3821
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017577039740216712,
+      "loss": 3.6371,
+      "step": 3822
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001757565943050929,
+      "loss": 3.6254,
+      "step": 3823
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001757427878197732,
+      "loss": 3.633,
+      "step": 3824
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017572897794682546,
+      "loss": 3.6268,
+      "step": 3825
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017571516468686734,
+      "loss": 3.6307,
+      "step": 3826
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017570134804051666,
+      "loss": 3.6321,
+      "step": 3827
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017568752800839135,
+      "loss": 3.6273,
+      "step": 3828
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017567370459110952,
+      "loss": 3.6288,
+      "step": 3829
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017565987778928946,
+      "loss": 3.6305,
+      "step": 3830
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017564604760354955,
+      "loss": 3.6442,
+      "step": 3831
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017563221403450835,
+      "loss": 3.6272,
+      "step": 3832
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017561837708278462,
+      "loss": 3.6235,
+      "step": 3833
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017560453674899713,
+      "loss": 3.6401,
+      "step": 3834
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017559069303376497,
+      "loss": 3.6557,
+      "step": 3835
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017557684593770732,
+      "loss": 3.6204,
+      "step": 3836
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001755629954614434,
+      "loss": 3.6379,
+      "step": 3837
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017554914160559278,
+      "loss": 3.6332,
+      "step": 3838
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017553528437077505,
+      "loss": 3.6346,
+      "step": 3839
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017552142375760994,
+      "loss": 3.6351,
+      "step": 3840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017550755976671742,
+      "loss": 3.628,
+      "step": 3841
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017549369239871755,
+      "loss": 3.6386,
+      "step": 3842
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017547982165423053,
+      "loss": 3.6354,
+      "step": 3843
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017546594753387677,
+      "loss": 3.6442,
+      "step": 3844
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017545207003827676,
+      "loss": 3.622,
+      "step": 3845
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017543818916805124,
+      "loss": 3.6442,
+      "step": 3846
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017542430492382098,
+      "loss": 3.6288,
+      "step": 3847
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017541041730620695,
+      "loss": 3.6347,
+      "step": 3848
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017539652631583033,
+      "loss": 3.6403,
+      "step": 3849
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017538263195331235,
+      "loss": 3.6455,
+      "step": 3850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017536873421927445,
+      "loss": 3.6374,
+      "step": 3851
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017535483311433827,
+      "loss": 3.6238,
+      "step": 3852
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001753409286391255,
+      "loss": 3.6335,
+      "step": 3853
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017532702079425797,
+      "loss": 3.6313,
+      "step": 3854
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017531310958035783,
+      "loss": 3.6537,
+      "step": 3855
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017529919499804713,
+      "loss": 3.6496,
+      "step": 3856
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017528527704794834,
+      "loss": 3.6474,
+      "step": 3857
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017527135573068388,
+      "loss": 3.6126,
+      "step": 3858
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017525743104687637,
+      "loss": 3.6199,
+      "step": 3859
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001752435029971486,
+      "loss": 3.6171,
+      "step": 3860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017522957158212356,
+      "loss": 3.6309,
+      "step": 3861
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001752156368024243,
+      "loss": 3.6349,
+      "step": 3862
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017520169865867406,
+      "loss": 3.6336,
+      "step": 3863
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001751877571514962,
+      "loss": 3.6398,
+      "step": 3864
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017517381228151436,
+      "loss": 3.6223,
+      "step": 3865
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001751598640493521,
+      "loss": 3.6109,
+      "step": 3866
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017514591245563335,
+      "loss": 3.6313,
+      "step": 3867
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017513195750098207,
+      "loss": 3.6431,
+      "step": 3868
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017511799918602238,
+      "loss": 3.6032,
+      "step": 3869
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017510403751137862,
+      "loss": 3.6035,
+      "step": 3870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001750900724776752,
+      "loss": 3.6325,
+      "step": 3871
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017507610408553674,
+      "loss": 3.6224,
+      "step": 3872
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017506213233558796,
+      "loss": 3.6396,
+      "step": 3873
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017504815722845372,
+      "loss": 3.6336,
+      "step": 3874
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017503417876475914,
+      "loss": 3.6489,
+      "step": 3875
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017502019694512936,
+      "loss": 3.6296,
+      "step": 3876
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017500621177018972,
+      "loss": 3.6245,
+      "step": 3877
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017499222324056574,
+      "loss": 3.6144,
+      "step": 3878
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017497823135688304,
+      "loss": 3.6399,
+      "step": 3879
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017496423611976743,
+      "loss": 3.6154,
+      "step": 3880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001749502375298448,
+      "loss": 3.6312,
+      "step": 3881
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017493623558774134,
+      "loss": 3.6121,
+      "step": 3882
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017492223029408325,
+      "loss": 3.6367,
+      "step": 3883
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017490822164949687,
+      "loss": 3.6423,
+      "step": 3884
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001748942096546088,
+      "loss": 3.6429,
+      "step": 3885
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001748801943100457,
+      "loss": 3.6009,
+      "step": 3886
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017486617561643446,
+      "loss": 3.6385,
+      "step": 3887
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000174852153574402,
+      "loss": 3.6292,
+      "step": 3888
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017483812818457555,
+      "loss": 3.6364,
+      "step": 3889
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017482409944758232,
+      "loss": 3.6233,
+      "step": 3890
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017481006736404982,
+      "loss": 3.6409,
+      "step": 3891
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017479603193460558,
+      "loss": 3.6258,
+      "step": 3892
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017478199315987738,
+      "loss": 3.6278,
+      "step": 3893
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001747679510404931,
+      "loss": 3.6137,
+      "step": 3894
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017475390557708076,
+      "loss": 3.6185,
+      "step": 3895
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001747398567702686,
+      "loss": 3.6139,
+      "step": 3896
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017472580462068486,
+      "loss": 3.6547,
+      "step": 3897
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017471174912895814,
+      "loss": 3.6225,
+      "step": 3898
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017469769029571704,
+      "loss": 3.649,
+      "step": 3899
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017468362812159032,
+      "loss": 3.6174,
+      "step": 3900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017466956260720693,
+      "loss": 3.6342,
+      "step": 3901
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017465549375319598,
+      "loss": 3.6254,
+      "step": 3902
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017464142156018664,
+      "loss": 3.6124,
+      "step": 3903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001746273460288084,
+      "loss": 3.6147,
+      "step": 3904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001746132671596907,
+      "loss": 3.6197,
+      "step": 3905
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017459918495346327,
+      "loss": 3.6203,
+      "step": 3906
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017458509941075594,
+      "loss": 3.6314,
+      "step": 3907
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017457101053219867,
+      "loss": 3.6589,
+      "step": 3908
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001745569183184216,
+      "loss": 3.6213,
+      "step": 3909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017454282277005502,
+      "loss": 3.6352,
+      "step": 3910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017452872388772937,
+      "loss": 3.6164,
+      "step": 3911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001745146216720752,
+      "loss": 3.6362,
+      "step": 3912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017450051612372325,
+      "loss": 3.6195,
+      "step": 3913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017448640724330438,
+      "loss": 3.6392,
+      "step": 3914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017447229503144963,
+      "loss": 3.6413,
+      "step": 3915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001744581794887902,
+      "loss": 3.6304,
+      "step": 3916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017444406061595737,
+      "loss": 3.5989,
+      "step": 3917
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017442993841358266,
+      "loss": 3.62,
+      "step": 3918
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017441581288229764,
+      "loss": 3.645,
+      "step": 3919
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017440168402273412,
+      "loss": 3.6145,
+      "step": 3920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017438755183552397,
+      "loss": 3.6355,
+      "step": 3921
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017437341632129934,
+      "loss": 3.6315,
+      "step": 3922
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017435927748069236,
+      "loss": 3.6133,
+      "step": 3923
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017434513531433546,
+      "loss": 3.6175,
+      "step": 3924
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017433098982286112,
+      "loss": 3.6135,
+      "step": 3925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000174316841006902,
+      "loss": 3.6176,
+      "step": 3926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001743026888670909,
+      "loss": 3.6181,
+      "step": 3927
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017428853340406084,
+      "loss": 3.6122,
+      "step": 3928
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017427437461844489,
+      "loss": 3.621,
+      "step": 3929
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001742602125108763,
+      "loss": 3.6223,
+      "step": 3930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017424604708198848,
+      "loss": 3.6037,
+      "step": 3931
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000174231878332415,
+      "loss": 3.6077,
+      "step": 3932
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017421770626278957,
+      "loss": 3.6511,
+      "step": 3933
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000174203530873746,
+      "loss": 3.6413,
+      "step": 3934
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017418935216591833,
+      "loss": 3.6202,
+      "step": 3935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001741751701399407,
+      "loss": 3.6319,
+      "step": 3936
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017416098479644738,
+      "loss": 3.6053,
+      "step": 3937
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017414679613607286,
+      "loss": 3.6278,
+      "step": 3938
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001741326041594517,
+      "loss": 3.6394,
+      "step": 3939
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017411840886721866,
+      "loss": 3.6159,
+      "step": 3940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001741042102600086,
+      "loss": 3.6141,
+      "step": 3941
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017409000833845662,
+      "loss": 3.6403,
+      "step": 3942
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017407580310319784,
+      "loss": 3.6172,
+      "step": 3943
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017406159455486765,
+      "loss": 3.6231,
+      "step": 3944
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017404738269410146,
+      "loss": 3.6316,
+      "step": 3945
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017403316752153498,
+      "loss": 3.6185,
+      "step": 3946
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017401894903780398,
+      "loss": 3.6245,
+      "step": 3947
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017400472724354434,
+      "loss": 3.6179,
+      "step": 3948
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017399050213939217,
+      "loss": 3.6295,
+      "step": 3949
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001739762737259837,
+      "loss": 3.624,
+      "step": 3950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017396204200395524,
+      "loss": 3.6128,
+      "step": 3951
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001739478069739434,
+      "loss": 3.6112,
+      "step": 3952
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017393356863658482,
+      "loss": 3.6175,
+      "step": 3953
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017391932699251626,
+      "loss": 3.6276,
+      "step": 3954
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017390508204237473,
+      "loss": 3.6406,
+      "step": 3955
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017389083378679737,
+      "loss": 3.6254,
+      "step": 3956
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017387658222642135,
+      "loss": 3.6191,
+      "step": 3957
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017386232736188418,
+      "loss": 3.6448,
+      "step": 3958
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017384806919382335,
+      "loss": 3.6137,
+      "step": 3959
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001738338077228766,
+      "loss": 3.6274,
+      "step": 3960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001738195429496817,
+      "loss": 3.6203,
+      "step": 3961
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017380527487487676,
+      "loss": 3.6351,
+      "step": 3962
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017379100349909984,
+      "loss": 3.6264,
+      "step": 3963
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001737767288229893,
+      "loss": 3.644,
+      "step": 3964
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017376245084718353,
+      "loss": 3.6129,
+      "step": 3965
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017374816957232113,
+      "loss": 3.6168,
+      "step": 3966
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017373388499904087,
+      "loss": 3.5981,
+      "step": 3967
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017371959712798155,
+      "loss": 3.596,
+      "step": 3968
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017370530595978232,
+      "loss": 3.6153,
+      "step": 3969
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017369101149508224,
+      "loss": 3.6026,
+      "step": 3970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017367671373452073,
+      "loss": 3.5993,
+      "step": 3971
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017366241267873723,
+      "loss": 3.6286,
+      "step": 3972
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001736481083283713,
+      "loss": 3.6127,
+      "step": 3973
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001736338006840628,
+      "loss": 3.6314,
+      "step": 3974
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001736194897464516,
+      "loss": 3.6274,
+      "step": 3975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017360517551617778,
+      "loss": 3.6221,
+      "step": 3976
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017359085799388156,
+      "loss": 3.6097,
+      "step": 3977
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017357653718020325,
+      "loss": 3.6195,
+      "step": 3978
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017356221307578338,
+      "loss": 3.62,
+      "step": 3979
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017354788568126262,
+      "loss": 3.617,
+      "step": 3980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017353355499728174,
+      "loss": 3.6152,
+      "step": 3981
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017351922102448172,
+      "loss": 3.5942,
+      "step": 3982
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001735048837635036,
+      "loss": 3.6007,
+      "step": 3983
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017349054321498866,
+      "loss": 3.5999,
+      "step": 3984
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001734761993795783,
+      "loss": 3.622,
+      "step": 3985
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017346185225791403,
+      "loss": 3.6163,
+      "step": 3986
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017344750185063754,
+      "loss": 3.6249,
+      "step": 3987
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017343314815839063,
+      "loss": 3.5989,
+      "step": 3988
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001734187911818153,
+      "loss": 3.6229,
+      "step": 3989
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017340443092155367,
+      "loss": 3.636,
+      "step": 3990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017339006737824802,
+      "loss": 3.6078,
+      "step": 3991
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017337570055254075,
+      "loss": 3.6288,
+      "step": 3992
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001733613304450744,
+      "loss": 3.5931,
+      "step": 3993
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017334695705649173,
+      "loss": 3.6097,
+      "step": 3994
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017333258038743557,
+      "loss": 3.6276,
+      "step": 3995
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017331820043854892,
+      "loss": 3.6145,
+      "step": 3996
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017330381721047494,
+      "loss": 3.6012,
+      "step": 3997
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017328943070385688,
+      "loss": 3.6432,
+      "step": 3998
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017327504091933825,
+      "loss": 3.6078,
+      "step": 3999
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001732606478575626,
+      "loss": 3.6472,
+      "step": 4000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017324625151917368,
+      "loss": 3.624,
+      "step": 4001
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017323185190481536,
+      "loss": 3.6121,
+      "step": 4002
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001732174490151317,
+      "loss": 3.6304,
+      "step": 4003
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017320304285076685,
+      "loss": 3.6258,
+      "step": 4004
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001731886334123651,
+      "loss": 3.6192,
+      "step": 4005
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000173174220700571,
+      "loss": 3.6247,
+      "step": 4006
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001731598047160291,
+      "loss": 3.6277,
+      "step": 4007
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017314538545938415,
+      "loss": 3.6244,
+      "step": 4008
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001731309629312811,
+      "loss": 3.5957,
+      "step": 4009
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000173116537132365,
+      "loss": 3.5917,
+      "step": 4010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000173102108063281,
+      "loss": 3.5878,
+      "step": 4011
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017308767572467453,
+      "loss": 3.6202,
+      "step": 4012
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000173073240117191,
+      "loss": 3.6033,
+      "step": 4013
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017305880124147613,
+      "loss": 3.6088,
+      "step": 4014
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001730443590981756,
+      "loss": 3.5877,
+      "step": 4015
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017302991368793545,
+      "loss": 3.6035,
+      "step": 4016
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017301546501140172,
+      "loss": 3.6066,
+      "step": 4017
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017300101306922058,
+      "loss": 3.6358,
+      "step": 4018
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017298655786203847,
+      "loss": 3.6114,
+      "step": 4019
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017297209939050185,
+      "loss": 3.5963,
+      "step": 4020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017295763765525743,
+      "loss": 3.6046,
+      "step": 4021
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017294317265695197,
+      "loss": 3.6051,
+      "step": 4022
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017292870439623249,
+      "loss": 3.6108,
+      "step": 4023
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000172914232873746,
+      "loss": 3.6166,
+      "step": 4024
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017289975809013978,
+      "loss": 3.5988,
+      "step": 4025
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017288528004606127,
+      "loss": 3.6203,
+      "step": 4026
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017287079874215796,
+      "loss": 3.635,
+      "step": 4027
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017285631417907754,
+      "loss": 3.6126,
+      "step": 4028
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017284182635746785,
+      "loss": 3.6328,
+      "step": 4029
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017282733527797684,
+      "loss": 3.5968,
+      "step": 4030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017281284094125264,
+      "loss": 3.5914,
+      "step": 4031
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017279834334794352,
+      "loss": 3.6239,
+      "step": 4032
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001727838424986979,
+      "loss": 3.6414,
+      "step": 4033
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001727693383941643,
+      "loss": 3.6035,
+      "step": 4034
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017275483103499148,
+      "loss": 3.6252,
+      "step": 4035
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017274032042182823,
+      "loss": 3.5998,
+      "step": 4036
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017272580655532357,
+      "loss": 3.6342,
+      "step": 4037
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017271128943612666,
+      "loss": 3.6213,
+      "step": 4038
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017269676906488677,
+      "loss": 3.6317,
+      "step": 4039
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017268224544225328,
+      "loss": 3.6201,
+      "step": 4040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017266771856887583,
+      "loss": 3.5939,
+      "step": 4041
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017265318844540413,
+      "loss": 3.5904,
+      "step": 4042
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000172638655072488,
+      "loss": 3.6241,
+      "step": 4043
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017262411845077747,
+      "loss": 3.6408,
+      "step": 4044
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017260957858092278,
+      "loss": 3.6196,
+      "step": 4045
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001725950354635741,
+      "loss": 3.5941,
+      "step": 4046
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017258048909938198,
+      "loss": 3.603,
+      "step": 4047
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017256593948899695,
+      "loss": 3.6182,
+      "step": 4048
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017255138663306978,
+      "loss": 3.6139,
+      "step": 4049
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001725368305322513,
+      "loss": 3.5956,
+      "step": 4050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017252227118719264,
+      "loss": 3.5922,
+      "step": 4051
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001725077085985449,
+      "loss": 3.6027,
+      "step": 4052
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017249314276695938,
+      "loss": 3.5861,
+      "step": 4053
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001724785736930876,
+      "loss": 3.6254,
+      "step": 4054
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017246400137758115,
+      "loss": 3.6159,
+      "step": 4055
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017244942582109177,
+      "loss": 3.6028,
+      "step": 4056
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017243484702427132,
+      "loss": 3.5997,
+      "step": 4057
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017242026498777194,
+      "loss": 3.6046,
+      "step": 4058
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001724056797122457,
+      "loss": 3.5952,
+      "step": 4059
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017239109119834506,
+      "loss": 3.6062,
+      "step": 4060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017237649944672241,
+      "loss": 3.5923,
+      "step": 4061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001723619044580304,
+      "loss": 3.5962,
+      "step": 4062
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017234730623292177,
+      "loss": 3.6279,
+      "step": 4063
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017233270477204945,
+      "loss": 3.6108,
+      "step": 4064
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017231810007606652,
+      "loss": 3.6097,
+      "step": 4065
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017230349214562615,
+      "loss": 3.5976,
+      "step": 4066
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017228888098138172,
+      "loss": 3.6008,
+      "step": 4067
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017227426658398667,
+      "loss": 3.5837,
+      "step": 4068
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017225964895409466,
+      "loss": 3.6069,
+      "step": 4069
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017224502809235948,
+      "loss": 3.6245,
+      "step": 4070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017223040399943508,
+      "loss": 3.6139,
+      "step": 4071
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017221577667597546,
+      "loss": 3.5663,
+      "step": 4072
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017220114612263488,
+      "loss": 3.6167,
+      "step": 4073
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017218651234006768,
+      "loss": 3.6022,
+      "step": 4074
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017217187532892836,
+      "loss": 3.6151,
+      "step": 4075
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001721572350898716,
+      "loss": 3.6261,
+      "step": 4076
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017214259162355212,
+      "loss": 3.6224,
+      "step": 4077
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017212794493062496,
+      "loss": 3.6063,
+      "step": 4078
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001721132950117451,
+      "loss": 3.5857,
+      "step": 4079
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001720986418675678,
+      "loss": 3.5887,
+      "step": 4080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017208398549874846,
+      "loss": 3.5875,
+      "step": 4081
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017206932590594257,
+      "loss": 3.6118,
+      "step": 4082
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017205466308980576,
+      "loss": 3.6064,
+      "step": 4083
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017203999705099382,
+      "loss": 3.6034,
+      "step": 4084
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017202532779016276,
+      "loss": 3.5944,
+      "step": 4085
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017201065530796864,
+      "loss": 3.5985,
+      "step": 4086
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017199597960506767,
+      "loss": 3.6145,
+      "step": 4087
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017198130068211626,
+      "loss": 3.5952,
+      "step": 4088
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017196661853977093,
+      "loss": 3.5863,
+      "step": 4089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001719519331786883,
+      "loss": 3.6176,
+      "step": 4090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017193724459952522,
+      "loss": 3.5874,
+      "step": 4091
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017192255280293862,
+      "loss": 3.6056,
+      "step": 4092
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017190785778958562,
+      "loss": 3.6084,
+      "step": 4093
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017189315956012347,
+      "loss": 3.5913,
+      "step": 4094
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017187845811520951,
+      "loss": 3.626,
+      "step": 4095
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001718637534555013,
+      "loss": 3.5818,
+      "step": 4096
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017184904558165652,
+      "loss": 3.6058,
+      "step": 4097
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017183433449433294,
+      "loss": 3.5943,
+      "step": 4098
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017181962019418857,
+      "loss": 3.6011,
+      "step": 4099
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017180490268188148,
+      "loss": 3.6222,
+      "step": 4100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017179018195806996,
+      "loss": 3.5808,
+      "step": 4101
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017177545802341232,
+      "loss": 3.597,
+      "step": 4102
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001717607308785672,
+      "loss": 3.6072,
+      "step": 4103
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017174600052419318,
+      "loss": 3.5974,
+      "step": 4104
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017173126696094917,
+      "loss": 3.6207,
+      "step": 4105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017171653018949404,
+      "loss": 3.5803,
+      "step": 4106
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017170179021048699,
+      "loss": 3.608,
+      "step": 4107
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001716870470245872,
+      "loss": 3.5809,
+      "step": 4108
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001716723006324541,
+      "loss": 3.594,
+      "step": 4109
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001716575510347472,
+      "loss": 3.5894,
+      "step": 4110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017164279823212627,
+      "loss": 3.5838,
+      "step": 4111
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017162804222525102,
+      "loss": 3.6049,
+      "step": 4112
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017161328301478146,
+      "loss": 3.5989,
+      "step": 4113
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017159852060137773,
+      "loss": 3.6137,
+      "step": 4114
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017158375498570007,
+      "loss": 3.6133,
+      "step": 4115
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017156898616840885,
+      "loss": 3.6009,
+      "step": 4116
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017155421415016467,
+      "loss": 3.59,
+      "step": 4117
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017153943893162817,
+      "loss": 3.5915,
+      "step": 4118
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017152466051346016,
+      "loss": 3.5743,
+      "step": 4119
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001715098788963217,
+      "loss": 3.5693,
+      "step": 4120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017149509408087379,
+      "loss": 3.6068,
+      "step": 4121
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017148030606777776,
+      "loss": 3.6053,
+      "step": 4122
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000171465514857695,
+      "loss": 3.604,
+      "step": 4123
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000171450720451287,
+      "loss": 3.6029,
+      "step": 4124
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017143592284921553,
+      "loss": 3.6008,
+      "step": 4125
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017142112205214237,
+      "loss": 3.6025,
+      "step": 4126
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017140631806072948,
+      "loss": 3.6075,
+      "step": 4127
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017139151087563903,
+      "loss": 3.6022,
+      "step": 4128
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017137670049753324,
+      "loss": 3.5703,
+      "step": 4129
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017136188692707448,
+      "loss": 3.6002,
+      "step": 4130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017134707016492535,
+      "loss": 3.5904,
+      "step": 4131
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017133225021174852,
+      "loss": 3.6182,
+      "step": 4132
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001713174270682068,
+      "loss": 3.5999,
+      "step": 4133
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017130260073496315,
+      "loss": 3.5905,
+      "step": 4134
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017128777121268077,
+      "loss": 3.5897,
+      "step": 4135
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017127293850202277,
+      "loss": 3.5967,
+      "step": 4136
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017125810260365269,
+      "loss": 3.569,
+      "step": 4137
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000171243263518234,
+      "loss": 3.6313,
+      "step": 4138
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001712284212464304,
+      "loss": 3.5955,
+      "step": 4139
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017121357578890568,
+      "loss": 3.6107,
+      "step": 4140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017119872714632388,
+      "loss": 3.606,
+      "step": 4141
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017118387531934908,
+      "loss": 3.6073,
+      "step": 4142
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001711690203086455,
+      "loss": 3.5902,
+      "step": 4143
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017115416211487761,
+      "loss": 3.5986,
+      "step": 4144
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017113930073870987,
+      "loss": 3.5929,
+      "step": 4145
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000171124436180807,
+      "loss": 3.5962,
+      "step": 4146
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017110956844183382,
+      "loss": 3.6043,
+      "step": 4147
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017109469752245535,
+      "loss": 3.5941,
+      "step": 4148
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017107982342333658,
+      "loss": 3.602,
+      "step": 4149
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017106494614514282,
+      "loss": 3.5965,
+      "step": 4150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017105006568853953,
+      "loss": 3.6014,
+      "step": 4151
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017103518205419213,
+      "loss": 3.5977,
+      "step": 4152
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017102029524276638,
+      "loss": 3.5874,
+      "step": 4153
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017100540525492807,
+      "loss": 3.5939,
+      "step": 4154
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017099051209134314,
+      "loss": 3.5639,
+      "step": 4155
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001709756157526777,
+      "loss": 3.5823,
+      "step": 4156
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017096071623959802,
+      "loss": 3.6195,
+      "step": 4157
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001709458135527705,
+      "loss": 3.5855,
+      "step": 4158
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017093090769286157,
+      "loss": 3.6008,
+      "step": 4159
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017091599866053805,
+      "loss": 3.6005,
+      "step": 4160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017090108645646663,
+      "loss": 3.5859,
+      "step": 4161
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017088617108131435,
+      "loss": 3.5915,
+      "step": 4162
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017087125253574824,
+      "loss": 3.6004,
+      "step": 4163
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017085633082043557,
+      "loss": 3.611,
+      "step": 4164
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017084140593604373,
+      "loss": 3.6089,
+      "step": 4165
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001708264778832402,
+      "loss": 3.6085,
+      "step": 4166
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001708115466626927,
+      "loss": 3.6107,
+      "step": 4167
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000170796612275069,
+      "loss": 3.5762,
+      "step": 4168
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017078167472103702,
+      "loss": 3.589,
+      "step": 4169
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017076673400126492,
+      "loss": 3.5983,
+      "step": 4170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017075179011642088,
+      "loss": 3.6104,
+      "step": 4171
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017073684306717328,
+      "loss": 3.5761,
+      "step": 4172
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017072189285419063,
+      "loss": 3.5888,
+      "step": 4173
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001707069394781416,
+      "loss": 3.5794,
+      "step": 4174
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017069198293969497,
+      "loss": 3.5964,
+      "step": 4175
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001706770232395197,
+      "loss": 3.6026,
+      "step": 4176
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017066206037828484,
+      "loss": 3.575,
+      "step": 4177
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017064709435665962,
+      "loss": 3.6056,
+      "step": 4178
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001706321251753134,
+      "loss": 3.6381,
+      "step": 4179
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017061715283491567,
+      "loss": 3.5734,
+      "step": 4180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017060217733613615,
+      "loss": 3.6022,
+      "step": 4181
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017058719867964453,
+      "loss": 3.5784,
+      "step": 4182
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017057221686611076,
+      "loss": 3.5559,
+      "step": 4183
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017055723189620496,
+      "loss": 3.5959,
+      "step": 4184
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017054224377059728,
+      "loss": 3.5938,
+      "step": 4185
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001705272524899581,
+      "loss": 3.5739,
+      "step": 4186
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017051225805495788,
+      "loss": 3.6,
+      "step": 4187
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001704972604662673,
+      "loss": 3.5631,
+      "step": 4188
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017048225972455709,
+      "loss": 3.6164,
+      "step": 4189
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001704672558304982,
+      "loss": 3.5735,
+      "step": 4190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017045224878476162,
+      "loss": 3.5849,
+      "step": 4191
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017043723858801863,
+      "loss": 3.5959,
+      "step": 4192
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017042222524094053,
+      "loss": 3.593,
+      "step": 4193
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017040720874419884,
+      "loss": 3.578,
+      "step": 4194
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017039218909846508,
+      "loss": 3.5937,
+      "step": 4195
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001703771663044111,
+      "loss": 3.563,
+      "step": 4196
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017036214036270875,
+      "loss": 3.5826,
+      "step": 4197
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001703471112740301,
+      "loss": 3.5894,
+      "step": 4198
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017033207903904734,
+      "loss": 3.6008,
+      "step": 4199
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017031704365843276,
+      "loss": 3.5803,
+      "step": 4200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017030200513285884,
+      "loss": 3.5717,
+      "step": 4201
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001702869634629982,
+      "loss": 3.5693,
+      "step": 4202
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017027191864952355,
+      "loss": 3.6093,
+      "step": 4203
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017025687069310784,
+      "loss": 3.5967,
+      "step": 4204
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000170241819594424,
+      "loss": 3.5698,
+      "step": 4205
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017022676535414526,
+      "loss": 3.5727,
+      "step": 4206
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017021170797294492,
+      "loss": 3.5824,
+      "step": 4207
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017019664745149645,
+      "loss": 3.5955,
+      "step": 4208
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017018158379047336,
+      "loss": 3.5932,
+      "step": 4209
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001701665169905495,
+      "loss": 3.6015,
+      "step": 4210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001701514470523986,
+      "loss": 3.5784,
+      "step": 4211
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017013637397669478,
+      "loss": 3.6082,
+      "step": 4212
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001701212977641121,
+      "loss": 3.5964,
+      "step": 4213
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017010621841532494,
+      "loss": 3.5557,
+      "step": 4214
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001700911359310077,
+      "loss": 3.5816,
+      "step": 4215
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017007605031183493,
+      "loss": 3.5806,
+      "step": 4216
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017006096155848134,
+      "loss": 3.5845,
+      "step": 4217
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001700458696716218,
+      "loss": 3.5936,
+      "step": 4218
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017003077465193128,
+      "loss": 3.585,
+      "step": 4219
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017001567650008493,
+      "loss": 3.5959,
+      "step": 4220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017000057521675797,
+      "loss": 3.5675,
+      "step": 4221
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001699854708026259,
+      "loss": 3.5939,
+      "step": 4222
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016997036325836423,
+      "loss": 3.5706,
+      "step": 4223
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016995525258464863,
+      "loss": 3.5721,
+      "step": 4224
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016994013878215495,
+      "loss": 3.5692,
+      "step": 4225
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016992502185155916,
+      "loss": 3.5717,
+      "step": 4226
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016990990179353738,
+      "loss": 3.5777,
+      "step": 4227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016989477860876586,
+      "loss": 3.5991,
+      "step": 4228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016987965229792095,
+      "loss": 3.5649,
+      "step": 4229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016986452286167923,
+      "loss": 3.5949,
+      "step": 4230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016984939030071736,
+      "loss": 3.584,
+      "step": 4231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016983425461571217,
+      "loss": 3.598,
+      "step": 4232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016981911580734053,
+      "loss": 3.5896,
+      "step": 4233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016980397387627961,
+      "loss": 3.5906,
+      "step": 4234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001697888288232066,
+      "loss": 3.5852,
+      "step": 4235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001697736806487989,
+      "loss": 3.567,
+      "step": 4236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000169758529353734,
+      "loss": 3.575,
+      "step": 4237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016974337493868956,
+      "loss": 3.6001,
+      "step": 4238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001697282174043433,
+      "loss": 3.5982,
+      "step": 4239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016971305675137323,
+      "loss": 3.595,
+      "step": 4240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016969789298045739,
+      "loss": 3.5761,
+      "step": 4241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000169682726092274,
+      "loss": 3.5745,
+      "step": 4242
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001696675560875014,
+      "loss": 3.5913,
+      "step": 4243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016965238296681804,
+      "loss": 3.5839,
+      "step": 4244
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016963720673090257,
+      "loss": 3.5797,
+      "step": 4245
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016962202738043377,
+      "loss": 3.5661,
+      "step": 4246
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001696068449160905,
+      "loss": 3.5997,
+      "step": 4247
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016959165933855185,
+      "loss": 3.5685,
+      "step": 4248
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016957647064849699,
+      "loss": 3.5657,
+      "step": 4249
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001695612788466052,
+      "loss": 3.6222,
+      "step": 4250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016954608393355598,
+      "loss": 3.589,
+      "step": 4251
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016953088591002891,
+      "loss": 3.593,
+      "step": 4252
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016951568477670373,
+      "loss": 3.5872,
+      "step": 4253
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001695004805342604,
+      "loss": 3.6085,
+      "step": 4254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016948527318337879,
+      "loss": 3.5902,
+      "step": 4255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001694700627247391,
+      "loss": 3.5745,
+      "step": 4256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016945484915902172,
+      "loss": 3.59,
+      "step": 4257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016943963248690697,
+      "loss": 3.5543,
+      "step": 4258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001694244127090755,
+      "loss": 3.5627,
+      "step": 4259
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016940918982620792,
+      "loss": 3.5754,
+      "step": 4260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016939396383898518,
+      "loss": 3.592,
+      "step": 4261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016937873474808822,
+      "loss": 3.5804,
+      "step": 4262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001693635025541982,
+      "loss": 3.5863,
+      "step": 4263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016934826725799638,
+      "loss": 3.5743,
+      "step": 4264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016933302886016412,
+      "loss": 3.5792,
+      "step": 4265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000169317787361383,
+      "loss": 3.5981,
+      "step": 4266
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016930254276233475,
+      "loss": 3.5717,
+      "step": 4267
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001692872950637011,
+      "loss": 3.5767,
+      "step": 4268
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016927204426616402,
+      "loss": 3.5642,
+      "step": 4269
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016925679037040569,
+      "loss": 3.5707,
+      "step": 4270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016924153337710826,
+      "loss": 3.5727,
+      "step": 4271
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016922627328695415,
+      "loss": 3.5717,
+      "step": 4272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016921101010062588,
+      "loss": 3.5895,
+      "step": 4273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001691957438188061,
+      "loss": 3.6009,
+      "step": 4274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016918047444217755,
+      "loss": 3.6044,
+      "step": 4275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001691652019714232,
+      "loss": 3.5792,
+      "step": 4276
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016914992640722614,
+      "loss": 3.5918,
+      "step": 4277
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001691346477502695,
+      "loss": 3.5886,
+      "step": 4278
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016911936600123671,
+      "loss": 3.5908,
+      "step": 4279
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016910408116081123,
+      "loss": 3.5874,
+      "step": 4280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016908879322967665,
+      "loss": 3.5589,
+      "step": 4281
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016907350220851677,
+      "loss": 3.5854,
+      "step": 4282
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016905820809801542,
+      "loss": 3.5929,
+      "step": 4283
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016904291089885672,
+      "loss": 3.5906,
+      "step": 4284
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001690276106117248,
+      "loss": 3.5868,
+      "step": 4285
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016901230723730394,
+      "loss": 3.5713,
+      "step": 4286
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016899700077627867,
+      "loss": 3.6019,
+      "step": 4287
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016898169122933352,
+      "loss": 3.5633,
+      "step": 4288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016896637859715324,
+      "loss": 3.5972,
+      "step": 4289
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016895106288042266,
+      "loss": 3.5744,
+      "step": 4290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001689357440798268,
+      "loss": 3.5816,
+      "step": 4291
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016892042219605085,
+      "loss": 3.5874,
+      "step": 4292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016890509722978,
+      "loss": 3.5914,
+      "step": 4293
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016888976918169975,
+      "loss": 3.5809,
+      "step": 4294
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016887443805249555,
+      "loss": 3.5882,
+      "step": 4295
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001688591038428532,
+      "loss": 3.5979,
+      "step": 4296
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016884376655345847,
+      "loss": 3.5737,
+      "step": 4297
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016882842618499733,
+      "loss": 3.585,
+      "step": 4298
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001688130827381559,
+      "loss": 3.5868,
+      "step": 4299
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001687977362136204,
+      "loss": 3.5989,
+      "step": 4300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016878238661207723,
+      "loss": 3.5834,
+      "step": 4301
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016876703393421295,
+      "loss": 3.6003,
+      "step": 4302
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016875167818071409,
+      "loss": 3.5563,
+      "step": 4303
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016873631935226756,
+      "loss": 3.5839,
+      "step": 4304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016872095744956028,
+      "loss": 3.5947,
+      "step": 4305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016870559247327924,
+      "loss": 3.5657,
+      "step": 4306
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001686902244241117,
+      "loss": 3.5752,
+      "step": 4307
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000168674853302745,
+      "loss": 3.5816,
+      "step": 4308
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016865947910986662,
+      "loss": 3.5902,
+      "step": 4309
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001686441018461642,
+      "loss": 3.5887,
+      "step": 4310
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001686287215123254,
+      "loss": 3.5675,
+      "step": 4311
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016861333810903824,
+      "loss": 3.589,
+      "step": 4312
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016859795163699066,
+      "loss": 3.6019,
+      "step": 4313
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016858256209687086,
+      "loss": 3.5748,
+      "step": 4314
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016856716948936712,
+      "loss": 3.5828,
+      "step": 4315
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016855177381516794,
+      "loss": 3.5519,
+      "step": 4316
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016853637507496184,
+      "loss": 3.5743,
+      "step": 4317
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016852097326943753,
+      "loss": 3.5872,
+      "step": 4318
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016850556839928386,
+      "loss": 3.5853,
+      "step": 4319
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001684901604651899,
+      "loss": 3.5985,
+      "step": 4320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001684747494678447,
+      "loss": 3.5692,
+      "step": 4321
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016845933540793754,
+      "loss": 3.5703,
+      "step": 4322
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001684439182861578,
+      "loss": 3.5658,
+      "step": 4323
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001684284981031951,
+      "loss": 3.5714,
+      "step": 4324
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000168413074859739,
+      "loss": 3.6068,
+      "step": 4325
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016839764855647942,
+      "loss": 3.5614,
+      "step": 4326
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016838221919410619,
+      "loss": 3.5885,
+      "step": 4327
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016836678677330951,
+      "loss": 3.561,
+      "step": 4328
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016835135129477952,
+      "loss": 3.5857,
+      "step": 4329
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001683359127592066,
+      "loss": 3.5872,
+      "step": 4330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016832047116728126,
+      "loss": 3.5841,
+      "step": 4331
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016830502651969415,
+      "loss": 3.5885,
+      "step": 4332
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000168289578817136,
+      "loss": 3.5975,
+      "step": 4333
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016827412806029774,
+      "loss": 3.5866,
+      "step": 4334
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001682586742498704,
+      "loss": 3.5947,
+      "step": 4335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016824321738654515,
+      "loss": 3.5986,
+      "step": 4336
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016822775747101328,
+      "loss": 3.5664,
+      "step": 4337
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001682122945039663,
+      "loss": 3.6,
+      "step": 4338
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016819682848609574,
+      "loss": 3.5729,
+      "step": 4339
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001681813594180934,
+      "loss": 3.5674,
+      "step": 4340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016816588730065107,
+      "loss": 3.5737,
+      "step": 4341
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001681504121344608,
+      "loss": 3.5819,
+      "step": 4342
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001681349339202147,
+      "loss": 3.5883,
+      "step": 4343
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000168119452658605,
+      "loss": 3.5728,
+      "step": 4344
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016810396835032417,
+      "loss": 3.572,
+      "step": 4345
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016808848099606472,
+      "loss": 3.5685,
+      "step": 4346
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016807299059651932,
+      "loss": 3.5905,
+      "step": 4347
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001680574971523808,
+      "loss": 3.5978,
+      "step": 4348
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016804200066434212,
+      "loss": 3.578,
+      "step": 4349
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016802650113309636,
+      "loss": 3.5608,
+      "step": 4350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001680109985593367,
+      "loss": 3.5808,
+      "step": 4351
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016799549294375655,
+      "loss": 3.5782,
+      "step": 4352
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016797998428704945,
+      "loss": 3.6019,
+      "step": 4353
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001679644725899089,
+      "loss": 3.575,
+      "step": 4354
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001679489578530288,
+      "loss": 3.565,
+      "step": 4355
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016793344007710295,
+      "loss": 3.5658,
+      "step": 4356
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016791791926282546,
+      "loss": 3.6091,
+      "step": 4357
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016790239541089046,
+      "loss": 3.5678,
+      "step": 4358
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016788686852199227,
+      "loss": 3.5773,
+      "step": 4359
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016787133859682535,
+      "loss": 3.5363,
+      "step": 4360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001678558056360843,
+      "loss": 3.5705,
+      "step": 4361
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001678402696404638,
+      "loss": 3.5745,
+      "step": 4362
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016782473061065874,
+      "loss": 3.5619,
+      "step": 4363
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00016780918854736407,
+      "loss": 3.6015,
+      "step": 4364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016779364345127494,
+      "loss": 3.6281,
+      "step": 4365
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016777809532308663,
+      "loss": 3.5842,
+      "step": 4366
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016776254416349447,
+      "loss": 3.624,
+      "step": 4367
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016774698997319405,
+      "loss": 3.5882,
+      "step": 4368
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000167731432752881,
+      "loss": 3.5547,
+      "step": 4369
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016771587250325118,
+      "loss": 3.5908,
+      "step": 4370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016770030922500046,
+      "loss": 3.5812,
+      "step": 4371
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016768474291882499,
+      "loss": 3.5647,
+      "step": 4372
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016766917358542092,
+      "loss": 3.6326,
+      "step": 4373
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001676536012254846,
+      "loss": 3.5772,
+      "step": 4374
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016763802583971254,
+      "loss": 3.5616,
+      "step": 4375
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016762244742880134,
+      "loss": 3.5657,
+      "step": 4376
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016760686599344774,
+      "loss": 3.5737,
+      "step": 4377
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016759128153434865,
+      "loss": 3.5984,
+      "step": 4378
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016757569405220107,
+      "loss": 3.5911,
+      "step": 4379
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016756010354770215,
+      "loss": 3.5588,
+      "step": 4380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001675445100215492,
+      "loss": 3.5568,
+      "step": 4381
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016752891347443966,
+      "loss": 3.5816,
+      "step": 4382
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016751331390707107,
+      "loss": 3.5703,
+      "step": 4383
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016749771132014115,
+      "loss": 3.5681,
+      "step": 4384
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016748210571434767,
+      "loss": 3.5699,
+      "step": 4385
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016746649709038867,
+      "loss": 3.5846,
+      "step": 4386
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001674508854489622,
+      "loss": 3.5598,
+      "step": 4387
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016743527079076659,
+      "loss": 3.5838,
+      "step": 4388
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016741965311650007,
+      "loss": 3.5616,
+      "step": 4389
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016740403242686126,
+      "loss": 3.5486,
+      "step": 4390
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00016738840872254876,
+      "loss": 3.5648,
+      "step": 4391
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001673727820042613,
+      "loss": 3.6231,
+      "step": 4392
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001673571522726979,
+      "loss": 3.7122,
+      "step": 4393
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001673415195285575,
+      "loss": 3.6611,
+      "step": 4394
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001673258837725394,
+      "loss": 3.5855,
+      "step": 4395
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001673102450053428,
+      "loss": 3.6263,
+      "step": 4396
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016729460322766716,
+      "loss": 3.6314,
+      "step": 4397
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016727895844021216,
+      "loss": 3.7147,
+      "step": 4398
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016726331064367744,
+      "loss": 3.6443,
+      "step": 4399
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016724765983876285,
+      "loss": 3.6279,
+      "step": 4400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016723200602616843,
+      "loss": 3.699,
+      "step": 4401
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016721634920659426,
+      "loss": 3.5908,
+      "step": 4402
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016720068938074062,
+      "loss": 3.6559,
+      "step": 4403
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001671850265493079,
+      "loss": 3.6273,
+      "step": 4404
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016716936071299663,
+      "loss": 3.6234,
+      "step": 4405
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016715369187250744,
+      "loss": 3.6043,
+      "step": 4406
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016713802002854116,
+      "loss": 3.6155,
+      "step": 4407
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001671223451817987,
+      "loss": 3.599,
+      "step": 4408
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016710666733298117,
+      "loss": 3.6013,
+      "step": 4409
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001670909864827897,
+      "loss": 3.5922,
+      "step": 4410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016707530263192566,
+      "loss": 3.5935,
+      "step": 4411
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001670596157810905,
+      "loss": 3.5841,
+      "step": 4412
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016704392593098582,
+      "loss": 3.6385,
+      "step": 4413
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001670282330823134,
+      "loss": 3.6014,
+      "step": 4414
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001670125372357751,
+      "loss": 3.6089,
+      "step": 4415
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016699683839207285,
+      "loss": 3.6337,
+      "step": 4416
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016698113655190885,
+      "loss": 3.5962,
+      "step": 4417
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016696543171598535,
+      "loss": 3.5983,
+      "step": 4418
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016694972388500479,
+      "loss": 3.6004,
+      "step": 4419
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016693401305966968,
+      "loss": 3.5893,
+      "step": 4420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016691829924068266,
+      "loss": 3.5935,
+      "step": 4421
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016690258242874658,
+      "loss": 3.5834,
+      "step": 4422
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001668868626245644,
+      "loss": 3.6035,
+      "step": 4423
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016687113982883918,
+      "loss": 3.6095,
+      "step": 4424
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00016685541404227409,
+      "loss": 3.5948,
+      "step": 4425
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001668396852655725,
+      "loss": 3.6129,
+      "step": 4426
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016682395349943793,
+      "loss": 3.6181,
+      "step": 4427
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016680821874457392,
+      "loss": 3.5896,
+      "step": 4428
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016679248100168423,
+      "loss": 3.5845,
+      "step": 4429
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016677674027147275,
+      "loss": 3.5926,
+      "step": 4430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016676099655464348,
+      "loss": 3.5832,
+      "step": 4431
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016674524985190058,
+      "loss": 3.5901,
+      "step": 4432
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016672950016394833,
+      "loss": 3.5776,
+      "step": 4433
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016671374749149112,
+      "loss": 3.5824,
+      "step": 4434
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001666979918352335,
+      "loss": 3.6075,
+      "step": 4435
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001666822331958802,
+      "loss": 3.6093,
+      "step": 4436
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016666647157413596,
+      "loss": 3.5886,
+      "step": 4437
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016665070697070573,
+      "loss": 3.5849,
+      "step": 4438
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001666349393862946,
+      "loss": 3.5985,
+      "step": 4439
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016661916882160783,
+      "loss": 3.5721,
+      "step": 4440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016660339527735073,
+      "loss": 3.5694,
+      "step": 4441
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016658761875422876,
+      "loss": 3.5991,
+      "step": 4442
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016657183925294753,
+      "loss": 3.6052,
+      "step": 4443
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016655605677421285,
+      "loss": 3.6088,
+      "step": 4444
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016654027131873052,
+      "loss": 3.5848,
+      "step": 4445
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016652448288720657,
+      "loss": 3.5868,
+      "step": 4446
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016650869148034718,
+      "loss": 3.5934,
+      "step": 4447
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016649289709885857,
+      "loss": 3.6033,
+      "step": 4448
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016647709974344722,
+      "loss": 3.5673,
+      "step": 4449
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016646129941481962,
+      "loss": 3.5794,
+      "step": 4450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016644549611368245,
+      "loss": 3.5991,
+      "step": 4451
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016642968984074256,
+      "loss": 3.593,
+      "step": 4452
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016641388059670685,
+      "loss": 3.5594,
+      "step": 4453
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016639806838228242,
+      "loss": 3.5737,
+      "step": 4454
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016638225319817648,
+      "loss": 3.5851,
+      "step": 4455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00016636643504509631,
+      "loss": 3.5907,
+      "step": 4456
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016635061392374943,
+      "loss": 3.6051,
+      "step": 4457
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001663347898348435,
+      "loss": 3.5701,
+      "step": 4458
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016631896277908618,
+      "loss": 3.6013,
+      "step": 4459
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001663031327571854,
+      "loss": 3.5978,
+      "step": 4460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016628729976984907,
+      "loss": 3.6001,
+      "step": 4461
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016627146381778546,
+      "loss": 3.6053,
+      "step": 4462
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001662556249017027,
+      "loss": 3.5905,
+      "step": 4463
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016623978302230934,
+      "loss": 3.5728,
+      "step": 4464
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016622393818031378,
+      "loss": 3.5815,
+      "step": 4465
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016620809037642476,
+      "loss": 3.5702,
+      "step": 4466
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016619223961135106,
+      "loss": 3.5721,
+      "step": 4467
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001661763858858016,
+      "loss": 3.5779,
+      "step": 4468
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016616052920048548,
+      "loss": 3.5647,
+      "step": 4469
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016614466955611188,
+      "loss": 3.5639,
+      "step": 4470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016612880695339013,
+      "loss": 3.5514,
+      "step": 4471
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016611294139302966,
+      "loss": 3.5607,
+      "step": 4472
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016609707287574014,
+      "loss": 3.5622,
+      "step": 4473
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016608120140223123,
+      "loss": 3.5675,
+      "step": 4474
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016606532697321276,
+      "loss": 3.5938,
+      "step": 4475
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001660494495893948,
+      "loss": 3.5528,
+      "step": 4476
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016603356925148747,
+      "loss": 3.5329,
+      "step": 4477
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000166017685960201,
+      "loss": 3.543,
+      "step": 4478
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001660017997162457,
+      "loss": 3.6745,
+      "step": 4479
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001659859105203322,
+      "loss": 3.6014,
+      "step": 4480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016597001837317113,
+      "loss": 3.5484,
+      "step": 4481
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016595412327547322,
+      "loss": 3.5405,
+      "step": 4482
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016593822522794943,
+      "loss": 3.556,
+      "step": 4483
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016592232423131081,
+      "loss": 3.5079,
+      "step": 4484
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016590642028626852,
+      "loss": 3.5421,
+      "step": 4485
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00016589051339353387,
+      "loss": 3.4636,
+      "step": 4486
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001658746035538183,
+      "loss": 3.5281,
+      "step": 4487
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001658586907678334,
+      "loss": 3.5252,
+      "step": 4488
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001658427750362909,
+      "loss": 3.3596,
+      "step": 4489
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016582685635990253,
+      "loss": 3.3503,
+      "step": 4490
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016581093473938042,
+      "loss": 3.2967,
+      "step": 4491
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001657950101754365,
+      "loss": 3.3424,
+      "step": 4492
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016577908266878315,
+      "loss": 3.3234,
+      "step": 4493
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001657631522201327,
+      "loss": 3.4101,
+      "step": 4494
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016574721883019756,
+      "loss": 3.5165,
+      "step": 4495
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016573128249969043,
+      "loss": 3.2259,
+      "step": 4496
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016571534322932404,
+      "loss": 3.3537,
+      "step": 4497
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016569940101981133,
+      "loss": 3.3455,
+      "step": 4498
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016568345587186526,
+      "loss": 3.2798,
+      "step": 4499
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016566750778619906,
+      "loss": 3.3639,
+      "step": 4500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016565155676352594,
+      "loss": 3.2723,
+      "step": 4501
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016563560280455933,
+      "loss": 3.2003,
+      "step": 4502
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016561964591001283,
+      "loss": 3.2522,
+      "step": 4503
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016560368608060005,
+      "loss": 3.2607,
+      "step": 4504
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016558772331703482,
+      "loss": 3.2199,
+      "step": 4505
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001655717576200311,
+      "loss": 3.209,
+      "step": 4506
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016555578899030295,
+      "loss": 3.2179,
+      "step": 4507
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001655398174285646,
+      "loss": 3.2284,
+      "step": 4508
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016552384293553034,
+      "loss": 3.2046,
+      "step": 4509
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016550786551191467,
+      "loss": 3.1694,
+      "step": 4510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016549188515843218,
+      "loss": 3.1744,
+      "step": 4511
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016547590187579756,
+      "loss": 3.1971,
+      "step": 4512
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016545991566472573,
+      "loss": 3.1928,
+      "step": 4513
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016544392652593167,
+      "loss": 3.1763,
+      "step": 4514
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016542793446013045,
+      "loss": 3.1357,
+      "step": 4515
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00016541193946803737,
+      "loss": 3.1748,
+      "step": 4516
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001653959415503678,
+      "loss": 3.1491,
+      "step": 4517
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016537994070783723,
+      "loss": 3.1641,
+      "step": 4518
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016536393694116134,
+      "loss": 3.1461,
+      "step": 4519
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016534793025105592,
+      "loss": 3.1502,
+      "step": 4520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001653319206382368,
+      "loss": 3.1389,
+      "step": 4521
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016531590810342007,
+      "loss": 3.1483,
+      "step": 4522
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001652998926473219,
+      "loss": 3.1385,
+      "step": 4523
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016528387427065862,
+      "loss": 3.1677,
+      "step": 4524
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016526785297414658,
+      "loss": 3.1161,
+      "step": 4525
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016525182875850236,
+      "loss": 3.1171,
+      "step": 4526
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016523580162444267,
+      "loss": 3.1138,
+      "step": 4527
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016521977157268436,
+      "loss": 3.1171,
+      "step": 4528
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001652037386039443,
+      "loss": 3.1224,
+      "step": 4529
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016518770271893967,
+      "loss": 3.1061,
+      "step": 4530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016517166391838761,
+      "loss": 3.0869,
+      "step": 4531
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016515562220300545,
+      "loss": 3.0843,
+      "step": 4532
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016513957757351074,
+      "loss": 3.1011,
+      "step": 4533
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016512353003062105,
+      "loss": 3.0826,
+      "step": 4534
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016510747957505407,
+      "loss": 3.086,
+      "step": 4535
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016509142620752768,
+      "loss": 3.1003,
+      "step": 4536
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016507536992875995,
+      "loss": 3.0953,
+      "step": 4537
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016505931073946888,
+      "loss": 3.0788,
+      "step": 4538
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001650432486403728,
+      "loss": 3.0915,
+      "step": 4539
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016502718363219007,
+      "loss": 3.0866,
+      "step": 4540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016501111571563923,
+      "loss": 3.0822,
+      "step": 4541
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016499504489143893,
+      "loss": 3.0674,
+      "step": 4542
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016497897116030788,
+      "loss": 3.0711,
+      "step": 4543
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016496289452296508,
+      "loss": 3.0474,
+      "step": 4544
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016494681498012946,
+      "loss": 3.0405,
+      "step": 4545
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016493073253252027,
+      "loss": 3.0602,
+      "step": 4546
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016491464718085673,
+      "loss": 3.0607,
+      "step": 4547
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016489855892585836,
+      "loss": 3.0388,
+      "step": 4548
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016488246776824464,
+      "loss": 3.0219,
+      "step": 4549
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016486637370873527,
+      "loss": 3.0496,
+      "step": 4550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016485027674805004,
+      "loss": 3.0364,
+      "step": 4551
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016483417688690892,
+      "loss": 3.0228,
+      "step": 4552
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000164818074126032,
+      "loss": 2.9991,
+      "step": 4553
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001648019684661395,
+      "loss": 3.0114,
+      "step": 4554
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016478585990795165,
+      "loss": 3.0067,
+      "step": 4555
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016476974845218903,
+      "loss": 2.9832,
+      "step": 4556
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016475363409957215,
+      "loss": 2.9895,
+      "step": 4557
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001647375168508218,
+      "loss": 3.0044,
+      "step": 4558
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016472139670665876,
+      "loss": 3.0065,
+      "step": 4559
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016470527366780404,
+      "loss": 2.9925,
+      "step": 4560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016468914773497875,
+      "loss": 2.9668,
+      "step": 4561
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016467301890890415,
+      "loss": 2.9717,
+      "step": 4562
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001646568871903016,
+      "loss": 2.9645,
+      "step": 4563
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016464075257989258,
+      "loss": 2.9289,
+      "step": 4564
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001646246150783987,
+      "loss": 2.9339,
+      "step": 4565
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016460847468654176,
+      "loss": 2.9462,
+      "step": 4566
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016459233140504362,
+      "loss": 2.9138,
+      "step": 4567
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016457618523462632,
+      "loss": 2.9316,
+      "step": 4568
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016456003617601196,
+      "loss": 2.9058,
+      "step": 4569
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016454388422992286,
+      "loss": 2.9325,
+      "step": 4570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016452772939708142,
+      "loss": 2.9237,
+      "step": 4571
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016451157167821014,
+      "loss": 2.9117,
+      "step": 4572
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016449541107403168,
+      "loss": 2.8966,
+      "step": 4573
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016447924758526886,
+      "loss": 2.8871,
+      "step": 4574
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001644630812126446,
+      "loss": 2.8721,
+      "step": 4575
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016444691195688192,
+      "loss": 2.8634,
+      "step": 4576
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000164430739818704,
+      "loss": 2.8686,
+      "step": 4577
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016441456479883418,
+      "loss": 2.863,
+      "step": 4578
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016439838689799583,
+      "loss": 2.8417,
+      "step": 4579
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016438220611691255,
+      "loss": 2.8554,
+      "step": 4580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016436602245630807,
+      "loss": 2.8409,
+      "step": 4581
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001643498359169062,
+      "loss": 2.8215,
+      "step": 4582
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001643336464994308,
+      "loss": 2.8146,
+      "step": 4583
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016431745420460603,
+      "loss": 2.8003,
+      "step": 4584
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001643012590331561,
+      "loss": 2.7693,
+      "step": 4585
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016428506098580532,
+      "loss": 2.7894,
+      "step": 4586
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016426886006327816,
+      "loss": 2.7764,
+      "step": 4587
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001642526562662992,
+      "loss": 2.7565,
+      "step": 4588
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001642364495955932,
+      "loss": 2.7844,
+      "step": 4589
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000164220240051885,
+      "loss": 2.7587,
+      "step": 4590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016420402763589955,
+      "loss": 2.7128,
+      "step": 4591
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016418781234836196,
+      "loss": 2.7001,
+      "step": 4592
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016417159418999745,
+      "loss": 2.7444,
+      "step": 4593
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016415537316153146,
+      "loss": 2.7976,
+      "step": 4594
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016413914926368946,
+      "loss": 2.7426,
+      "step": 4595
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016412292249719698,
+      "loss": 2.6758,
+      "step": 4596
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016410669286277986,
+      "loss": 2.6934,
+      "step": 4597
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016409046036116394,
+      "loss": 2.6197,
+      "step": 4598
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016407422499307525,
+      "loss": 2.622,
+      "step": 4599
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016405798675923992,
+      "loss": 2.7212,
+      "step": 4600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016404174566038418,
+      "loss": 3.4241,
+      "step": 4601
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016402550169723442,
+      "loss": 3.3946,
+      "step": 4602
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016400925487051725,
+      "loss": 3.4429,
+      "step": 4603
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001639930051809592,
+      "loss": 3.389,
+      "step": 4604
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001639767526292871,
+      "loss": 4.335,
+      "step": 4605
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016396049721622785,
+      "loss": 3.3793,
+      "step": 4606
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016394423894250845,
+      "loss": 3.7642,
+      "step": 4607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001639279778088561,
+      "loss": 3.3729,
+      "step": 4608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016391171381599805,
+      "loss": 3.9017,
+      "step": 4609
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016389544696466174,
+      "loss": 3.2018,
+      "step": 4610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001638791772555747,
+      "loss": 3.3507,
+      "step": 4611
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001638629046894646,
+      "loss": 3.2632,
+      "step": 4612
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016384662926705924,
+      "loss": 3.4194,
+      "step": 4613
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016383035098908657,
+      "loss": 3.1757,
+      "step": 4614
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001638140698562746,
+      "loss": 3.1636,
+      "step": 4615
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016379778586935153,
+      "loss": 3.3051,
+      "step": 4616
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016378149902904565,
+      "loss": 3.1997,
+      "step": 4617
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001637652093360854,
+      "loss": 3.0937,
+      "step": 4618
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001637489167911994,
+      "loss": 3.0984,
+      "step": 4619
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016373262139511628,
+      "loss": 3.081,
+      "step": 4620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016371632314856488,
+      "loss": 3.0645,
+      "step": 4621
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016370002205227412,
+      "loss": 3.0775,
+      "step": 4622
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016368371810697308,
+      "loss": 3.0325,
+      "step": 4623
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000163667411313391,
+      "loss": 3.0051,
+      "step": 4624
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016365110167225718,
+      "loss": 3.026,
+      "step": 4625
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001636347891843011,
+      "loss": 3.0199,
+      "step": 4626
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016361847385025228,
+      "loss": 3.0042,
+      "step": 4627
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001636021556708405,
+      "loss": 2.9736,
+      "step": 4628
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016358583464679555,
+      "loss": 2.9564,
+      "step": 4629
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016356951077884743,
+      "loss": 2.9352,
+      "step": 4630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001635531840677262,
+      "loss": 2.9327,
+      "step": 4631
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001635368545141621,
+      "loss": 2.9228,
+      "step": 4632
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001635205221188855,
+      "loss": 2.8764,
+      "step": 4633
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001635041868826268,
+      "loss": 2.8851,
+      "step": 4634
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016348784880611668,
+      "loss": 2.8802,
+      "step": 4635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001634715078900858,
+      "loss": 2.8876,
+      "step": 4636
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016345516413526507,
+      "loss": 2.8316,
+      "step": 4637
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016343881754238542,
+      "loss": 2.8408,
+      "step": 4638
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016342246811217802,
+      "loss": 2.8213,
+      "step": 4639
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016340611584537404,
+      "loss": 2.8098,
+      "step": 4640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001633897607427049,
+      "loss": 2.8051,
+      "step": 4641
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016337340280490206,
+      "loss": 2.7937,
+      "step": 4642
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001633570420326971,
+      "loss": 2.7614,
+      "step": 4643
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016334067842682185,
+      "loss": 2.741,
+      "step": 4644
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016332431198800808,
+      "loss": 2.7501,
+      "step": 4645
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016330794271698785,
+      "loss": 2.7131,
+      "step": 4646
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001632915706144933,
+      "loss": 2.7013,
+      "step": 4647
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001632751956812566,
+      "loss": 2.7016,
+      "step": 4648
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001632588179180102,
+      "loss": 2.6898,
+      "step": 4649
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016324243732548655,
+      "loss": 2.6749,
+      "step": 4650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016322605390441832,
+      "loss": 2.6531,
+      "step": 4651
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016320966765553827,
+      "loss": 2.6324,
+      "step": 4652
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016319327857957925,
+      "loss": 2.6125,
+      "step": 4653
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016317688667727426,
+      "loss": 2.5763,
+      "step": 4654
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016316049194935647,
+      "loss": 2.5829,
+      "step": 4655
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016314409439655911,
+      "loss": 2.5729,
+      "step": 4656
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001631276940196156,
+      "loss": 2.5267,
+      "step": 4657
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001631112908192594,
+      "loss": 2.4851,
+      "step": 4658
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016309488479622424,
+      "loss": 2.4753,
+      "step": 4659
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016307847595124376,
+      "loss": 2.4587,
+      "step": 4660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000163062064285052,
+      "loss": 2.4506,
+      "step": 4661
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016304564979838286,
+      "loss": 2.4179,
+      "step": 4662
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016302923249197053,
+      "loss": 2.3813,
+      "step": 4663
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001630128123665493,
+      "loss": 2.3469,
+      "step": 4664
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016299638942285352,
+      "loss": 2.3498,
+      "step": 4665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016297996366161776,
+      "loss": 2.3684,
+      "step": 4666
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016296353508357661,
+      "loss": 2.2973,
+      "step": 4667
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016294710368946493,
+      "loss": 2.2688,
+      "step": 4668
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016293066948001755,
+      "loss": 2.2811,
+      "step": 4669
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016291423245596955,
+      "loss": 2.2326,
+      "step": 4670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016289779261805602,
+      "loss": 2.191,
+      "step": 4671
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016288134996701227,
+      "loss": 2.1937,
+      "step": 4672
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016286490450357372,
+      "loss": 2.1492,
+      "step": 4673
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016284845622847592,
+      "loss": 2.1007,
+      "step": 4674
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016283200514245446,
+      "loss": 2.1436,
+      "step": 4675
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016281555124624518,
+      "loss": 2.0587,
+      "step": 4676
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001627990945405839,
+      "loss": 2.0658,
+      "step": 4677
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016278263502620676,
+      "loss": 2.0624,
+      "step": 4678
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016276617270384987,
+      "loss": 2.8624,
+      "step": 4679
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016274970757424953,
+      "loss": 5.6045,
+      "step": 4680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016273323963814212,
+      "loss": 4.8294,
+      "step": 4681
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016271676889626418,
+      "loss": 4.3782,
+      "step": 4682
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016270029534935242,
+      "loss": 3.5223,
+      "step": 4683
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016268381899814358,
+      "loss": 3.0787,
+      "step": 4684
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001626673398433746,
+      "loss": 3.3574,
+      "step": 4685
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016265085788578248,
+      "loss": 3.4203,
+      "step": 4686
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016263437312610438,
+      "loss": 2.9845,
+      "step": 4687
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016261788556507764,
+      "loss": 2.7585,
+      "step": 4688
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016260139520343962,
+      "loss": 2.6132,
+      "step": 4689
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001625849020419279,
+      "loss": 2.737,
+      "step": 4690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016256840608128013,
+      "loss": 2.828,
+      "step": 4691
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001625519073222341,
+      "loss": 3.1699,
+      "step": 4692
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016253540576552768,
+      "loss": 3.0695,
+      "step": 4693
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016251890141189899,
+      "loss": 2.7196,
+      "step": 4694
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016250239426208614,
+      "loss": 3.0027,
+      "step": 4695
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016248588431682742,
+      "loss": 2.968,
+      "step": 4696
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016246937157686123,
+      "loss": 2.92,
+      "step": 4697
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016245285604292618,
+      "loss": 2.9245,
+      "step": 4698
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016243633771576086,
+      "loss": 2.6826,
+      "step": 4699
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001624198165961041,
+      "loss": 2.7777,
+      "step": 4700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001624032926846948,
+      "loss": 2.7154,
+      "step": 4701
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016238676598227198,
+      "loss": 2.8193,
+      "step": 4702
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016237023648957486,
+      "loss": 2.6361,
+      "step": 4703
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016235370420734268,
+      "loss": 2.6437,
+      "step": 4704
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016233716913631487,
+      "loss": 2.4743,
+      "step": 4705
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016232063127723093,
+      "loss": 2.4874,
+      "step": 4706
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016230409063083058,
+      "loss": 2.419,
+      "step": 4707
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001622875471978536,
+      "loss": 2.4376,
+      "step": 4708
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016227100097903984,
+      "loss": 2.3134,
+      "step": 4709
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001622544519751294,
+      "loss": 2.3361,
+      "step": 4710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016223790018686244,
+      "loss": 2.282,
+      "step": 4711
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016222134561497923,
+      "loss": 2.263,
+      "step": 4712
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016220478826022017,
+      "loss": 2.2793,
+      "step": 4713
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016218822812332583,
+      "loss": 2.2546,
+      "step": 4714
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016217166520503682,
+      "loss": 2.2409,
+      "step": 4715
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001621550995060939,
+      "loss": 2.2455,
+      "step": 4716
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001621385310272381,
+      "loss": 2.1548,
+      "step": 4717
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016212195976921033,
+      "loss": 2.1797,
+      "step": 4718
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016210538573275182,
+      "loss": 2.133,
+      "step": 4719
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001620888089186038,
+      "loss": 2.0877,
+      "step": 4720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001620722293275077,
+      "loss": 2.1103,
+      "step": 4721
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001620556469602051,
+      "loss": 2.0506,
+      "step": 4722
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016203906181743753,
+      "loss": 2.0179,
+      "step": 4723
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016202247389994687,
+      "loss": 2.0303,
+      "step": 4724
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016200588320847497,
+      "loss": 1.9717,
+      "step": 4725
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016198928974376392,
+      "loss": 1.9761,
+      "step": 4726
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016197269350655578,
+      "loss": 1.9901,
+      "step": 4727
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001619560944975929,
+      "loss": 1.9471,
+      "step": 4728
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016193949271761762,
+      "loss": 1.9107,
+      "step": 4729
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016192288816737252,
+      "loss": 1.9584,
+      "step": 4730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001619062808476002,
+      "loss": 1.9172,
+      "step": 4731
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016188967075904344,
+      "loss": 1.8762,
+      "step": 4732
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016187305790244514,
+      "loss": 1.9011,
+      "step": 4733
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016185644227854833,
+      "loss": 1.8929,
+      "step": 4734
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016183982388809609,
+      "loss": 1.8601,
+      "step": 4735
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016182320273183177,
+      "loss": 1.8587,
+      "step": 4736
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001618065788104987,
+      "loss": 1.8998,
+      "step": 4737
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016178995212484041,
+      "loss": 1.8242,
+      "step": 4738
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016177332267560055,
+      "loss": 1.7934,
+      "step": 4739
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016175669046352282,
+      "loss": 1.8255,
+      "step": 4740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016174005548935118,
+      "loss": 1.8245,
+      "step": 4741
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016172341775382959,
+      "loss": 1.7824,
+      "step": 4742
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001617067772577022,
+      "loss": 1.7635,
+      "step": 4743
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016169013400171327,
+      "loss": 1.7492,
+      "step": 4744
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016167348798660719,
+      "loss": 1.757,
+      "step": 4745
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016165683921312836,
+      "loss": 1.7717,
+      "step": 4746
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016164018768202149,
+      "loss": 1.6942,
+      "step": 4747
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016162353339403134,
+      "loss": 1.7315,
+      "step": 4748
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016160687634990276,
+      "loss": 1.6954,
+      "step": 4749
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016159021655038068,
+      "loss": 1.7266,
+      "step": 4750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001615735539962103,
+      "loss": 1.6957,
+      "step": 4751
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016155688868813684,
+      "loss": 1.726,
+      "step": 4752
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016154022062690566,
+      "loss": 1.9142,
+      "step": 4753
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016152354981326225,
+      "loss": 5.9257,
+      "step": 4754
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001615068762479522,
+      "loss": 8.3002,
+      "step": 4755
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016149019993172124,
+      "loss": 8.5642,
+      "step": 4756
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001614735208653152,
+      "loss": 6.3459,
+      "step": 4757
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016145683904948016,
+      "loss": 5.07,
+      "step": 4758
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016144015448496215,
+      "loss": 4.8956,
+      "step": 4759
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016142346717250736,
+      "loss": 4.8995,
+      "step": 4760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001614067771128622,
+      "loss": 5.2644,
+      "step": 4761
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016139008430677315,
+      "loss": 5.1125,
+      "step": 4762
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016137338875498673,
+      "loss": 4.9205,
+      "step": 4763
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016135669045824972,
+      "loss": 4.7913,
+      "step": 4764
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016133998941730891,
+      "loss": 4.6312,
+      "step": 4765
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001613232856329113,
+      "loss": 4.3462,
+      "step": 4766
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016130657910580395,
+      "loss": 4.1544,
+      "step": 4767
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001612898698367341,
+      "loss": 4.0536,
+      "step": 4768
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016127315782644904,
+      "loss": 4.0136,
+      "step": 4769
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016125644307569623,
+      "loss": 4.0636,
+      "step": 4770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016123972558522326,
+      "loss": 4.0476,
+      "step": 4771
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001612230053557778,
+      "loss": 3.9291,
+      "step": 4772
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016120628238810772,
+      "loss": 3.8573,
+      "step": 4773
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001611895566829609,
+      "loss": 3.7645,
+      "step": 4774
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016117282824108545,
+      "loss": 3.6419,
+      "step": 4775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016115609706322955,
+      "loss": 3.5798,
+      "step": 4776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016113936315014147,
+      "loss": 3.5881,
+      "step": 4777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001611226265025697,
+      "loss": 3.5816,
+      "step": 4778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016110588712126272,
+      "loss": 3.5478,
+      "step": 4779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016108914500696927,
+      "loss": 3.5256,
+      "step": 4780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016107240016043813,
+      "loss": 3.4558,
+      "step": 4781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001610556525824182,
+      "loss": 3.441,
+      "step": 4782
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016103890227365858,
+      "loss": 3.4061,
+      "step": 4783
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001610221492349084,
+      "loss": 3.4016,
+      "step": 4784
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001610053934669169,
+      "loss": 3.3753,
+      "step": 4785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016098863497043353,
+      "loss": 3.3414,
+      "step": 4786
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016097187374620783,
+      "loss": 3.3024,
+      "step": 4787
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016095510979498946,
+      "loss": 3.2973,
+      "step": 4788
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016093834311752816,
+      "loss": 3.2341,
+      "step": 4789
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016092157371457385,
+      "loss": 3.2065,
+      "step": 4790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016090480158687656,
+      "loss": 3.2045,
+      "step": 4791
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001608880267351864,
+      "loss": 3.1504,
+      "step": 4792
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016087124916025364,
+      "loss": 3.1522,
+      "step": 4793
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001608544688628287,
+      "loss": 3.1422,
+      "step": 4794
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016083768584366205,
+      "loss": 3.1193,
+      "step": 4795
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016082090010350429,
+      "loss": 3.0922,
+      "step": 4796
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016080411164310623,
+      "loss": 3.0949,
+      "step": 4797
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001607873204632187,
+      "loss": 3.0849,
+      "step": 4798
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001607705265645927,
+      "loss": 3.0697,
+      "step": 4799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016075372994797936,
+      "loss": 3.0822,
+      "step": 4800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016073693061412992,
+      "loss": 3.0714,
+      "step": 4801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016072012856379572,
+      "loss": 3.0294,
+      "step": 4802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016070332379772824,
+      "loss": 3.0341,
+      "step": 4803
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001606865163166791,
+      "loss": 3.0036,
+      "step": 4804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001606697061214,
+      "loss": 3.0115,
+      "step": 4805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016065289321264277,
+      "loss": 2.9686,
+      "step": 4806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016063607759115938,
+      "loss": 2.9419,
+      "step": 4807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016061925925770195,
+      "loss": 2.9862,
+      "step": 4808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016060243821302268,
+      "loss": 2.9511,
+      "step": 4809
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016058561445787386,
+      "loss": 2.9342,
+      "step": 4810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016056878799300796,
+      "loss": 2.916,
+      "step": 4811
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016055195881917755,
+      "loss": 2.8813,
+      "step": 4812
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016053512693713534,
+      "loss": 2.8746,
+      "step": 4813
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016051829234763415,
+      "loss": 2.8706,
+      "step": 4814
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016050145505142687,
+      "loss": 2.887,
+      "step": 4815
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016048461504926658,
+      "loss": 2.8751,
+      "step": 4816
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016046777234190643,
+      "loss": 2.8282,
+      "step": 4817
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016045092693009978,
+      "loss": 2.8477,
+      "step": 4818
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016043407881459996,
+      "loss": 2.8576,
+      "step": 4819
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016041722799616063,
+      "loss": 2.8183,
+      "step": 4820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016040037447553532,
+      "loss": 2.8333,
+      "step": 4821
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016038351825347786,
+      "loss": 2.7982,
+      "step": 4822
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001603666593307422,
+      "loss": 2.8404,
+      "step": 4823
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016034979770808232,
+      "loss": 2.8203,
+      "step": 4824
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016033293338625236,
+      "loss": 2.8006,
+      "step": 4825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016031606636600655,
+      "loss": 2.8039,
+      "step": 4826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001602991966480994,
+      "loss": 2.7688,
+      "step": 4827
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016028232423328525,
+      "loss": 2.763,
+      "step": 4828
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016026544912231883,
+      "loss": 2.8032,
+      "step": 4829
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016024857131595485,
+      "loss": 2.7624,
+      "step": 4830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016023169081494823,
+      "loss": 2.7957,
+      "step": 4831
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001602148076200539,
+      "loss": 2.7228,
+      "step": 4832
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016019792173202694,
+      "loss": 2.783,
+      "step": 4833
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016018103315162269,
+      "loss": 2.7408,
+      "step": 4834
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016016414187959642,
+      "loss": 2.7748,
+      "step": 4835
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016014724791670357,
+      "loss": 2.7194,
+      "step": 4836
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001601303512636998,
+      "loss": 2.7076,
+      "step": 4837
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001601134519213408,
+      "loss": 2.7149,
+      "step": 4838
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016009654989038237,
+      "loss": 2.7292,
+      "step": 4839
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016007964517158048,
+      "loss": 2.7231,
+      "step": 4840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001600627377656912,
+      "loss": 2.6758,
+      "step": 4841
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016004582767347074,
+      "loss": 2.6973,
+      "step": 4842
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016002891489567538,
+      "loss": 2.6886,
+      "step": 4843
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001600119994330616,
+      "loss": 2.6508,
+      "step": 4844
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015999508128638587,
+      "loss": 2.6667,
+      "step": 4845
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015997816045640494,
+      "loss": 2.6836,
+      "step": 4846
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001599612369438755,
+      "loss": 2.6638,
+      "step": 4847
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015994431074955461,
+      "loss": 2.6097,
+      "step": 4848
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001599273818741992,
+      "loss": 2.6743,
+      "step": 4849
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015991045031856643,
+      "loss": 2.6374,
+      "step": 4850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001598935160834136,
+      "loss": 2.618,
+      "step": 4851
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015987657916949804,
+      "loss": 2.6233,
+      "step": 4852
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015985963957757736,
+      "loss": 2.624,
+      "step": 4853
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001598426973084091,
+      "loss": 2.5675,
+      "step": 4854
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015982575236275106,
+      "loss": 2.5803,
+      "step": 4855
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001598088047413611,
+      "loss": 2.5576,
+      "step": 4856
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015979185444499718,
+      "loss": 2.5494,
+      "step": 4857
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015977490147441744,
+      "loss": 2.5602,
+      "step": 4858
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015975794583038012,
+      "loss": 2.5538,
+      "step": 4859
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015974098751364352,
+      "loss": 2.5032,
+      "step": 4860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015972402652496617,
+      "loss": 2.5222,
+      "step": 4861
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001597070628651066,
+      "loss": 2.4938,
+      "step": 4862
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015969009653482355,
+      "loss": 2.495,
+      "step": 4863
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015967312753487588,
+      "loss": 2.4998,
+      "step": 4864
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015965615586602242,
+      "loss": 2.4427,
+      "step": 4865
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015963918152902236,
+      "loss": 2.4624,
+      "step": 4866
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001596222045246348,
+      "loss": 2.4576,
+      "step": 4867
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001596052248536191,
+      "loss": 2.4504,
+      "step": 4868
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015958824251673467,
+      "loss": 2.415,
+      "step": 4869
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000159571257514741,
+      "loss": 2.381,
+      "step": 4870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015955426984839784,
+      "loss": 2.3932,
+      "step": 4871
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015953727951846492,
+      "loss": 2.3767,
+      "step": 4872
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015952028652570214,
+      "loss": 2.3408,
+      "step": 4873
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015950329087086954,
+      "loss": 2.3479,
+      "step": 4874
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015948629255472719,
+      "loss": 2.3137,
+      "step": 4875
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001594692915780354,
+      "loss": 2.3066,
+      "step": 4876
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015945228794155458,
+      "loss": 2.3187,
+      "step": 4877
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015943528164604514,
+      "loss": 2.304,
+      "step": 4878
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015941827269226778,
+      "loss": 2.2602,
+      "step": 4879
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015940126108098318,
+      "loss": 2.2619,
+      "step": 4880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015938424681295223,
+      "loss": 2.2181,
+      "step": 4881
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015936722988893587,
+      "loss": 2.2084,
+      "step": 4882
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001593502103096952,
+      "loss": 2.2091,
+      "step": 4883
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001593331880759914,
+      "loss": 2.3077,
+      "step": 4884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015931616318858583,
+      "loss": 2.2197,
+      "step": 4885
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001592991356482399,
+      "loss": 2.2143,
+      "step": 4886
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015928210545571522,
+      "loss": 2.2857,
+      "step": 4887
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015926507261177346,
+      "loss": 2.7632,
+      "step": 4888
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001592480371171764,
+      "loss": 2.8855,
+      "step": 4889
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015923099897268596,
+      "loss": 2.5521,
+      "step": 4890
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001592139581790642,
+      "loss": 2.5666,
+      "step": 4891
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001591969147370733,
+      "loss": 2.4322,
+      "step": 4892
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015917986864747542,
+      "loss": 2.5287,
+      "step": 4893
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015916281991103313,
+      "loss": 2.518,
+      "step": 4894
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015914576852850879,
+      "loss": 2.291,
+      "step": 4895
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015912871450066512,
+      "loss": 2.3438,
+      "step": 4896
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015911165782826482,
+      "loss": 2.3836,
+      "step": 4897
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015909459851207075,
+      "loss": 2.3936,
+      "step": 4898
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015907753655284594,
+      "loss": 2.3311,
+      "step": 4899
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015906047195135344,
+      "loss": 2.3257,
+      "step": 4900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015904340470835658,
+      "loss": 2.2363,
+      "step": 4901
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015902633482461854,
+      "loss": 2.2529,
+      "step": 4902
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001590092623009029,
+      "loss": 2.2959,
+      "step": 4903
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001589921871379732,
+      "loss": 2.2355,
+      "step": 4904
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001589751093365931,
+      "loss": 2.1704,
+      "step": 4905
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001589580288975265,
+      "loss": 2.1263,
+      "step": 4906
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001589409458215372,
+      "loss": 2.1701,
+      "step": 4907
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001589238601093894,
+      "loss": 2.1729,
+      "step": 4908
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015890677176184708,
+      "loss": 2.0946,
+      "step": 4909
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001588896807796747,
+      "loss": 2.1455,
+      "step": 4910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015887258716363657,
+      "loss": 2.0976,
+      "step": 4911
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015885549091449724,
+      "loss": 2.0603,
+      "step": 4912
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015883839203302132,
+      "loss": 2.0527,
+      "step": 4913
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001588212905199736,
+      "loss": 2.04,
+      "step": 4914
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00015880418637611893,
+      "loss": 2.021,
+      "step": 4915
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001587870796022223,
+      "loss": 2.0158,
+      "step": 4916
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001587699701990488,
+      "loss": 1.9873,
+      "step": 4917
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015875285816736368,
+      "loss": 1.993,
+      "step": 4918
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015873574350793227,
+      "loss": 1.9733,
+      "step": 4919
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015871862622152008,
+      "loss": 1.9788,
+      "step": 4920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015870150630889262,
+      "loss": 1.9497,
+      "step": 4921
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015868438377081564,
+      "loss": 1.9196,
+      "step": 4922
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015866725860805487,
+      "loss": 1.919,
+      "step": 4923
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015865013082137634,
+      "loss": 1.9039,
+      "step": 4924
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015863300041154603,
+      "loss": 1.9184,
+      "step": 4925
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015861586737933014,
+      "loss": 1.896,
+      "step": 4926
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015859873172549492,
+      "loss": 1.8734,
+      "step": 4927
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015858159345080684,
+      "loss": 1.8622,
+      "step": 4928
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015856445255603232,
+      "loss": 1.8298,
+      "step": 4929
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015854730904193805,
+      "loss": 1.837,
+      "step": 4930
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015853016290929075,
+      "loss": 1.8232,
+      "step": 4931
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015851301415885732,
+      "loss": 1.8592,
+      "step": 4932
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015849586279140473,
+      "loss": 1.8424,
+      "step": 4933
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015847870880770012,
+      "loss": 1.865,
+      "step": 4934
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015846155220851062,
+      "loss": 1.8051,
+      "step": 4935
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015844439299460365,
+      "loss": 1.7913,
+      "step": 4936
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015842723116674665,
+      "loss": 1.807,
+      "step": 4937
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015841006672570712,
+      "loss": 1.7944,
+      "step": 4938
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015839289967225283,
+      "loss": 1.7579,
+      "step": 4939
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015837573000715156,
+      "loss": 1.7486,
+      "step": 4940
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015835855773117121,
+      "loss": 1.7397,
+      "step": 4941
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015834138284507985,
+      "loss": 1.7496,
+      "step": 4942
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001583242053496456,
+      "loss": 1.7211,
+      "step": 4943
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00015830702524563677,
+      "loss": 1.7355,
+      "step": 4944
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001582898425338217,
+      "loss": 1.7144,
+      "step": 4945
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001582726572149689,
+      "loss": 1.6851,
+      "step": 4946
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015825546928984702,
+      "loss": 1.7052,
+      "step": 4947
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015823827875922482,
+      "loss": 1.6597,
+      "step": 4948
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001582210856238711,
+      "loss": 1.6817,
+      "step": 4949
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015820388988455481,
+      "loss": 1.682,
+      "step": 4950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015818669154204514,
+      "loss": 1.6508,
+      "step": 4951
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015816949059711116,
+      "loss": 1.6898,
+      "step": 4952
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015815228705052232,
+      "loss": 1.6176,
+      "step": 4953
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015813508090304796,
+      "loss": 1.6375,
+      "step": 4954
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015811787215545768,
+      "loss": 1.622,
+      "step": 4955
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015810066080852112,
+      "loss": 1.6185,
+      "step": 4956
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001580834468630081,
+      "loss": 1.5781,
+      "step": 4957
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015806623031968848,
+      "loss": 1.5919,
+      "step": 4958
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015804901117933234,
+      "loss": 1.5709,
+      "step": 4959
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001580317894427097,
+      "loss": 1.583,
+      "step": 4960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001580145651105909,
+      "loss": 1.5462,
+      "step": 4961
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015799733818374633,
+      "loss": 1.5725,
+      "step": 4962
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001579801086629464,
+      "loss": 1.5945,
+      "step": 4963
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015796287654896177,
+      "loss": 1.5774,
+      "step": 4964
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015794564184256306,
+      "loss": 1.5307,
+      "step": 4965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015792840454452117,
+      "loss": 1.5313,
+      "step": 4966
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015791116465560707,
+      "loss": 1.563,
+      "step": 4967
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015789392217659172,
+      "loss": 1.4964,
+      "step": 4968
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015787667710824642,
+      "loss": 1.5352,
+      "step": 4969
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001578594294513424,
+      "loss": 1.5153,
+      "step": 4970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015784217920665106,
+      "loss": 1.4822,
+      "step": 4971
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015782492637494392,
+      "loss": 1.4843,
+      "step": 4972
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015780767095699267,
+      "loss": 1.4727,
+      "step": 4973
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015779041295356904,
+      "loss": 1.4531,
+      "step": 4974
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015777315236544488,
+      "loss": 1.4555,
+      "step": 4975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015775588919339216,
+      "loss": 1.4632,
+      "step": 4976
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00015773862343818306,
+      "loss": 1.4478,
+      "step": 4977
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015772135510058974,
+      "loss": 1.4469,
+      "step": 4978
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001577040841813845,
+      "loss": 1.4562,
+      "step": 4979
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001576868106813399,
+      "loss": 1.466,
+      "step": 4980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015766953460122842,
+      "loss": 1.4379,
+      "step": 4981
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015765225594182276,
+      "loss": 1.4358,
+      "step": 4982
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015763497470389572,
+      "loss": 1.4586,
+      "step": 4983
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015761769088822023,
+      "loss": 1.437,
+      "step": 4984
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015760040449556924,
+      "loss": 1.4752,
+      "step": 4985
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015758311552671597,
+      "loss": 1.3903,
+      "step": 4986
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015756582398243364,
+      "loss": 1.402,
+      "step": 4987
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015754852986349564,
+      "loss": 1.355,
+      "step": 4988
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015753123317067543,
+      "loss": 1.4023,
+      "step": 4989
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015751393390474666,
+      "loss": 1.4027,
+      "step": 4990
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015749663206648303,
+      "loss": 1.3795,
+      "step": 4991
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015747932765665832,
+      "loss": 1.442,
+      "step": 4992
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015746202067604652,
+      "loss": 1.4285,
+      "step": 4993
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015744471112542172,
+      "loss": 1.3531,
+      "step": 4994
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015742739900555803,
+      "loss": 1.4236,
+      "step": 4995
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001574100843172298,
+      "loss": 1.3463,
+      "step": 4996
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015739276706121138,
+      "loss": 1.3974,
+      "step": 4997
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015737544723827735,
+      "loss": 1.3717,
+      "step": 4998
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015735812484920234,
+      "loss": 1.3629,
+      "step": 4999
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015734079989476106,
+      "loss": 1.3485,
+      "step": 5000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001573234723757284,
+      "loss": 1.3412,
+      "step": 5001
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015730614229287932,
+      "loss": 1.3648,
+      "step": 5002
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015728880964698894,
+      "loss": 1.2797,
+      "step": 5003
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001572714744388325,
+      "loss": 1.3116,
+      "step": 5004
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015725413666918528,
+      "loss": 1.2754,
+      "step": 5005
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015723679633882268,
+      "loss": 1.3109,
+      "step": 5006
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015721945344852035,
+      "loss": 1.2752,
+      "step": 5007
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00015720210799905388,
+      "loss": 1.2794,
+      "step": 5008
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001571847599911991,
+      "loss": 1.2703,
+      "step": 5009
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015716740942573186,
+      "loss": 1.2465,
+      "step": 5010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015715005630342823,
+      "loss": 1.281,
+      "step": 5011
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001571327006250643,
+      "loss": 1.2747,
+      "step": 5012
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015711534239141632,
+      "loss": 1.2547,
+      "step": 5013
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015709798160326065,
+      "loss": 1.2663,
+      "step": 5014
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015708061826137372,
+      "loss": 1.2311,
+      "step": 5015
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015706325236653217,
+      "loss": 1.2024,
+      "step": 5016
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015704588391951261,
+      "loss": 1.2115,
+      "step": 5017
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000157028512921092,
+      "loss": 1.1821,
+      "step": 5018
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001570111393720471,
+      "loss": 1.2021,
+      "step": 5019
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015699376327315507,
+      "loss": 1.1842,
+      "step": 5020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000156976384625193,
+      "loss": 1.16,
+      "step": 5021
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015695900342893815,
+      "loss": 1.1826,
+      "step": 5022
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015694161968516794,
+      "loss": 1.1608,
+      "step": 5023
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001569242333946599,
+      "loss": 1.1103,
+      "step": 5024
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001569068445581915,
+      "loss": 1.1269,
+      "step": 5025
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015688945317654063,
+      "loss": 1.1796,
+      "step": 5026
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000156872059250485,
+      "loss": 1.1743,
+      "step": 5027
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015685466278080265,
+      "loss": 1.1242,
+      "step": 5028
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015683726376827158,
+      "loss": 1.0991,
+      "step": 5029
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015681986221367003,
+      "loss": 1.1155,
+      "step": 5030
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001568024581177762,
+      "loss": 1.1057,
+      "step": 5031
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001567850514813686,
+      "loss": 1.0602,
+      "step": 5032
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015676764230522565,
+      "loss": 1.0722,
+      "step": 5033
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015675023059012606,
+      "loss": 1.1192,
+      "step": 5034
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015673281633684855,
+      "loss": 1.0745,
+      "step": 5035
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015671539954617198,
+      "loss": 1.0607,
+      "step": 5036
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015669798021887533,
+      "loss": 1.0143,
+      "step": 5037
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00015668055835573767,
+      "loss": 1.0303,
+      "step": 5038
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001566631339575382,
+      "loss": 1.0428,
+      "step": 5039
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015664570702505625,
+      "loss": 1.0238,
+      "step": 5040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015662827755907123,
+      "loss": 1.0032,
+      "step": 5041
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001566108455603627,
+      "loss": 0.9886,
+      "step": 5042
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001565934110297103,
+      "loss": 0.9861,
+      "step": 5043
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001565759739678938,
+      "loss": 0.9521,
+      "step": 5044
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015655853437569308,
+      "loss": 0.9627,
+      "step": 5045
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015654109225388813,
+      "loss": 0.95,
+      "step": 5046
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015652364760325907,
+      "loss": 0.923,
+      "step": 5047
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015650620042458608,
+      "loss": 0.8986,
+      "step": 5048
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015648875071864955,
+      "loss": 0.9266,
+      "step": 5049
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001564712984862299,
+      "loss": 0.9261,
+      "step": 5050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015645384372810764,
+      "loss": 0.9062,
+      "step": 5051
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001564363864450635,
+      "loss": 0.944,
+      "step": 5052
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001564189266378783,
+      "loss": 0.9441,
+      "step": 5053
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015640146430733284,
+      "loss": 0.9771,
+      "step": 5054
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015638399945420818,
+      "loss": 0.943,
+      "step": 5055
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015636653207928546,
+      "loss": 0.9009,
+      "step": 5056
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001563490621833459,
+      "loss": 0.8931,
+      "step": 5057
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015633158976717081,
+      "loss": 0.8739,
+      "step": 5058
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001563141148315417,
+      "loss": 0.864,
+      "step": 5059
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015629663737724015,
+      "loss": 0.8701,
+      "step": 5060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015627915740504782,
+      "loss": 0.8571,
+      "step": 5061
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001562616749157465,
+      "loss": 0.864,
+      "step": 5062
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001562441899101181,
+      "loss": 0.8423,
+      "step": 5063
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001562267023889447,
+      "loss": 0.8665,
+      "step": 5064
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001562092123530084,
+      "loss": 0.8174,
+      "step": 5065
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015619171980309142,
+      "loss": 0.8182,
+      "step": 5066
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015617422473997614,
+      "loss": 0.7775,
+      "step": 5067
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015615672716444507,
+      "loss": 0.8056,
+      "step": 5068
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00015613922707728074,
+      "loss": 0.7836,
+      "step": 5069
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015612172447926592,
+      "loss": 0.7838,
+      "step": 5070
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015610421937118334,
+      "loss": 0.748,
+      "step": 5071
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015608671175381598,
+      "loss": 0.7844,
+      "step": 5072
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015606920162794684,
+      "loss": 0.7847,
+      "step": 5073
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001560516889943591,
+      "loss": 0.7903,
+      "step": 5074
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015603417385383602,
+      "loss": 0.8098,
+      "step": 5075
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015601665620716096,
+      "loss": 0.7927,
+      "step": 5076
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015599913605511738,
+      "loss": 0.8041,
+      "step": 5077
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001559816133984889,
+      "loss": 0.783,
+      "step": 5078
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015596408823805922,
+      "loss": 0.7964,
+      "step": 5079
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015594656057461216,
+      "loss": 0.7413,
+      "step": 5080
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015592903040893168,
+      "loss": 0.7789,
+      "step": 5081
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001559114977418018,
+      "loss": 0.7315,
+      "step": 5082
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015589396257400666,
+      "loss": 0.7326,
+      "step": 5083
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015587642490633058,
+      "loss": 0.7344,
+      "step": 5084
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015585888473955789,
+      "loss": 0.7829,
+      "step": 5085
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001558413420744731,
+      "loss": 0.7915,
+      "step": 5086
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015582379691186078,
+      "loss": 0.7101,
+      "step": 5087
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015580624925250568,
+      "loss": 0.734,
+      "step": 5088
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015578869909719265,
+      "loss": 0.7885,
+      "step": 5089
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015577114644670656,
+      "loss": 0.7186,
+      "step": 5090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015575359130183254,
+      "loss": 0.767,
+      "step": 5091
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015573603366335567,
+      "loss": 0.7878,
+      "step": 5092
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001557184735320613,
+      "loss": 0.7866,
+      "step": 5093
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001557009109087348,
+      "loss": 0.8819,
+      "step": 5094
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015568334579416158,
+      "loss": 1.4106,
+      "step": 5095
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015566577818912735,
+      "loss": 2.5233,
+      "step": 5096
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015564820809441777,
+      "loss": 5.1498,
+      "step": 5097
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015563063551081866,
+      "loss": 2.5373,
+      "step": 5098
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00015561306043911604,
+      "loss": 2.3618,
+      "step": 5099
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001555954828800959,
+      "loss": 2.1123,
+      "step": 5100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001555779028345444,
+      "loss": 1.4853,
+      "step": 5101
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015556032030324785,
+      "loss": 2.0159,
+      "step": 5102
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015554273528699258,
+      "loss": 1.6846,
+      "step": 5103
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015552514778656518,
+      "loss": 1.2757,
+      "step": 5104
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015550755780275214,
+      "loss": 1.3637,
+      "step": 5105
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015548996533634027,
+      "loss": 1.5923,
+      "step": 5106
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001554723703881164,
+      "loss": 1.4489,
+      "step": 5107
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015545477295886742,
+      "loss": 1.2239,
+      "step": 5108
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001554371730493804,
+      "loss": 1.212,
+      "step": 5109
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015541957066044252,
+      "loss": 1.3442,
+      "step": 5110
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015540196579284104,
+      "loss": 1.1842,
+      "step": 5111
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015538435844736332,
+      "loss": 1.1634,
+      "step": 5112
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001553667486247969,
+      "loss": 1.1774,
+      "step": 5113
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015534913632592943,
+      "loss": 1.1188,
+      "step": 5114
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001553315215515485,
+      "loss": 1.0483,
+      "step": 5115
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015531390430244203,
+      "loss": 1.078,
+      "step": 5116
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015529628457939794,
+      "loss": 1.0227,
+      "step": 5117
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001552786623832043,
+      "loss": 1.0194,
+      "step": 5118
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015526103771464918,
+      "loss": 1.0166,
+      "step": 5119
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015524341057452095,
+      "loss": 0.9548,
+      "step": 5120
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015522578096360796,
+      "loss": 0.9623,
+      "step": 5121
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015520814888269867,
+      "loss": 0.9285,
+      "step": 5122
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015519051433258177,
+      "loss": 0.9162,
+      "step": 5123
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015517287731404587,
+      "loss": 0.9216,
+      "step": 5124
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015515523782787987,
+      "loss": 0.8757,
+      "step": 5125
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015513759587487265,
+      "loss": 0.8574,
+      "step": 5126
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015511995145581326,
+      "loss": 0.8569,
+      "step": 5127
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001551023045714909,
+      "loss": 0.8643,
+      "step": 5128
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00015508465522269478,
+      "loss": 0.8654,
+      "step": 5129
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001550670034102143,
+      "loss": 0.8353,
+      "step": 5130
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015504934913483894,
+      "loss": 0.8117,
+      "step": 5131
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001550316923973583,
+      "loss": 0.8181,
+      "step": 5132
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015501403319856207,
+      "loss": 0.8046,
+      "step": 5133
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015499637153924012,
+      "loss": 0.8004,
+      "step": 5134
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001549787074201823,
+      "loss": 0.7624,
+      "step": 5135
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015496104084217866,
+      "loss": 0.7866,
+      "step": 5136
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001549433718060194,
+      "loss": 0.7789,
+      "step": 5137
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015492570031249472,
+      "loss": 0.7708,
+      "step": 5138
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015490802636239498,
+      "loss": 0.7787,
+      "step": 5139
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015489034995651074,
+      "loss": 0.7373,
+      "step": 5140
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015487267109563248,
+      "loss": 0.7481,
+      "step": 5141
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015485498978055093,
+      "loss": 0.7244,
+      "step": 5142
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001548373060120569,
+      "loss": 0.7317,
+      "step": 5143
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015481961979094133,
+      "loss": 0.7502,
+      "step": 5144
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015480193111799522,
+      "loss": 0.7309,
+      "step": 5145
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015478423999400965,
+      "loss": 0.7022,
+      "step": 5146
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015476654641977596,
+      "loss": 0.704,
+      "step": 5147
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015474885039608544,
+      "loss": 0.6789,
+      "step": 5148
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001547311519237296,
+      "loss": 0.6843,
+      "step": 5149
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015471345100349993,
+      "loss": 0.6577,
+      "step": 5150
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001546957476361882,
+      "loss": 0.6812,
+      "step": 5151
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015467804182258616,
+      "loss": 0.699,
+      "step": 5152
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001546603335634857,
+      "loss": 0.6706,
+      "step": 5153
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015464262285967885,
+      "loss": 0.6856,
+      "step": 5154
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015462490971195774,
+      "loss": 0.6748,
+      "step": 5155
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015460719412111452,
+      "loss": 0.6589,
+      "step": 5156
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015458947608794162,
+      "loss": 0.6627,
+      "step": 5157
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015457175561323148,
+      "loss": 0.6674,
+      "step": 5158
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015455403269777659,
+      "loss": 0.6692,
+      "step": 5159
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015453630734236967,
+      "loss": 0.6581,
+      "step": 5160
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00015451857954780347,
+      "loss": 0.6458,
+      "step": 5161
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001545008493148709,
+      "loss": 0.6743,
+      "step": 5162
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015448311664436487,
+      "loss": 0.6308,
+      "step": 5163
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001544653815370786,
+      "loss": 0.6381,
+      "step": 5164
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015444764399380522,
+      "loss": 0.6533,
+      "step": 5165
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001544299040153381,
+      "loss": 0.6391,
+      "step": 5166
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001544121616024706,
+      "loss": 0.6341,
+      "step": 5167
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015439441675599633,
+      "loss": 0.6484,
+      "step": 5168
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001543766694767089,
+      "loss": 0.6382,
+      "step": 5169
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015435891976540205,
+      "loss": 0.6068,
+      "step": 5170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015434116762286966,
+      "loss": 0.5907,
+      "step": 5171
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001543234130499057,
+      "loss": 0.61,
+      "step": 5172
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015430565604730427,
+      "loss": 0.5899,
+      "step": 5173
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001542878966158596,
+      "loss": 0.6204,
+      "step": 5174
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015427013475636584,
+      "loss": 0.6017,
+      "step": 5175
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001542523704696175,
+      "loss": 0.6095,
+      "step": 5176
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015423460375640913,
+      "loss": 0.6213,
+      "step": 5177
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015421683461753526,
+      "loss": 0.5794,
+      "step": 5178
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015419906305379067,
+      "loss": 0.5996,
+      "step": 5179
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015418128906597022,
+      "loss": 0.6044,
+      "step": 5180
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015416351265486883,
+      "loss": 0.5867,
+      "step": 5181
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015414573382128158,
+      "loss": 0.5712,
+      "step": 5182
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015412795256600364,
+      "loss": 0.557,
+      "step": 5183
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015411016888983024,
+      "loss": 0.6181,
+      "step": 5184
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015409238279355677,
+      "loss": 0.5911,
+      "step": 5185
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015407459427797874,
+      "loss": 0.569,
+      "step": 5186
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015405680334389178,
+      "loss": 0.566,
+      "step": 5187
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015403900999209157,
+      "loss": 0.616,
+      "step": 5188
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015402121422337386,
+      "loss": 0.5591,
+      "step": 5189
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015400341603853467,
+      "loss": 0.5898,
+      "step": 5190
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00015398561543836998,
+      "loss": 0.5645,
+      "step": 5191
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015396781242367594,
+      "loss": 0.5805,
+      "step": 5192
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015395000699524886,
+      "loss": 0.5684,
+      "step": 5193
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000153932199153885,
+      "loss": 0.5716,
+      "step": 5194
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015391438890038087,
+      "loss": 0.5574,
+      "step": 5195
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015389657623553297,
+      "loss": 0.5783,
+      "step": 5196
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001538787611601381,
+      "loss": 0.5738,
+      "step": 5197
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000153860943674993,
+      "loss": 0.5467,
+      "step": 5198
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015384312378089447,
+      "loss": 0.5754,
+      "step": 5199
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015382530147863967,
+      "loss": 0.5529,
+      "step": 5200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015380747676902563,
+      "loss": 0.5654,
+      "step": 5201
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015378964965284952,
+      "loss": 0.5322,
+      "step": 5202
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015377182013090877,
+      "loss": 0.5711,
+      "step": 5203
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015375398820400072,
+      "loss": 0.559,
+      "step": 5204
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015373615387292294,
+      "loss": 0.5512,
+      "step": 5205
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015371831713847312,
+      "loss": 0.5382,
+      "step": 5206
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000153700478001449,
+      "loss": 0.5427,
+      "step": 5207
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015368263646264838,
+      "loss": 0.5413,
+      "step": 5208
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001536647925228693,
+      "loss": 0.5274,
+      "step": 5209
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001536469461829098,
+      "loss": 0.5491,
+      "step": 5210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015362909744356807,
+      "loss": 0.5499,
+      "step": 5211
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015361124630564244,
+      "loss": 0.5386,
+      "step": 5212
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015359339276993126,
+      "loss": 0.5321,
+      "step": 5213
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015357553683723307,
+      "loss": 0.5264,
+      "step": 5214
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015355767850834648,
+      "loss": 0.5306,
+      "step": 5215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015353981778407015,
+      "loss": 0.5113,
+      "step": 5216
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000153521954665203,
+      "loss": 0.5053,
+      "step": 5217
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015350408915254393,
+      "loss": 0.5259,
+      "step": 5218
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015348622124689197,
+      "loss": 0.5184,
+      "step": 5219
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001534683509490463,
+      "loss": 0.549,
+      "step": 5220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015345047825980615,
+      "loss": 0.5076,
+      "step": 5221
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00015343260317997085,
+      "loss": 0.5144,
+      "step": 5222
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015341472571033996,
+      "loss": 0.537,
+      "step": 5223
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015339684585171298,
+      "loss": 0.5383,
+      "step": 5224
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015337896360488964,
+      "loss": 0.5343,
+      "step": 5225
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015336107897066968,
+      "loss": 0.5319,
+      "step": 5226
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015334319194985305,
+      "loss": 0.5471,
+      "step": 5227
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015332530254323973,
+      "loss": 0.5202,
+      "step": 5228
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015330741075162985,
+      "loss": 0.5084,
+      "step": 5229
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001532895165758236,
+      "loss": 0.5148,
+      "step": 5230
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015327162001662129,
+      "loss": 0.5278,
+      "step": 5231
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015325372107482342,
+      "loss": 0.522,
+      "step": 5232
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015323581975123046,
+      "loss": 0.5229,
+      "step": 5233
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015321791604664312,
+      "loss": 0.5276,
+      "step": 5234
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001532000099618621,
+      "loss": 0.5278,
+      "step": 5235
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015318210149768824,
+      "loss": 0.504,
+      "step": 5236
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015316419065492255,
+      "loss": 0.4718,
+      "step": 5237
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015314627743436606,
+      "loss": 0.4884,
+      "step": 5238
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015312836183682,
+      "loss": 0.5208,
+      "step": 5239
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001531104438630856,
+      "loss": 0.517,
+      "step": 5240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001530925235139643,
+      "loss": 0.4957,
+      "step": 5241
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015307460079025753,
+      "loss": 0.5114,
+      "step": 5242
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015305667569276695,
+      "loss": 0.4951,
+      "step": 5243
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015303874822229427,
+      "loss": 0.4914,
+      "step": 5244
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015302081837964123,
+      "loss": 0.4765,
+      "step": 5245
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015300288616560982,
+      "loss": 0.4703,
+      "step": 5246
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015298495158100204,
+      "loss": 0.476,
+      "step": 5247
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015296701462661998,
+      "loss": 0.4963,
+      "step": 5248
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000152949075303266,
+      "loss": 0.4803,
+      "step": 5249
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001529311336117423,
+      "loss": 0.4897,
+      "step": 5250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015291318955285146,
+      "loss": 0.4821,
+      "step": 5251
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015289524312739593,
+      "loss": 0.4766,
+      "step": 5252
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00015287729433617844,
+      "loss": 0.48,
+      "step": 5253
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015285934318000173,
+      "loss": 0.4886,
+      "step": 5254
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015284138965966865,
+      "loss": 0.5044,
+      "step": 5255
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015282343377598224,
+      "loss": 0.4819,
+      "step": 5256
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001528054755297455,
+      "loss": 0.4963,
+      "step": 5257
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015278751492176168,
+      "loss": 0.467,
+      "step": 5258
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015276955195283408,
+      "loss": 0.4637,
+      "step": 5259
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015275158662376604,
+      "loss": 0.4854,
+      "step": 5260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015273361893536116,
+      "loss": 0.4749,
+      "step": 5261
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000152715648888423,
+      "loss": 0.4959,
+      "step": 5262
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015269767648375525,
+      "loss": 0.4719,
+      "step": 5263
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015267970172216182,
+      "loss": 0.4786,
+      "step": 5264
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015266172460444655,
+      "loss": 0.4755,
+      "step": 5265
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015264374513141352,
+      "loss": 0.4767,
+      "step": 5266
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015262576330386682,
+      "loss": 0.4642,
+      "step": 5267
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015260777912261077,
+      "loss": 0.4594,
+      "step": 5268
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015258979258844966,
+      "loss": 0.4734,
+      "step": 5269
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015257180370218797,
+      "loss": 0.4583,
+      "step": 5270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015255381246463028,
+      "loss": 0.4715,
+      "step": 5271
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001525358188765812,
+      "loss": 0.4588,
+      "step": 5272
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015251782293884558,
+      "loss": 0.4557,
+      "step": 5273
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001524998246522282,
+      "loss": 0.4681,
+      "step": 5274
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015248182401753413,
+      "loss": 0.4441,
+      "step": 5275
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001524638210355684,
+      "loss": 0.4693,
+      "step": 5276
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015244581570713622,
+      "loss": 0.452,
+      "step": 5277
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015242780803304288,
+      "loss": 0.457,
+      "step": 5278
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001524097980140938,
+      "loss": 0.446,
+      "step": 5279
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015239178565109443,
+      "loss": 0.4754,
+      "step": 5280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001523737709448505,
+      "loss": 0.4544,
+      "step": 5281
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015235575389616757,
+      "loss": 0.4513,
+      "step": 5282
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00015233773450585157,
+      "loss": 0.4712,
+      "step": 5283
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015231971277470839,
+      "loss": 0.4611,
+      "step": 5284
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015230168870354404,
+      "loss": 0.4604,
+      "step": 5285
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015228366229316472,
+      "loss": 0.4421,
+      "step": 5286
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015226563354437658,
+      "loss": 0.4309,
+      "step": 5287
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015224760245798603,
+      "loss": 0.4441,
+      "step": 5288
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015222956903479946,
+      "loss": 0.4374,
+      "step": 5289
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001522115332756235,
+      "loss": 0.4645,
+      "step": 5290
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015219349518126474,
+      "loss": 0.4603,
+      "step": 5291
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015217545475252996,
+      "loss": 0.444,
+      "step": 5292
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015215741199022605,
+      "loss": 0.4509,
+      "step": 5293
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015213936689515998,
+      "loss": 0.4431,
+      "step": 5294
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015212131946813878,
+      "loss": 0.4655,
+      "step": 5295
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015210326970996964,
+      "loss": 0.4447,
+      "step": 5296
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015208521762145985,
+      "loss": 0.4476,
+      "step": 5297
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015206716320341685,
+      "loss": 0.4468,
+      "step": 5298
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015204910645664804,
+      "loss": 0.4618,
+      "step": 5299
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001520310473819611,
+      "loss": 0.4271,
+      "step": 5300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015201298598016368,
+      "loss": 0.4258,
+      "step": 5301
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015199492225206359,
+      "loss": 0.4287,
+      "step": 5302
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015197685619846874,
+      "loss": 0.4611,
+      "step": 5303
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015195878782018718,
+      "loss": 0.4174,
+      "step": 5304
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015194071711802697,
+      "loss": 0.4319,
+      "step": 5305
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015192264409279632,
+      "loss": 0.4445,
+      "step": 5306
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015190456874530363,
+      "loss": 0.4121,
+      "step": 5307
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015188649107635726,
+      "loss": 0.4216,
+      "step": 5308
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015186841108676577,
+      "loss": 0.4286,
+      "step": 5309
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015185032877733777,
+      "loss": 0.416,
+      "step": 5310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000151832244148882,
+      "loss": 0.4212,
+      "step": 5311
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015181415720220737,
+      "loss": 0.416,
+      "step": 5312
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001517960679381228,
+      "loss": 0.4429,
+      "step": 5313
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00015177797635743725,
+      "loss": 0.4129,
+      "step": 5314
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015175988246095995,
+      "loss": 0.4171,
+      "step": 5315
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015174178624950018,
+      "loss": 0.43,
+      "step": 5316
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015172368772386726,
+      "loss": 0.4163,
+      "step": 5317
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015170558688487062,
+      "loss": 0.4044,
+      "step": 5318
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015168748373331988,
+      "loss": 0.4081,
+      "step": 5319
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015166937827002473,
+      "loss": 0.443,
+      "step": 5320
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015165127049579491,
+      "loss": 0.4163,
+      "step": 5321
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001516331604114403,
+      "loss": 0.429,
+      "step": 5322
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015161504801777088,
+      "loss": 0.4246,
+      "step": 5323
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015159693331559675,
+      "loss": 0.4191,
+      "step": 5324
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015157881630572807,
+      "loss": 0.4248,
+      "step": 5325
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015156069698897513,
+      "loss": 0.4298,
+      "step": 5326
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015154257536614834,
+      "loss": 0.4114,
+      "step": 5327
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015152445143805824,
+      "loss": 0.421,
+      "step": 5328
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015150632520551539,
+      "loss": 0.4135,
+      "step": 5329
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015148819666933045,
+      "loss": 0.4153,
+      "step": 5330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015147006583031427,
+      "loss": 0.4105,
+      "step": 5331
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015145193268927776,
+      "loss": 0.4209,
+      "step": 5332
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015143379724703198,
+      "loss": 0.4097,
+      "step": 5333
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015141565950438794,
+      "loss": 0.4316,
+      "step": 5334
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015139751946215694,
+      "loss": 0.4151,
+      "step": 5335
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015137937712115027,
+      "loss": 0.4351,
+      "step": 5336
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015136123248217934,
+      "loss": 0.4284,
+      "step": 5337
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015134308554605572,
+      "loss": 0.4133,
+      "step": 5338
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000151324936313591,
+      "loss": 0.4218,
+      "step": 5339
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015130678478559695,
+      "loss": 0.42,
+      "step": 5340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001512886309628853,
+      "loss": 0.3915,
+      "step": 5341
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015127047484626816,
+      "loss": 0.4064,
+      "step": 5342
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015125231643655748,
+      "loss": 0.4042,
+      "step": 5343
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00015123415573456534,
+      "loss": 0.3999,
+      "step": 5344
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001512159927411041,
+      "loss": 0.4031,
+      "step": 5345
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000151197827456986,
+      "loss": 0.3951,
+      "step": 5346
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001511796598830236,
+      "loss": 0.3829,
+      "step": 5347
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015116149002002933,
+      "loss": 0.3935,
+      "step": 5348
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015114331786881597,
+      "loss": 0.4084,
+      "step": 5349
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001511251434301962,
+      "loss": 0.4034,
+      "step": 5350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015110696670498288,
+      "loss": 0.4004,
+      "step": 5351
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000151088787693989,
+      "loss": 0.401,
+      "step": 5352
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001510706063980276,
+      "loss": 0.4026,
+      "step": 5353
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015105242281791188,
+      "loss": 0.3869,
+      "step": 5354
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015103423695445509,
+      "loss": 0.4018,
+      "step": 5355
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015101604880847058,
+      "loss": 0.381,
+      "step": 5356
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015099785838077186,
+      "loss": 0.4078,
+      "step": 5357
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015097966567217242,
+      "loss": 0.378,
+      "step": 5358
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015096147068348608,
+      "loss": 0.3886,
+      "step": 5359
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001509432734155265,
+      "loss": 0.3778,
+      "step": 5360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001509250738691076,
+      "loss": 0.3731,
+      "step": 5361
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015090687204504338,
+      "loss": 0.3894,
+      "step": 5362
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015088866794414786,
+      "loss": 0.3768,
+      "step": 5363
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015087046156723536,
+      "loss": 0.3767,
+      "step": 5364
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015085225291511998,
+      "loss": 0.4178,
+      "step": 5365
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015083404198861628,
+      "loss": 0.3668,
+      "step": 5366
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015081582878853866,
+      "loss": 0.3835,
+      "step": 5367
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015079761331570173,
+      "loss": 0.4054,
+      "step": 5368
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001507793955709202,
+      "loss": 0.381,
+      "step": 5369
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015076117555500883,
+      "loss": 0.3818,
+      "step": 5370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015074295326878256,
+      "loss": 0.382,
+      "step": 5371
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001507247287130564,
+      "loss": 0.3852,
+      "step": 5372
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015070650188864538,
+      "loss": 0.3676,
+      "step": 5373
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015068827279636477,
+      "loss": 0.3823,
+      "step": 5374
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00015067004143702983,
+      "loss": 0.3881,
+      "step": 5375
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000150651807811456,
+      "loss": 0.3729,
+      "step": 5376
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015063357192045875,
+      "loss": 0.3804,
+      "step": 5377
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015061533376485375,
+      "loss": 0.3806,
+      "step": 5378
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015059709334545662,
+      "loss": 0.3883,
+      "step": 5379
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015057885066308325,
+      "loss": 0.3875,
+      "step": 5380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015056060571854953,
+      "loss": 0.3777,
+      "step": 5381
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001505423585126714,
+      "loss": 0.3816,
+      "step": 5382
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001505241090462651,
+      "loss": 0.3681,
+      "step": 5383
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015050585732014678,
+      "loss": 0.3731,
+      "step": 5384
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015048760333513273,
+      "loss": 0.3721,
+      "step": 5385
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015046934709203943,
+      "loss": 0.3739,
+      "step": 5386
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015045108859168335,
+      "loss": 0.3586,
+      "step": 5387
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015043282783488114,
+      "loss": 0.3816,
+      "step": 5388
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015041456482244945,
+      "loss": 0.3713,
+      "step": 5389
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001503962995552052,
+      "loss": 0.3722,
+      "step": 5390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015037803203396525,
+      "loss": 0.3604,
+      "step": 5391
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001503597622595466,
+      "loss": 0.3542,
+      "step": 5392
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015034149023276648,
+      "loss": 0.3559,
+      "step": 5393
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015032321595444202,
+      "loss": 0.3585,
+      "step": 5394
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015030493942539056,
+      "loss": 0.3969,
+      "step": 5395
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015028666064642953,
+      "loss": 0.3688,
+      "step": 5396
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001502683796183765,
+      "loss": 0.3616,
+      "step": 5397
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000150250096342049,
+      "loss": 0.3546,
+      "step": 5398
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015023181081826484,
+      "loss": 0.3512,
+      "step": 5399
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015021352304784186,
+      "loss": 0.375,
+      "step": 5400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001501952330315979,
+      "loss": 0.3806,
+      "step": 5401
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015017694077035107,
+      "loss": 0.3711,
+      "step": 5402
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015015864626491946,
+      "loss": 0.3725,
+      "step": 5403
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015014034951612134,
+      "loss": 0.3504,
+      "step": 5404
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015012205052477494,
+      "loss": 0.3565,
+      "step": 5405
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00015010374929169885,
+      "loss": 0.3802,
+      "step": 5406
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001500854458177114,
+      "loss": 0.3448,
+      "step": 5407
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001500671401036314,
+      "loss": 0.3579,
+      "step": 5408
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001500488321502775,
+      "loss": 0.3757,
+      "step": 5409
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015003052195846855,
+      "loss": 0.3474,
+      "step": 5410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00015001220952902346,
+      "loss": 0.3809,
+      "step": 5411
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014999389486276127,
+      "loss": 0.3546,
+      "step": 5412
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014997557796050113,
+      "loss": 0.3621,
+      "step": 5413
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014995725882306225,
+      "loss": 0.3579,
+      "step": 5414
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014993893745126395,
+      "loss": 0.3594,
+      "step": 5415
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001499206138459257,
+      "loss": 0.3638,
+      "step": 5416
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014990228800786697,
+      "loss": 0.3905,
+      "step": 5417
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014988395993790747,
+      "loss": 0.3461,
+      "step": 5418
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014986562963686687,
+      "loss": 0.356,
+      "step": 5419
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014984729710556505,
+      "loss": 0.3303,
+      "step": 5420
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014982896234482188,
+      "loss": 0.3649,
+      "step": 5421
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014981062535545744,
+      "loss": 0.3485,
+      "step": 5422
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014979228613829184,
+      "loss": 0.3443,
+      "step": 5423
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001497739446941453,
+      "loss": 0.3662,
+      "step": 5424
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014975560102383815,
+      "loss": 0.3615,
+      "step": 5425
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014973725512819084,
+      "loss": 0.3657,
+      "step": 5426
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014971890700802388,
+      "loss": 0.3467,
+      "step": 5427
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014970055666415792,
+      "loss": 0.3596,
+      "step": 5428
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014968220409741366,
+      "loss": 0.3519,
+      "step": 5429
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001496638493086119,
+      "loss": 0.3602,
+      "step": 5430
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014964549229857363,
+      "loss": 0.3307,
+      "step": 5431
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014962713306811983,
+      "loss": 0.3359,
+      "step": 5432
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014960877161807165,
+      "loss": 0.349,
+      "step": 5433
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014959040794925032,
+      "loss": 0.3469,
+      "step": 5434
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014957204206247714,
+      "loss": 0.3442,
+      "step": 5435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00014955367395857354,
+      "loss": 0.3582,
+      "step": 5436
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014953530363836102,
+      "loss": 0.3547,
+      "step": 5437
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014951693110266124,
+      "loss": 0.3298,
+      "step": 5438
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014949855635229588,
+      "loss": 0.3359,
+      "step": 5439
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014948017938808678,
+      "loss": 0.3419,
+      "step": 5440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014946180021085587,
+      "loss": 0.3538,
+      "step": 5441
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014944341882142513,
+      "loss": 0.3369,
+      "step": 5442
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001494250352206167,
+      "loss": 0.3311,
+      "step": 5443
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014940664940925284,
+      "loss": 0.3377,
+      "step": 5444
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014938826138815578,
+      "loss": 0.3367,
+      "step": 5445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014936987115814795,
+      "loss": 0.3513,
+      "step": 5446
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001493514787200519,
+      "loss": 0.3427,
+      "step": 5447
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001493330840746902,
+      "loss": 0.3367,
+      "step": 5448
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001493146872228856,
+      "loss": 0.3436,
+      "step": 5449
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014929628816546083,
+      "loss": 0.3469,
+      "step": 5450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001492778869032389,
+      "loss": 0.3615,
+      "step": 5451
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014925948343704273,
+      "loss": 0.3411,
+      "step": 5452
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014924107776769546,
+      "loss": 0.3512,
+      "step": 5453
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014922266989602032,
+      "loss": 0.3312,
+      "step": 5454
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014920425982284055,
+      "loss": 0.3315,
+      "step": 5455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014918584754897954,
+      "loss": 0.3478,
+      "step": 5456
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014916743307526086,
+      "loss": 0.3313,
+      "step": 5457
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014914901640250807,
+      "loss": 0.3295,
+      "step": 5458
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014913059753154484,
+      "loss": 0.335,
+      "step": 5459
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014911217646319499,
+      "loss": 0.3227,
+      "step": 5460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014909375319828238,
+      "loss": 0.3478,
+      "step": 5461
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014907532773763104,
+      "loss": 0.3278,
+      "step": 5462
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014905690008206504,
+      "loss": 0.3437,
+      "step": 5463
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014903847023240859,
+      "loss": 0.3375,
+      "step": 5464
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001490200381894859,
+      "loss": 0.3359,
+      "step": 5465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014900160395412137,
+      "loss": 0.3348,
+      "step": 5466
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00014898316752713953,
+      "loss": 0.3245,
+      "step": 5467
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014896472890936493,
+      "loss": 0.3367,
+      "step": 5468
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014894628810162226,
+      "loss": 0.3292,
+      "step": 5469
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014892784510473625,
+      "loss": 0.3385,
+      "step": 5470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001489093999195318,
+      "loss": 0.3315,
+      "step": 5471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014889095254683388,
+      "loss": 0.3249,
+      "step": 5472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014887250298746757,
+      "loss": 0.341,
+      "step": 5473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014885405124225804,
+      "loss": 0.3424,
+      "step": 5474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001488355973120305,
+      "loss": 0.3315,
+      "step": 5475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014881714119761033,
+      "loss": 0.3205,
+      "step": 5476
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000148798682899823,
+      "loss": 0.3456,
+      "step": 5477
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014878022241949409,
+      "loss": 0.343,
+      "step": 5478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014876175975744917,
+      "loss": 0.3414,
+      "step": 5479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001487432949145141,
+      "loss": 0.3251,
+      "step": 5480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014872482789151463,
+      "loss": 0.3364,
+      "step": 5481
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001487063586892768,
+      "loss": 0.3517,
+      "step": 5482
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014868788730862652,
+      "loss": 0.3354,
+      "step": 5483
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001486694137503901,
+      "loss": 0.3408,
+      "step": 5484
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014865093801539364,
+      "loss": 0.3262,
+      "step": 5485
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014863246010446356,
+      "loss": 0.3201,
+      "step": 5486
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014861398001842621,
+      "loss": 0.3255,
+      "step": 5487
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001485954977581082,
+      "loss": 0.3451,
+      "step": 5488
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001485770133243361,
+      "loss": 0.3372,
+      "step": 5489
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014855852671793665,
+      "loss": 0.3338,
+      "step": 5490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001485400379397367,
+      "loss": 0.3298,
+      "step": 5491
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014852154699056315,
+      "loss": 0.3195,
+      "step": 5492
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014850305387124302,
+      "loss": 0.3369,
+      "step": 5493
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001484845585826034,
+      "loss": 0.338,
+      "step": 5494
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001484660611254715,
+      "loss": 0.327,
+      "step": 5495
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014844756150067466,
+      "loss": 0.3215,
+      "step": 5496
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00014842905970904025,
+      "loss": 0.3231,
+      "step": 5497
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001484105557513958,
+      "loss": 0.3412,
+      "step": 5498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001483920496285689,
+      "loss": 0.3258,
+      "step": 5499
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001483735413413872,
+      "loss": 0.3339,
+      "step": 5500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014835503089067858,
+      "loss": 0.3298,
+      "step": 5501
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014833651827727082,
+      "loss": 0.3198,
+      "step": 5502
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014831800350199198,
+      "loss": 0.3494,
+      "step": 5503
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014829948656567012,
+      "loss": 0.3316,
+      "step": 5504
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014828096746913343,
+      "loss": 0.3378,
+      "step": 5505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014826244621321018,
+      "loss": 0.3328,
+      "step": 5506
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014824392279872873,
+      "loss": 0.3266,
+      "step": 5507
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001482253972265175,
+      "loss": 0.3308,
+      "step": 5508
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014820686949740516,
+      "loss": 0.3143,
+      "step": 5509
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001481883396122203,
+      "loss": 0.3203,
+      "step": 5510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001481698075717917,
+      "loss": 0.3103,
+      "step": 5511
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014815127337694815,
+      "loss": 0.3054,
+      "step": 5512
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014813273702851874,
+      "loss": 0.3197,
+      "step": 5513
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001481141985273324,
+      "loss": 0.3306,
+      "step": 5514
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001480956578742183,
+      "loss": 0.3266,
+      "step": 5515
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014807711507000565,
+      "loss": 0.333,
+      "step": 5516
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014805857011552384,
+      "loss": 0.3144,
+      "step": 5517
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014804002301160228,
+      "loss": 0.3259,
+      "step": 5518
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001480214737590705,
+      "loss": 0.3328,
+      "step": 5519
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001480029223587581,
+      "loss": 0.3227,
+      "step": 5520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014798436881149484,
+      "loss": 0.3292,
+      "step": 5521
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001479658131181105,
+      "loss": 0.3216,
+      "step": 5522
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000147947255279435,
+      "loss": 0.3404,
+      "step": 5523
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014792869529629837,
+      "loss": 0.3241,
+      "step": 5524
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014791013316953067,
+      "loss": 0.3252,
+      "step": 5525
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014789156889996214,
+      "loss": 0.3081,
+      "step": 5526
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00014787300248842304,
+      "loss": 0.3341,
+      "step": 5527
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001478544339357438,
+      "loss": 0.3108,
+      "step": 5528
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014783586324275486,
+      "loss": 0.3092,
+      "step": 5529
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014781729041028684,
+      "loss": 0.3044,
+      "step": 5530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001477987154391704,
+      "loss": 0.3283,
+      "step": 5531
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014778013833023632,
+      "loss": 0.3292,
+      "step": 5532
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014776155908431546,
+      "loss": 0.3202,
+      "step": 5533
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014774297770223883,
+      "loss": 0.3219,
+      "step": 5534
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001477243941848374,
+      "loss": 0.3185,
+      "step": 5535
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014770580853294238,
+      "loss": 0.3154,
+      "step": 5536
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014768722074738502,
+      "loss": 0.3326,
+      "step": 5537
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014766863082899667,
+      "loss": 0.3126,
+      "step": 5538
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014765003877860878,
+      "loss": 0.3004,
+      "step": 5539
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014763144459705284,
+      "loss": 0.3166,
+      "step": 5540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014761284828516055,
+      "loss": 0.3104,
+      "step": 5541
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001475942498437636,
+      "loss": 0.3212,
+      "step": 5542
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001475756492736938,
+      "loss": 0.3129,
+      "step": 5543
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014755704657578312,
+      "loss": 0.3141,
+      "step": 5544
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014753844175086352,
+      "loss": 0.3191,
+      "step": 5545
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001475198347997671,
+      "loss": 0.306,
+      "step": 5546
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001475012257233261,
+      "loss": 0.3132,
+      "step": 5547
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014748261452237284,
+      "loss": 0.3205,
+      "step": 5548
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014746400119773964,
+      "loss": 0.3082,
+      "step": 5549
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014744538575025908,
+      "loss": 0.3081,
+      "step": 5550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014742676818076366,
+      "loss": 0.2918,
+      "step": 5551
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001474081484900861,
+      "loss": 0.3172,
+      "step": 5552
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014738952667905916,
+      "loss": 0.3108,
+      "step": 5553
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014737090274851577,
+      "loss": 0.3234,
+      "step": 5554
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001473522766992888,
+      "loss": 0.3113,
+      "step": 5555
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014733364853221134,
+      "loss": 0.3125,
+      "step": 5556
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014731501824811658,
+      "loss": 0.3142,
+      "step": 5557
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001472963858478377,
+      "loss": 0.3169,
+      "step": 5558
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00014727775133220808,
+      "loss": 0.3199,
+      "step": 5559
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014725911470206117,
+      "loss": 0.3037,
+      "step": 5560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001472404759582305,
+      "loss": 0.3104,
+      "step": 5561
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014722183510154969,
+      "loss": 0.2972,
+      "step": 5562
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001472031921328524,
+      "loss": 0.3086,
+      "step": 5563
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014718454705297255,
+      "loss": 0.3129,
+      "step": 5564
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000147165899862744,
+      "loss": 0.2985,
+      "step": 5565
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014714725056300072,
+      "loss": 0.3179,
+      "step": 5566
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014712859915457684,
+      "loss": 0.3051,
+      "step": 5567
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014710994563830657,
+      "loss": 0.3212,
+      "step": 5568
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014709129001502414,
+      "loss": 0.3108,
+      "step": 5569
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014707263228556398,
+      "loss": 0.3276,
+      "step": 5570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001470539724507606,
+      "loss": 0.3172,
+      "step": 5571
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014703531051144845,
+      "loss": 0.3204,
+      "step": 5572
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014701664646846232,
+      "loss": 0.292,
+      "step": 5573
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014699798032263687,
+      "loss": 0.3233,
+      "step": 5574
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014697931207480699,
+      "loss": 0.3271,
+      "step": 5575
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014696064172580765,
+      "loss": 0.3012,
+      "step": 5576
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014694196927647388,
+      "loss": 0.3266,
+      "step": 5577
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001469232947276408,
+      "loss": 0.3104,
+      "step": 5578
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001469046180801436,
+      "loss": 0.315,
+      "step": 5579
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014688593933481764,
+      "loss": 0.3107,
+      "step": 5580
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014686725849249834,
+      "loss": 0.3138,
+      "step": 5581
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001468485755540212,
+      "loss": 0.3046,
+      "step": 5582
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014682989052022184,
+      "loss": 0.3081,
+      "step": 5583
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014681120339193594,
+      "loss": 0.3243,
+      "step": 5584
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014679251416999927,
+      "loss": 0.3108,
+      "step": 5585
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014677382285524775,
+      "loss": 0.2993,
+      "step": 5586
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014675512944851732,
+      "loss": 0.3138,
+      "step": 5587
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014673643395064407,
+      "loss": 0.3198,
+      "step": 5588
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00014671773636246418,
+      "loss": 0.3203,
+      "step": 5589
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014669903668481388,
+      "loss": 0.3103,
+      "step": 5590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014668033491852954,
+      "loss": 0.3123,
+      "step": 5591
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014666163106444758,
+      "loss": 0.3009,
+      "step": 5592
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014664292512340456,
+      "loss": 0.3106,
+      "step": 5593
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014662421709623713,
+      "loss": 0.3056,
+      "step": 5594
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014660550698378195,
+      "loss": 0.2936,
+      "step": 5595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001465867947868759,
+      "loss": 0.306,
+      "step": 5596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014656808050635586,
+      "loss": 0.3077,
+      "step": 5597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014654936414305886,
+      "loss": 0.2953,
+      "step": 5598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014653064569782195,
+      "loss": 0.3106,
+      "step": 5599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014651192517148236,
+      "loss": 0.298,
+      "step": 5600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014649320256487734,
+      "loss": 0.3099,
+      "step": 5601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001464744778788443,
+      "loss": 0.3138,
+      "step": 5602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014645575111422074,
+      "loss": 0.2892,
+      "step": 5603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014643702227184417,
+      "loss": 0.3149,
+      "step": 5604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014641829135255224,
+      "loss": 0.3007,
+      "step": 5605
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014639955835718272,
+      "loss": 0.2974,
+      "step": 5606
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014638082328657343,
+      "loss": 0.3071,
+      "step": 5607
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001463620861415623,
+      "loss": 0.3053,
+      "step": 5608
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014634334692298744,
+      "loss": 0.3049,
+      "step": 5609
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001463246056316869,
+      "loss": 0.3011,
+      "step": 5610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014630586226849882,
+      "loss": 0.3332,
+      "step": 5611
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014628711683426166,
+      "loss": 0.2995,
+      "step": 5612
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001462683693298137,
+      "loss": 0.3281,
+      "step": 5613
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014624961975599353,
+      "loss": 0.3081,
+      "step": 5614
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001462308681136396,
+      "loss": 0.3002,
+      "step": 5615
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001462121144035907,
+      "loss": 0.3066,
+      "step": 5616
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014619335862668553,
+      "loss": 0.3127,
+      "step": 5617
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000146174600783763,
+      "loss": 0.3131,
+      "step": 5618
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014615584087566201,
+      "loss": 0.3263,
+      "step": 5619
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00014613707890322164,
+      "loss": 0.3135,
+      "step": 5620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014611831486728102,
+      "loss": 0.3076,
+      "step": 5621
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001460995487686794,
+      "loss": 0.334,
+      "step": 5622
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014608078060825605,
+      "loss": 0.3036,
+      "step": 5623
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014606201038685045,
+      "loss": 0.2991,
+      "step": 5624
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014604323810530207,
+      "loss": 0.2872,
+      "step": 5625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014602446376445048,
+      "loss": 0.3024,
+      "step": 5626
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001460056873651354,
+      "loss": 0.307,
+      "step": 5627
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001459869089081966,
+      "loss": 0.3334,
+      "step": 5628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014596812839447402,
+      "loss": 0.313,
+      "step": 5629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014594934582480752,
+      "loss": 0.2863,
+      "step": 5630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014593056120003724,
+      "loss": 0.3115,
+      "step": 5631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014591177452100327,
+      "loss": 0.2837,
+      "step": 5632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014589298578854593,
+      "loss": 0.3125,
+      "step": 5633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001458741950035055,
+      "loss": 0.302,
+      "step": 5634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001458554021667224,
+      "loss": 0.311,
+      "step": 5635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014583660727903713,
+      "loss": 0.3058,
+      "step": 5636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014581781034129037,
+      "loss": 0.303,
+      "step": 5637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014579901135432276,
+      "loss": 0.304,
+      "step": 5638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001457802103189751,
+      "loss": 0.3106,
+      "step": 5639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001457614072360883,
+      "loss": 0.2992,
+      "step": 5640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014574260210650336,
+      "loss": 0.299,
+      "step": 5641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014572379493106127,
+      "loss": 0.2883,
+      "step": 5642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014570498571060326,
+      "loss": 0.3071,
+      "step": 5643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014568617444597051,
+      "loss": 0.3108,
+      "step": 5644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014566736113800444,
+      "loss": 0.307,
+      "step": 5645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001456485457875464,
+      "loss": 0.3073,
+      "step": 5646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014562972839543797,
+      "loss": 0.3126,
+      "step": 5647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001456109089625208,
+      "loss": 0.3057,
+      "step": 5648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014559208748963652,
+      "loss": 0.3021,
+      "step": 5649
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00014557326397762697,
+      "loss": 0.2994,
+      "step": 5650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000145554438427334,
+      "loss": 0.3085,
+      "step": 5651
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014553561083959963,
+      "loss": 0.2859,
+      "step": 5652
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014551678121526592,
+      "loss": 0.3234,
+      "step": 5653
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014549794955517507,
+      "loss": 0.3115,
+      "step": 5654
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014547911586016929,
+      "loss": 0.2917,
+      "step": 5655
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001454602801310909,
+      "loss": 0.2972,
+      "step": 5656
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001454414423687824,
+      "loss": 0.2962,
+      "step": 5657
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014542260257408625,
+      "loss": 0.288,
+      "step": 5658
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014540376074784512,
+      "loss": 0.3122,
+      "step": 5659
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014538491689090172,
+      "loss": 0.3112,
+      "step": 5660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001453660710040988,
+      "loss": 0.3142,
+      "step": 5661
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001453472230882793,
+      "loss": 0.3142,
+      "step": 5662
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001453283731442862,
+      "loss": 0.3047,
+      "step": 5663
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014530952117296258,
+      "loss": 0.3115,
+      "step": 5664
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001452906671751515,
+      "loss": 0.289,
+      "step": 5665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014527181115169633,
+      "loss": 0.2876,
+      "step": 5666
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001452529531034404,
+      "loss": 0.3133,
+      "step": 5667
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001452340930312271,
+      "loss": 0.3285,
+      "step": 5668
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014521523093589993,
+      "loss": 0.3166,
+      "step": 5669
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014519636681830256,
+      "loss": 0.3031,
+      "step": 5670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001451775006792787,
+      "loss": 0.2961,
+      "step": 5671
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014515863251967213,
+      "loss": 0.3132,
+      "step": 5672
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014513976234032676,
+      "loss": 0.2937,
+      "step": 5673
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014512089014208655,
+      "loss": 0.3018,
+      "step": 5674
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014510201592579552,
+      "loss": 0.3011,
+      "step": 5675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014508313969229786,
+      "loss": 0.2896,
+      "step": 5676
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014506426144243783,
+      "loss": 0.3037,
+      "step": 5677
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014504538117705978,
+      "loss": 0.2883,
+      "step": 5678
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014502649889700813,
+      "loss": 0.3083,
+      "step": 5679
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00014500761460312735,
+      "loss": 0.3021,
+      "step": 5680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001449887282962621,
+      "loss": 0.2907,
+      "step": 5681
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014496983997725708,
+      "loss": 0.295,
+      "step": 5682
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014495094964695705,
+      "loss": 0.2998,
+      "step": 5683
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001449320573062069,
+      "loss": 0.2919,
+      "step": 5684
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001449131629558516,
+      "loss": 0.2934,
+      "step": 5685
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014489426659673618,
+      "loss": 0.2907,
+      "step": 5686
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014487536822970585,
+      "loss": 0.305,
+      "step": 5687
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014485646785560578,
+      "loss": 0.305,
+      "step": 5688
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014483756547528136,
+      "loss": 0.3157,
+      "step": 5689
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014481866108957795,
+      "loss": 0.2785,
+      "step": 5690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001447997546993411,
+      "loss": 0.2905,
+      "step": 5691
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014478084630541635,
+      "loss": 0.2951,
+      "step": 5692
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014476193590864944,
+      "loss": 0.3103,
+      "step": 5693
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014474302350988614,
+      "loss": 0.281,
+      "step": 5694
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014472410910997232,
+      "loss": 0.2833,
+      "step": 5695
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014470519270975388,
+      "loss": 0.3084,
+      "step": 5696
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014468627431007692,
+      "loss": 0.3134,
+      "step": 5697
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014466735391178756,
+      "loss": 0.294,
+      "step": 5698
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014464843151573202,
+      "loss": 0.2893,
+      "step": 5699
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001446295071227566,
+      "loss": 0.2993,
+      "step": 5700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001446105807337077,
+      "loss": 0.2877,
+      "step": 5701
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001445916523494318,
+      "loss": 0.295,
+      "step": 5702
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014457272197077555,
+      "loss": 0.3055,
+      "step": 5703
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014455378959858558,
+      "loss": 0.2907,
+      "step": 5704
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014453485523370862,
+      "loss": 0.2858,
+      "step": 5705
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014451591887699154,
+      "loss": 0.2877,
+      "step": 5706
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014449698052928126,
+      "loss": 0.3034,
+      "step": 5707
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014447804019142484,
+      "loss": 0.2995,
+      "step": 5708
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014445909786426933,
+      "loss": 0.2922,
+      "step": 5709
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014444015354866203,
+      "loss": 0.2903,
+      "step": 5710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014442120724545015,
+      "loss": 0.2968,
+      "step": 5711
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00014440225895548113,
+      "loss": 0.2842,
+      "step": 5712
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014438330867960238,
+      "loss": 0.2911,
+      "step": 5713
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001443643564186615,
+      "loss": 0.285,
+      "step": 5714
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014434540217350613,
+      "loss": 0.2986,
+      "step": 5715
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014432644594498402,
+      "loss": 0.2693,
+      "step": 5716
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014430748773394293,
+      "loss": 0.2868,
+      "step": 5717
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014428852754123084,
+      "loss": 0.2951,
+      "step": 5718
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014426956536769579,
+      "loss": 0.2918,
+      "step": 5719
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014425060121418579,
+      "loss": 0.2954,
+      "step": 5720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014423163508154904,
+      "loss": 0.2893,
+      "step": 5721
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014421266697063381,
+      "loss": 0.2997,
+      "step": 5722
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014419369688228848,
+      "loss": 0.2893,
+      "step": 5723
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014417472481736146,
+      "loss": 0.3073,
+      "step": 5724
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014415575077670128,
+      "loss": 0.2853,
+      "step": 5725
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001441367747611566,
+      "loss": 0.2966,
+      "step": 5726
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014411779677157616,
+      "loss": 0.3112,
+      "step": 5727
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001440988168088087,
+      "loss": 0.2897,
+      "step": 5728
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014407983487370308,
+      "loss": 0.2843,
+      "step": 5729
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001440608509671083,
+      "loss": 0.2926,
+      "step": 5730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014404186508987346,
+      "loss": 0.2971,
+      "step": 5731
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001440228772428477,
+      "loss": 0.2908,
+      "step": 5732
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014400388742688024,
+      "loss": 0.2882,
+      "step": 5733
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014398489564282046,
+      "loss": 0.2815,
+      "step": 5734
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014396590189151765,
+      "loss": 0.2946,
+      "step": 5735
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014394690617382144,
+      "loss": 0.293,
+      "step": 5736
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014392790849058137,
+      "loss": 0.2695,
+      "step": 5737
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014390890884264713,
+      "loss": 0.2941,
+      "step": 5738
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014388990723086846,
+      "loss": 0.2848,
+      "step": 5739
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014387090365609525,
+      "loss": 0.2858,
+      "step": 5740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014385189811917746,
+      "loss": 0.2799,
+      "step": 5741
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014383289062096507,
+      "loss": 0.3103,
+      "step": 5742
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00014381388116230817,
+      "loss": 0.2956,
+      "step": 5743
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001437948697440571,
+      "loss": 0.2738,
+      "step": 5744
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014377585636706202,
+      "loss": 0.2963,
+      "step": 5745
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014375684103217339,
+      "loss": 0.3044,
+      "step": 5746
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001437378237402416,
+      "loss": 0.2817,
+      "step": 5747
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001437188044921173,
+      "loss": 0.2918,
+      "step": 5748
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014369978328865107,
+      "loss": 0.275,
+      "step": 5749
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014368076013069364,
+      "loss": 0.3013,
+      "step": 5750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001436617350190959,
+      "loss": 0.2855,
+      "step": 5751
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014364270795470867,
+      "loss": 0.2971,
+      "step": 5752
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014362367893838298,
+      "loss": 0.2663,
+      "step": 5753
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014360464797096993,
+      "loss": 0.2941,
+      "step": 5754
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014358561505332067,
+      "loss": 0.2781,
+      "step": 5755
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014356658018628644,
+      "loss": 0.2862,
+      "step": 5756
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001435475433707186,
+      "loss": 0.2773,
+      "step": 5757
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014352850460746855,
+      "loss": 0.2837,
+      "step": 5758
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014350946389738787,
+      "loss": 0.2755,
+      "step": 5759
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014349042124132808,
+      "loss": 0.294,
+      "step": 5760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014347137664014095,
+      "loss": 0.2848,
+      "step": 5761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014345233009467823,
+      "loss": 0.2772,
+      "step": 5762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014343328160579174,
+      "loss": 0.2977,
+      "step": 5763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014341423117433352,
+      "loss": 0.2842,
+      "step": 5764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014339517880115554,
+      "loss": 0.2875,
+      "step": 5765
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014337612448710996,
+      "loss": 0.3046,
+      "step": 5766
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014335706823304895,
+      "loss": 0.2802,
+      "step": 5767
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014333801003982484,
+      "loss": 0.2868,
+      "step": 5768
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014331894990829,
+      "loss": 0.2909,
+      "step": 5769
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014329988783929693,
+      "loss": 0.3049,
+      "step": 5770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001432808238336982,
+      "loss": 0.2792,
+      "step": 5771
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014326175789234637,
+      "loss": 0.2794,
+      "step": 5772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00014324269001609428,
+      "loss": 0.2888,
+      "step": 5773
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001432236202057947,
+      "loss": 0.2883,
+      "step": 5774
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001432045484623005,
+      "loss": 0.3076,
+      "step": 5775
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014318547478646472,
+      "loss": 0.2805,
+      "step": 5776
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014316639917914044,
+      "loss": 0.2995,
+      "step": 5777
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014314732164118077,
+      "loss": 0.2816,
+      "step": 5778
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014312824217343903,
+      "loss": 0.2947,
+      "step": 5779
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014310916077676851,
+      "loss": 0.297,
+      "step": 5780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014309007745202267,
+      "loss": 0.2954,
+      "step": 5781
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000143070992200055,
+      "loss": 0.3005,
+      "step": 5782
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001430519050217191,
+      "loss": 0.3091,
+      "step": 5783
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014303281591786866,
+      "loss": 0.2763,
+      "step": 5784
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001430137248893574,
+      "loss": 0.2965,
+      "step": 5785
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014299463193703923,
+      "loss": 0.2924,
+      "step": 5786
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014297553706176807,
+      "loss": 0.2903,
+      "step": 5787
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014295644026439795,
+      "loss": 0.2906,
+      "step": 5788
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014293734154578297,
+      "loss": 0.2766,
+      "step": 5789
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014291824090677737,
+      "loss": 0.2973,
+      "step": 5790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014289913834823536,
+      "loss": 0.2775,
+      "step": 5791
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001428800338710114,
+      "loss": 0.2785,
+      "step": 5792
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014286092747595988,
+      "loss": 0.2855,
+      "step": 5793
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014284181916393533,
+      "loss": 0.275,
+      "step": 5794
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014282270893579244,
+      "loss": 0.2886,
+      "step": 5795
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014280359679238588,
+      "loss": 0.2908,
+      "step": 5796
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014278448273457044,
+      "loss": 0.2948,
+      "step": 5797
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014276536676320105,
+      "loss": 0.2826,
+      "step": 5798
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014274624887913263,
+      "loss": 0.2939,
+      "step": 5799
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014272712908322029,
+      "loss": 0.2698,
+      "step": 5800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001427080073763191,
+      "loss": 0.2824,
+      "step": 5801
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014268888375928434,
+      "loss": 0.292,
+      "step": 5802
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014266975823297134,
+      "loss": 0.297,
+      "step": 5803
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00014265063079823542,
+      "loss": 0.2895,
+      "step": 5804
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014263150145593214,
+      "loss": 0.2842,
+      "step": 5805
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000142612370206917,
+      "loss": 0.2644,
+      "step": 5806
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001425932370520457,
+      "loss": 0.3033,
+      "step": 5807
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014257410199217398,
+      "loss": 0.2758,
+      "step": 5808
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014255496502815765,
+      "loss": 0.2648,
+      "step": 5809
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001425358261608526,
+      "loss": 0.2917,
+      "step": 5810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014251668539111485,
+      "loss": 0.2912,
+      "step": 5811
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014249754271980047,
+      "loss": 0.269,
+      "step": 5812
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014247839814776563,
+      "loss": 0.2777,
+      "step": 5813
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001424592516758666,
+      "loss": 0.2863,
+      "step": 5814
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014244010330495964,
+      "loss": 0.3014,
+      "step": 5815
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014242095303590126,
+      "loss": 0.2785,
+      "step": 5816
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001424018008695479,
+      "loss": 0.2803,
+      "step": 5817
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014238264680675616,
+      "loss": 0.2695,
+      "step": 5818
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014236349084838275,
+      "loss": 0.3036,
+      "step": 5819
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014234433299528437,
+      "loss": 0.2769,
+      "step": 5820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014232517324831793,
+      "loss": 0.2722,
+      "step": 5821
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001423060116083403,
+      "loss": 0.2933,
+      "step": 5822
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001422868480762085,
+      "loss": 0.2805,
+      "step": 5823
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001422676826527797,
+      "loss": 0.2904,
+      "step": 5824
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000142248515338911,
+      "loss": 0.2914,
+      "step": 5825
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014222934613545966,
+      "loss": 0.2807,
+      "step": 5826
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001422101750432831,
+      "loss": 0.2727,
+      "step": 5827
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014219100206323867,
+      "loss": 0.2887,
+      "step": 5828
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014217182719618396,
+      "loss": 0.2698,
+      "step": 5829
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014215265044297657,
+      "loss": 0.2781,
+      "step": 5830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014213347180447414,
+      "loss": 0.3,
+      "step": 5831
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014211429128153447,
+      "loss": 0.29,
+      "step": 5832
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001420951088750154,
+      "loss": 0.2773,
+      "step": 5833
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014207592458577494,
+      "loss": 0.2791,
+      "step": 5834
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00014205673841467104,
+      "loss": 0.2784,
+      "step": 5835
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014203755036256183,
+      "loss": 0.2744,
+      "step": 5836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001420183604303055,
+      "loss": 0.2785,
+      "step": 5837
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014199916861876036,
+      "loss": 0.2775,
+      "step": 5838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014197997492878473,
+      "loss": 0.2818,
+      "step": 5839
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014196077936123706,
+      "loss": 0.2774,
+      "step": 5840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001419415819169759,
+      "loss": 0.266,
+      "step": 5841
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014192238259685985,
+      "loss": 0.2891,
+      "step": 5842
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014190318140174763,
+      "loss": 0.287,
+      "step": 5843
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014188397833249803,
+      "loss": 0.2742,
+      "step": 5844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014186477338996986,
+      "loss": 0.2726,
+      "step": 5845
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001418455665750221,
+      "loss": 0.277,
+      "step": 5846
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014182635788851378,
+      "loss": 0.2837,
+      "step": 5847
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014180714733130404,
+      "loss": 0.2752,
+      "step": 5848
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000141787934904252,
+      "loss": 0.2738,
+      "step": 5849
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014176872060821707,
+      "loss": 0.2803,
+      "step": 5850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014174950444405853,
+      "loss": 0.2897,
+      "step": 5851
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014173028641263584,
+      "loss": 0.2702,
+      "step": 5852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014171106651480857,
+      "loss": 0.2804,
+      "step": 5853
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001416918447514363,
+      "loss": 0.288,
+      "step": 5854
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014167262112337874,
+      "loss": 0.2746,
+      "step": 5855
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001416533956314957,
+      "loss": 0.283,
+      "step": 5856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000141634168276647,
+      "loss": 0.2769,
+      "step": 5857
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014161493905969264,
+      "loss": 0.2882,
+      "step": 5858
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014159570798149263,
+      "loss": 0.2686,
+      "step": 5859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001415764750429071,
+      "loss": 0.2739,
+      "step": 5860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014155724024479625,
+      "loss": 0.2765,
+      "step": 5861
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014153800358802037,
+      "loss": 0.2748,
+      "step": 5862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00014151876507343978,
+      "loss": 0.2786,
+      "step": 5863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000141499524701915,
+      "loss": 0.2802,
+      "step": 5864
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001414802824743065,
+      "loss": 0.293,
+      "step": 5865
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014146103839147496,
+      "loss": 0.2615,
+      "step": 5866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014144179245428103,
+      "loss": 0.2715,
+      "step": 5867
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014142254466358552,
+      "loss": 0.2782,
+      "step": 5868
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014140329502024927,
+      "loss": 0.2889,
+      "step": 5869
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014138404352513327,
+      "loss": 0.2618,
+      "step": 5870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014136479017909853,
+      "loss": 0.2668,
+      "step": 5871
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001413455349830061,
+      "loss": 0.2907,
+      "step": 5872
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014132627793771727,
+      "loss": 0.2802,
+      "step": 5873
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014130701904409328,
+      "loss": 0.2944,
+      "step": 5874
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014128775830299551,
+      "loss": 0.287,
+      "step": 5875
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014126849571528542,
+      "loss": 0.2783,
+      "step": 5876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014124923128182445,
+      "loss": 0.2845,
+      "step": 5877
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014122996500347433,
+      "loss": 0.2681,
+      "step": 5878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014121069688109665,
+      "loss": 0.2851,
+      "step": 5879
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014119142691555324,
+      "loss": 0.3051,
+      "step": 5880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001411721551077059,
+      "loss": 0.2909,
+      "step": 5881
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014115288145841667,
+      "loss": 0.2892,
+      "step": 5882
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001411336059685475,
+      "loss": 0.2857,
+      "step": 5883
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014111432863896052,
+      "loss": 0.2785,
+      "step": 5884
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001410950494705179,
+      "loss": 0.2787,
+      "step": 5885
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001410757684640819,
+      "loss": 0.2797,
+      "step": 5886
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001410564856205149,
+      "loss": 0.2858,
+      "step": 5887
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014103720094067935,
+      "loss": 0.3023,
+      "step": 5888
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014101791442543767,
+      "loss": 0.2823,
+      "step": 5889
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014099862607565256,
+      "loss": 0.287,
+      "step": 5890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014097933589218666,
+      "loss": 0.2861,
+      "step": 5891
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014096004387590275,
+      "loss": 0.2545,
+      "step": 5892
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014094075002766364,
+      "loss": 0.2818,
+      "step": 5893
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014092145434833227,
+      "loss": 0.281,
+      "step": 5894
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001409021568387717,
+      "loss": 0.2852,
+      "step": 5895
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00014088285749984492,
+      "loss": 0.2879,
+      "step": 5896
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014086355633241514,
+      "loss": 0.271,
+      "step": 5897
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014084425333734564,
+      "loss": 0.2887,
+      "step": 5898
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014082494851549974,
+      "loss": 0.2802,
+      "step": 5899
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014080564186774085,
+      "loss": 0.2759,
+      "step": 5900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001407863333949325,
+      "loss": 0.2724,
+      "step": 5901
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001407670230979382,
+      "loss": 0.2761,
+      "step": 5902
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014074771097762168,
+      "loss": 0.2727,
+      "step": 5903
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014072839703484666,
+      "loss": 0.2747,
+      "step": 5904
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014070908127047696,
+      "loss": 0.2703,
+      "step": 5905
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014068976368537646,
+      "loss": 0.272,
+      "step": 5906
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001406704442804092,
+      "loss": 0.2944,
+      "step": 5907
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001406511230564392,
+      "loss": 0.2611,
+      "step": 5908
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014063180001433064,
+      "loss": 0.2863,
+      "step": 5909
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014061247515494772,
+      "loss": 0.2731,
+      "step": 5910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001405931484791548,
+      "loss": 0.2767,
+      "step": 5911
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014057381998781624,
+      "loss": 0.269,
+      "step": 5912
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014055448968179653,
+      "loss": 0.3029,
+      "step": 5913
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014053515756196026,
+      "loss": 0.272,
+      "step": 5914
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014051582362917196,
+      "loss": 0.2759,
+      "step": 5915
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014049648788429646,
+      "loss": 0.2704,
+      "step": 5916
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001404771503281985,
+      "loss": 0.2612,
+      "step": 5917
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014045781096174296,
+      "loss": 0.2757,
+      "step": 5918
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014043846978579482,
+      "loss": 0.2735,
+      "step": 5919
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014041912680121913,
+      "loss": 0.2713,
+      "step": 5920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014039978200888103,
+      "loss": 0.289,
+      "step": 5921
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014038043540964563,
+      "loss": 0.2675,
+      "step": 5922
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014036108700437835,
+      "loss": 0.2789,
+      "step": 5923
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014034173679394445,
+      "loss": 0.2689,
+      "step": 5924
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001403223847792094,
+      "loss": 0.2793,
+      "step": 5925
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00014030303096103873,
+      "loss": 0.2882,
+      "step": 5926
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001402836753402981,
+      "loss": 0.2851,
+      "step": 5927
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014026431791785312,
+      "loss": 0.2801,
+      "step": 5928
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001402449586945696,
+      "loss": 0.2701,
+      "step": 5929
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014022559767131338,
+      "loss": 0.2643,
+      "step": 5930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001402062348489504,
+      "loss": 0.2723,
+      "step": 5931
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014018687022834667,
+      "loss": 0.2867,
+      "step": 5932
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014016750381036828,
+      "loss": 0.3103,
+      "step": 5933
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001401481355958814,
+      "loss": 0.2749,
+      "step": 5934
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014012876558575226,
+      "loss": 0.2966,
+      "step": 5935
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014010939378084723,
+      "loss": 0.3044,
+      "step": 5936
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014009002018203266,
+      "loss": 0.3084,
+      "step": 5937
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001400706447901751,
+      "loss": 0.3319,
+      "step": 5938
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001400512676061411,
+      "loss": 0.2876,
+      "step": 5939
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014003188863079733,
+      "loss": 0.3129,
+      "step": 5940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00014001250786501053,
+      "loss": 0.3032,
+      "step": 5941
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013999312530964746,
+      "loss": 0.2978,
+      "step": 5942
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013997374096557508,
+      "loss": 0.288,
+      "step": 5943
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001399543548336603,
+      "loss": 0.307,
+      "step": 5944
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001399349669147702,
+      "loss": 0.2974,
+      "step": 5945
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013991557720977195,
+      "loss": 0.2871,
+      "step": 5946
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013989618571953272,
+      "loss": 0.3001,
+      "step": 5947
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001398767924449198,
+      "loss": 0.2889,
+      "step": 5948
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001398573973868006,
+      "loss": 0.2843,
+      "step": 5949
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001398380005460425,
+      "loss": 0.2791,
+      "step": 5950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013981860192351312,
+      "loss": 0.278,
+      "step": 5951
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013979920152008,
+      "loss": 0.2716,
+      "step": 5952
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013977979933661085,
+      "loss": 0.2819,
+      "step": 5953
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001397603953739735,
+      "loss": 0.2787,
+      "step": 5954
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001397409896330357,
+      "loss": 0.2827,
+      "step": 5955
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013972158211466546,
+      "loss": 0.2813,
+      "step": 5956
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00013970217281973076,
+      "loss": 0.2644,
+      "step": 5957
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013968276174909967,
+      "loss": 0.2831,
+      "step": 5958
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001396633489036404,
+      "loss": 0.2775,
+      "step": 5959
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013964393428422115,
+      "loss": 0.2834,
+      "step": 5960
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001396245178917103,
+      "loss": 0.2766,
+      "step": 5961
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001396050997269762,
+      "loss": 0.2874,
+      "step": 5962
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013958567979088739,
+      "loss": 0.2879,
+      "step": 5963
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001395662580843124,
+      "loss": 0.2637,
+      "step": 5964
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013954683460811991,
+      "loss": 0.28,
+      "step": 5965
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013952740936317858,
+      "loss": 0.2839,
+      "step": 5966
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013950798235035726,
+      "loss": 0.2582,
+      "step": 5967
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013948855357052482,
+      "loss": 0.2723,
+      "step": 5968
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013946912302455018,
+      "loss": 0.2715,
+      "step": 5969
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013944969071330248,
+      "loss": 0.286,
+      "step": 5970
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013943025663765076,
+      "loss": 0.2629,
+      "step": 5971
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001394108207984642,
+      "loss": 0.2805,
+      "step": 5972
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013939138319661219,
+      "loss": 0.2956,
+      "step": 5973
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013937194383296396,
+      "loss": 0.2832,
+      "step": 5974
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013935250270838897,
+      "loss": 0.2757,
+      "step": 5975
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013933305982375676,
+      "loss": 0.2604,
+      "step": 5976
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013931361517993695,
+      "loss": 0.2743,
+      "step": 5977
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013929416877779915,
+      "loss": 0.2672,
+      "step": 5978
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013927472061821313,
+      "loss": 0.2718,
+      "step": 5979
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013925527070204872,
+      "loss": 0.2716,
+      "step": 5980
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013923581903017582,
+      "loss": 0.259,
+      "step": 5981
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013921636560346442,
+      "loss": 0.266,
+      "step": 5982
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013919691042278457,
+      "loss": 0.2741,
+      "step": 5983
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013917745348900645,
+      "loss": 0.2601,
+      "step": 5984
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013915799480300028,
+      "loss": 0.2809,
+      "step": 5985
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013913853436563626,
+      "loss": 0.2721,
+      "step": 5986
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013911907217778487,
+      "loss": 0.2715,
+      "step": 5987
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013909960824031654,
+      "loss": 0.2758,
+      "step": 5988
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013908014255410177,
+      "loss": 0.2895,
+      "step": 5989
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001390606751200112,
+      "loss": 0.275,
+      "step": 5990
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001390412059389155,
+      "loss": 0.271,
+      "step": 5991
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001390217350116855,
+      "loss": 0.2862,
+      "step": 5992
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013900226233919195,
+      "loss": 0.2737,
+      "step": 5993
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013898278792230588,
+      "loss": 0.2827,
+      "step": 5994
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013896331176189818,
+      "loss": 0.2809,
+      "step": 5995
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013894383385883998,
+      "loss": 0.2844,
+      "step": 5996
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013892435421400246,
+      "loss": 0.2682,
+      "step": 5997
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013890487282825682,
+      "loss": 0.2858,
+      "step": 5998
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001388853897024744,
+      "loss": 0.2656,
+      "step": 5999
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013886590483752655,
+      "loss": 0.2562,
+      "step": 6000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001388464182342848,
+      "loss": 0.274,
+      "step": 6001
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013882692989362062,
+      "loss": 0.2663,
+      "step": 6002
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013880743981640572,
+      "loss": 0.2796,
+      "step": 6003
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013878794800351176,
+      "loss": 0.2746,
+      "step": 6004
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001387684544558105,
+      "loss": 0.283,
+      "step": 6005
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013874895917417382,
+      "loss": 0.2598,
+      "step": 6006
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013872946215947365,
+      "loss": 0.2754,
+      "step": 6007
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000138709963412582,
+      "loss": 0.2765,
+      "step": 6008
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013869046293437097,
+      "loss": 0.2717,
+      "step": 6009
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013867096072571271,
+      "loss": 0.2765,
+      "step": 6010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013865145678747948,
+      "loss": 0.2758,
+      "step": 6011
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001386319511205436,
+      "loss": 0.26,
+      "step": 6012
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013861244372577746,
+      "loss": 0.2826,
+      "step": 6013
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013859293460405356,
+      "loss": 0.2898,
+      "step": 6014
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001385734237562444,
+      "loss": 0.2754,
+      "step": 6015
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013855391118322268,
+      "loss": 0.2552,
+      "step": 6016
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013853439688586106,
+      "loss": 0.2748,
+      "step": 6017
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013851488086503232,
+      "loss": 0.2608,
+      "step": 6018
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013849536312160937,
+      "loss": 0.2693,
+      "step": 6019
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013847584365646513,
+      "loss": 0.2753,
+      "step": 6020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013845632247047257,
+      "loss": 0.2618,
+      "step": 6021
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013843679956450483,
+      "loss": 0.2652,
+      "step": 6022
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001384172749394351,
+      "loss": 0.2528,
+      "step": 6023
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013839774859613656,
+      "loss": 0.2665,
+      "step": 6024
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001383782205354826,
+      "loss": 0.2693,
+      "step": 6025
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001383586907583466,
+      "loss": 0.2756,
+      "step": 6026
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000138339159265602,
+      "loss": 0.269,
+      "step": 6027
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001383196260581224,
+      "loss": 0.2822,
+      "step": 6028
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001383000911367814,
+      "loss": 0.2487,
+      "step": 6029
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013828055450245273,
+      "loss": 0.2696,
+      "step": 6030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013826101615601016,
+      "loss": 0.2568,
+      "step": 6031
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013824147609832753,
+      "loss": 0.2654,
+      "step": 6032
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013822193433027887,
+      "loss": 0.2926,
+      "step": 6033
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001382023908527381,
+      "loss": 0.2718,
+      "step": 6034
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013818284566657934,
+      "loss": 0.2846,
+      "step": 6035
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013816329877267672,
+      "loss": 0.2623,
+      "step": 6036
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013814375017190457,
+      "loss": 0.276,
+      "step": 6037
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013812419986513713,
+      "loss": 0.2799,
+      "step": 6038
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013810464785324884,
+      "loss": 0.2897,
+      "step": 6039
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013808509413711412,
+      "loss": 0.2725,
+      "step": 6040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001380655387176076,
+      "loss": 0.2779,
+      "step": 6041
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013804598159560381,
+      "loss": 0.2795,
+      "step": 6042
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013802642277197755,
+      "loss": 0.2886,
+      "step": 6043
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013800686224760354,
+      "loss": 0.2647,
+      "step": 6044
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013798730002335661,
+      "loss": 0.2666,
+      "step": 6045
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001379677361001117,
+      "loss": 0.2741,
+      "step": 6046
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013794817047874388,
+      "loss": 0.2649,
+      "step": 6047
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00013792860316012815,
+      "loss": 0.2571,
+      "step": 6048
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001379090341451397,
+      "loss": 0.2683,
+      "step": 6049
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013788946343465378,
+      "loss": 0.2609,
+      "step": 6050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013786989102954563,
+      "loss": 0.2626,
+      "step": 6051
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013785031693069072,
+      "loss": 0.2808,
+      "step": 6052
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001378307411389645,
+      "loss": 0.2487,
+      "step": 6053
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013781116365524246,
+      "loss": 0.2932,
+      "step": 6054
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001377915844804002,
+      "loss": 0.2655,
+      "step": 6055
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013777200361531346,
+      "loss": 0.2829,
+      "step": 6056
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013775242106085797,
+      "loss": 0.2862,
+      "step": 6057
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013773283681790956,
+      "loss": 0.258,
+      "step": 6058
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013771325088734419,
+      "loss": 0.2606,
+      "step": 6059
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001376936632700378,
+      "loss": 0.262,
+      "step": 6060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001376740739668665,
+      "loss": 0.2929,
+      "step": 6061
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013765448297870637,
+      "loss": 0.2661,
+      "step": 6062
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001376348903064337,
+      "loss": 0.2673,
+      "step": 6063
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001376152959509247,
+      "loss": 0.2711,
+      "step": 6064
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001375956999130558,
+      "loss": 0.2543,
+      "step": 6065
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001375761021937034,
+      "loss": 0.2512,
+      "step": 6066
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013755650279374403,
+      "loss": 0.2703,
+      "step": 6067
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001375369017140543,
+      "loss": 0.2726,
+      "step": 6068
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013751729895551085,
+      "loss": 0.2855,
+      "step": 6069
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013749769451899043,
+      "loss": 0.2665,
+      "step": 6070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013747808840536987,
+      "loss": 0.267,
+      "step": 6071
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013745848061552605,
+      "loss": 0.2734,
+      "step": 6072
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001374388711503359,
+      "loss": 0.2664,
+      "step": 6073
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013741926001067654,
+      "loss": 0.258,
+      "step": 6074
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013739964719742504,
+      "loss": 0.2694,
+      "step": 6075
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013738003271145859,
+      "loss": 0.2724,
+      "step": 6076
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013736041655365447,
+      "loss": 0.2637,
+      "step": 6077
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013734079872488998,
+      "loss": 0.2697,
+      "step": 6078
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001373211792260426,
+      "loss": 0.2584,
+      "step": 6079
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00013730155805798976,
+      "loss": 0.2714,
+      "step": 6080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001372819352216091,
+      "loss": 0.2642,
+      "step": 6081
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013726231071777817,
+      "loss": 0.2639,
+      "step": 6082
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013724268454737474,
+      "loss": 0.2556,
+      "step": 6083
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001372230567112766,
+      "loss": 0.2605,
+      "step": 6084
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013720342721036162,
+      "loss": 0.2774,
+      "step": 6085
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013718379604550767,
+      "loss": 0.2688,
+      "step": 6086
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013716416321759285,
+      "loss": 0.2616,
+      "step": 6087
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001371445287274952,
+      "loss": 0.2743,
+      "step": 6088
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013712489257609284,
+      "loss": 0.2742,
+      "step": 6089
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013710525476426407,
+      "loss": 0.2619,
+      "step": 6090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001370856152928872,
+      "loss": 0.2707,
+      "step": 6091
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001370659741628406,
+      "loss": 0.2637,
+      "step": 6092
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013704633137500273,
+      "loss": 0.266,
+      "step": 6093
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001370266869302521,
+      "loss": 0.2481,
+      "step": 6094
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013700704082946733,
+      "loss": 0.266,
+      "step": 6095
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001369873930735271,
+      "loss": 0.2471,
+      "step": 6096
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013696774366331018,
+      "loss": 0.2538,
+      "step": 6097
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001369480925996954,
+      "loss": 0.265,
+      "step": 6098
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001369284398835616,
+      "loss": 0.2652,
+      "step": 6099
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013690878551578782,
+      "loss": 0.26,
+      "step": 6100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013688912949725307,
+      "loss": 0.2589,
+      "step": 6101
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013686947182883652,
+      "loss": 0.278,
+      "step": 6102
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013684981251141734,
+      "loss": 0.2643,
+      "step": 6103
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013683015154587478,
+      "loss": 0.2644,
+      "step": 6104
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013681048893308826,
+      "loss": 0.2615,
+      "step": 6105
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001367908246739371,
+      "loss": 0.2781,
+      "step": 6106
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013677115876930084,
+      "loss": 0.266,
+      "step": 6107
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013675149122005905,
+      "loss": 0.2631,
+      "step": 6108
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013673182202709138,
+      "loss": 0.2531,
+      "step": 6109
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00013671215119127753,
+      "loss": 0.2819,
+      "step": 6110
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013669247871349728,
+      "loss": 0.2894,
+      "step": 6111
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013667280459463049,
+      "loss": 0.2701,
+      "step": 6112
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013665312883555709,
+      "loss": 0.272,
+      "step": 6113
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013663345143715712,
+      "loss": 0.2934,
+      "step": 6114
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013661377240031062,
+      "loss": 0.2819,
+      "step": 6115
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013659409172589774,
+      "loss": 0.2812,
+      "step": 6116
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001365744094147987,
+      "loss": 0.2827,
+      "step": 6117
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013655472546789387,
+      "loss": 0.2636,
+      "step": 6118
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013653503988606355,
+      "loss": 0.283,
+      "step": 6119
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013651535267018825,
+      "loss": 0.2781,
+      "step": 6120
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001364956638211484,
+      "loss": 0.2908,
+      "step": 6121
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013647597333982467,
+      "loss": 0.2731,
+      "step": 6122
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001364562812270977,
+      "loss": 0.2718,
+      "step": 6123
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013643658748384825,
+      "loss": 0.2839,
+      "step": 6124
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013641689211095706,
+      "loss": 0.2665,
+      "step": 6125
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013639719510930508,
+      "loss": 0.2727,
+      "step": 6126
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013637749647977323,
+      "loss": 0.2855,
+      "step": 6127
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013635779622324257,
+      "loss": 0.2841,
+      "step": 6128
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013633809434059418,
+      "loss": 0.2694,
+      "step": 6129
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013631839083270926,
+      "loss": 0.2717,
+      "step": 6130
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013629868570046904,
+      "loss": 0.2563,
+      "step": 6131
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013627897894475483,
+      "loss": 0.2714,
+      "step": 6132
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013625927056644806,
+      "loss": 0.2794,
+      "step": 6133
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013623956056643016,
+      "loss": 0.2689,
+      "step": 6134
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013621984894558267,
+      "loss": 0.2622,
+      "step": 6135
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001362001357047872,
+      "loss": 0.2591,
+      "step": 6136
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013618042084492546,
+      "loss": 0.2672,
+      "step": 6137
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013616070436687916,
+      "loss": 0.2606,
+      "step": 6138
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001361409862715302,
+      "loss": 0.2694,
+      "step": 6139
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013612126655976038,
+      "loss": 0.2629,
+      "step": 6140
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00013610154523245176,
+      "loss": 0.2735,
+      "step": 6141
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013608182229048635,
+      "loss": 0.2806,
+      "step": 6142
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013606209773474624,
+      "loss": 0.2518,
+      "step": 6143
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013604237156611372,
+      "loss": 0.2646,
+      "step": 6144
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013602264378547094,
+      "loss": 0.2678,
+      "step": 6145
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013600291439370025,
+      "loss": 0.2636,
+      "step": 6146
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001359831833916841,
+      "loss": 0.26,
+      "step": 6147
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013596345078030494,
+      "loss": 0.271,
+      "step": 6148
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013594371656044533,
+      "loss": 0.2697,
+      "step": 6149
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013592398073298786,
+      "loss": 0.2699,
+      "step": 6150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001359042432988153,
+      "loss": 0.2653,
+      "step": 6151
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013588450425881031,
+      "loss": 0.2777,
+      "step": 6152
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001358647636138558,
+      "loss": 0.2662,
+      "step": 6153
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001358450213648347,
+      "loss": 0.2662,
+      "step": 6154
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001358252775126299,
+      "loss": 0.2588,
+      "step": 6155
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013580553205812452,
+      "loss": 0.2527,
+      "step": 6156
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013578578500220167,
+      "loss": 0.2511,
+      "step": 6157
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013576603634574453,
+      "loss": 0.2623,
+      "step": 6158
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013574628608963637,
+      "loss": 0.273,
+      "step": 6159
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013572653423476056,
+      "loss": 0.2763,
+      "step": 6160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013570678078200047,
+      "loss": 0.2673,
+      "step": 6161
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001356870257322396,
+      "loss": 0.2896,
+      "step": 6162
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013566726908636152,
+      "loss": 0.2579,
+      "step": 6163
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013564751084524983,
+      "loss": 0.2763,
+      "step": 6164
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013562775100978822,
+      "loss": 0.2539,
+      "step": 6165
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013560798958086047,
+      "loss": 0.2686,
+      "step": 6166
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013558822655935043,
+      "loss": 0.2373,
+      "step": 6167
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013556846194614196,
+      "loss": 0.2639,
+      "step": 6168
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001355486957421191,
+      "loss": 0.2841,
+      "step": 6169
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001355289279481659,
+      "loss": 0.2529,
+      "step": 6170
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00013550915856516644,
+      "loss": 0.2443,
+      "step": 6171
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001354893875940049,
+      "loss": 0.2515,
+      "step": 6172
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013546961503556562,
+      "loss": 0.2494,
+      "step": 6173
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013544984089073288,
+      "loss": 0.269,
+      "step": 6174
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013543006516039113,
+      "loss": 0.2544,
+      "step": 6175
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013541028784542483,
+      "loss": 0.2531,
+      "step": 6176
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013539050894671847,
+      "loss": 0.2409,
+      "step": 6177
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013537072846515674,
+      "loss": 0.2534,
+      "step": 6178
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013535094640162432,
+      "loss": 0.2618,
+      "step": 6179
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013533116275700593,
+      "loss": 0.2545,
+      "step": 6180
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013531137753218647,
+      "loss": 0.267,
+      "step": 6181
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001352915907280508,
+      "loss": 0.2681,
+      "step": 6182
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013527180234548388,
+      "loss": 0.265,
+      "step": 6183
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001352520123853708,
+      "loss": 0.2601,
+      "step": 6184
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001352322208485966,
+      "loss": 0.2604,
+      "step": 6185
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013521242773604654,
+      "loss": 0.2552,
+      "step": 6186
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013519263304860582,
+      "loss": 0.266,
+      "step": 6187
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013517283678715983,
+      "loss": 0.2769,
+      "step": 6188
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013515303895259392,
+      "loss": 0.2852,
+      "step": 6189
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013513323954579353,
+      "loss": 0.2684,
+      "step": 6190
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013511343856764424,
+      "loss": 0.2728,
+      "step": 6191
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013509363601903166,
+      "loss": 0.2799,
+      "step": 6192
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013507383190084147,
+      "loss": 0.2775,
+      "step": 6193
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013505402621395943,
+      "loss": 0.2852,
+      "step": 6194
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013503421895927126,
+      "loss": 0.2513,
+      "step": 6195
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013501441013766296,
+      "loss": 0.2802,
+      "step": 6196
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013499459975002044,
+      "loss": 0.2584,
+      "step": 6197
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013497478779722974,
+      "loss": 0.2868,
+      "step": 6198
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013495497428017694,
+      "loss": 0.2622,
+      "step": 6199
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013493515919974823,
+      "loss": 0.2719,
+      "step": 6200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013491534255682982,
+      "loss": 0.2715,
+      "step": 6201
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00013489552435230802,
+      "loss": 0.2682,
+      "step": 6202
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013487570458706926,
+      "loss": 0.2596,
+      "step": 6203
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013485588326199997,
+      "loss": 0.2525,
+      "step": 6204
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013483606037798662,
+      "loss": 0.2559,
+      "step": 6205
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001348162359359158,
+      "loss": 0.2491,
+      "step": 6206
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013479640993667422,
+      "loss": 0.2744,
+      "step": 6207
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013477658238114857,
+      "loss": 0.2621,
+      "step": 6208
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013475675327022566,
+      "loss": 0.2542,
+      "step": 6209
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013473692260479233,
+      "loss": 0.2453,
+      "step": 6210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013471709038573554,
+      "loss": 0.2503,
+      "step": 6211
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001346972566139423,
+      "loss": 0.2699,
+      "step": 6212
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013467742129029964,
+      "loss": 0.2572,
+      "step": 6213
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013465758441569481,
+      "loss": 0.2584,
+      "step": 6214
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013463774599101487,
+      "loss": 0.2697,
+      "step": 6215
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001346179060171472,
+      "loss": 0.2637,
+      "step": 6216
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013459806449497914,
+      "loss": 0.2623,
+      "step": 6217
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001345782214253981,
+      "loss": 0.2622,
+      "step": 6218
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013455837680929154,
+      "loss": 0.2512,
+      "step": 6219
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013453853064754708,
+      "loss": 0.2493,
+      "step": 6220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001345186829410523,
+      "loss": 0.2587,
+      "step": 6221
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013449883369069491,
+      "loss": 0.2406,
+      "step": 6222
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001344789828973627,
+      "loss": 0.2613,
+      "step": 6223
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001344591305619435,
+      "loss": 0.2577,
+      "step": 6224
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013443927668532516,
+      "loss": 0.2627,
+      "step": 6225
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001344194212683957,
+      "loss": 0.2628,
+      "step": 6226
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013439956431204316,
+      "loss": 0.2521,
+      "step": 6227
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013437970581715565,
+      "loss": 0.2473,
+      "step": 6228
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013435984578462134,
+      "loss": 0.2588,
+      "step": 6229
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013433998421532846,
+      "loss": 0.2639,
+      "step": 6230
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013432012111016538,
+      "loss": 0.2493,
+      "step": 6231
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013430025647002044,
+      "loss": 0.253,
+      "step": 6232
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00013428039029578213,
+      "loss": 0.281,
+      "step": 6233
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013426052258833895,
+      "loss": 0.2591,
+      "step": 6234
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013424065334857946,
+      "loss": 0.2542,
+      "step": 6235
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001342207825773924,
+      "loss": 0.2685,
+      "step": 6236
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001342009102756664,
+      "loss": 0.2669,
+      "step": 6237
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013418103644429033,
+      "loss": 0.2613,
+      "step": 6238
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013416116108415306,
+      "loss": 0.2486,
+      "step": 6239
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013414128419614348,
+      "loss": 0.2501,
+      "step": 6240
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001341214057811506,
+      "loss": 0.2626,
+      "step": 6241
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013410152584006352,
+      "loss": 0.2717,
+      "step": 6242
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001340816443737714,
+      "loss": 0.2653,
+      "step": 6243
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013406176138316332,
+      "loss": 0.249,
+      "step": 6244
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013404187686912869,
+      "loss": 0.261,
+      "step": 6245
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001340219908325568,
+      "loss": 0.2426,
+      "step": 6246
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001340021032743371,
+      "loss": 0.261,
+      "step": 6247
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013398221419535898,
+      "loss": 0.2593,
+      "step": 6248
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013396232359651206,
+      "loss": 0.2532,
+      "step": 6249
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013394243147868594,
+      "loss": 0.2509,
+      "step": 6250
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013392253784277034,
+      "loss": 0.2776,
+      "step": 6251
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013390264268965495,
+      "loss": 0.2827,
+      "step": 6252
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013388274602022958,
+      "loss": 0.2793,
+      "step": 6253
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013386284783538417,
+      "loss": 0.2755,
+      "step": 6254
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013384294813600868,
+      "loss": 0.2706,
+      "step": 6255
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001338230469229931,
+      "loss": 0.2612,
+      "step": 6256
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013380314419722747,
+      "loss": 0.2741,
+      "step": 6257
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013378323995960208,
+      "loss": 0.272,
+      "step": 6258
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013376333421100705,
+      "loss": 0.2676,
+      "step": 6259
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001337434269523327,
+      "loss": 0.2521,
+      "step": 6260
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013372351818446938,
+      "loss": 0.245,
+      "step": 6261
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00013370360790830753,
+      "loss": 0.2435,
+      "step": 6262
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001336836961247377,
+      "loss": 0.2536,
+      "step": 6263
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013366378283465036,
+      "loss": 0.2593,
+      "step": 6264
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001336438680389362,
+      "loss": 0.2647,
+      "step": 6265
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013362395173848585,
+      "loss": 0.256,
+      "step": 6266
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013360403393419018,
+      "loss": 0.2743,
+      "step": 6267
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013358411462693993,
+      "loss": 0.2721,
+      "step": 6268
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013356419381762604,
+      "loss": 0.259,
+      "step": 6269
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013354427150713945,
+      "loss": 0.2441,
+      "step": 6270
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013352434769637125,
+      "loss": 0.2549,
+      "step": 6271
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013350442238621251,
+      "loss": 0.2633,
+      "step": 6272
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013348449557755435,
+      "loss": 0.2659,
+      "step": 6273
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001334645672712881,
+      "loss": 0.2528,
+      "step": 6274
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000133444637468305,
+      "loss": 0.2575,
+      "step": 6275
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001334247061694964,
+      "loss": 0.2601,
+      "step": 6276
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001334047733757538,
+      "loss": 0.2525,
+      "step": 6277
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013338483908796863,
+      "loss": 0.2562,
+      "step": 6278
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013336490330703252,
+      "loss": 0.2642,
+      "step": 6279
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013334496603383705,
+      "loss": 0.2715,
+      "step": 6280
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000133325027269274,
+      "loss": 0.2639,
+      "step": 6281
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013330508701423508,
+      "loss": 0.2575,
+      "step": 6282
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013328514526961214,
+      "loss": 0.2488,
+      "step": 6283
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013326520203629712,
+      "loss": 0.2374,
+      "step": 6284
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001332452573151819,
+      "loss": 0.2459,
+      "step": 6285
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001332253111071586,
+      "loss": 0.2461,
+      "step": 6286
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013320536341311927,
+      "loss": 0.2506,
+      "step": 6287
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001331854142339561,
+      "loss": 0.2618,
+      "step": 6288
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013316546357056134,
+      "loss": 0.2419,
+      "step": 6289
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013314551142382726,
+      "loss": 0.2509,
+      "step": 6290
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00013312555779464627,
+      "loss": 0.2599,
+      "step": 6291
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001331056026839108,
+      "loss": 0.2585,
+      "step": 6292
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001330856460925133,
+      "loss": 0.2707,
+      "step": 6293
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001330656880213464,
+      "loss": 0.2565,
+      "step": 6294
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001330457284713027,
+      "loss": 0.2516,
+      "step": 6295
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013302576744327486,
+      "loss": 0.2538,
+      "step": 6296
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001330058049381557,
+      "loss": 0.2407,
+      "step": 6297
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013298584095683806,
+      "loss": 0.2521,
+      "step": 6298
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001329658755002148,
+      "loss": 0.2453,
+      "step": 6299
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001329459085691789,
+      "loss": 0.2408,
+      "step": 6300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001329259401646234,
+      "loss": 0.2613,
+      "step": 6301
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013290597028744135,
+      "loss": 0.2594,
+      "step": 6302
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013288599893852599,
+      "loss": 0.2364,
+      "step": 6303
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013286602611877052,
+      "loss": 0.2711,
+      "step": 6304
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013284605182906816,
+      "loss": 0.2419,
+      "step": 6305
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013282607607031233,
+      "loss": 0.2552,
+      "step": 6306
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013280609884339646,
+      "loss": 0.2589,
+      "step": 6307
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000132786120149214,
+      "loss": 0.2663,
+      "step": 6308
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013276613998865856,
+      "loss": 0.2779,
+      "step": 6309
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013274615836262373,
+      "loss": 0.2754,
+      "step": 6310
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001327261752720032,
+      "loss": 0.2609,
+      "step": 6311
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013270619071769076,
+      "loss": 0.2662,
+      "step": 6312
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013268620470058016,
+      "loss": 0.2601,
+      "step": 6313
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001326662172215653,
+      "loss": 0.2626,
+      "step": 6314
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013264622828154017,
+      "loss": 0.2713,
+      "step": 6315
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013262623788139872,
+      "loss": 0.2566,
+      "step": 6316
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013260624602203508,
+      "loss": 0.2667,
+      "step": 6317
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013258625270434337,
+      "loss": 0.2655,
+      "step": 6318
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013256625792921784,
+      "loss": 0.2567,
+      "step": 6319
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013254626169755272,
+      "loss": 0.2681,
+      "step": 6320
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013252626401024235,
+      "loss": 0.2449,
+      "step": 6321
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013250626486818113,
+      "loss": 0.2665,
+      "step": 6322
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013248626427226357,
+      "loss": 0.2489,
+      "step": 6323
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00013246626222338418,
+      "loss": 0.2612,
+      "step": 6324
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001324462587224376,
+      "loss": 0.2583,
+      "step": 6325
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013242625377031842,
+      "loss": 0.2614,
+      "step": 6326
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013240624736792143,
+      "loss": 0.2628,
+      "step": 6327
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013238623951614137,
+      "loss": 0.2314,
+      "step": 6328
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013236623021587317,
+      "loss": 0.2382,
+      "step": 6329
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013234621946801167,
+      "loss": 0.2469,
+      "step": 6330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013232620727345192,
+      "loss": 0.2641,
+      "step": 6331
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013230619363308898,
+      "loss": 0.2569,
+      "step": 6332
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013228617854781793,
+      "loss": 0.2568,
+      "step": 6333
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000132266162018534,
+      "loss": 0.2569,
+      "step": 6334
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013224614404613237,
+      "loss": 0.2383,
+      "step": 6335
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013222612463150842,
+      "loss": 0.2687,
+      "step": 6336
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013220610377555748,
+      "loss": 0.2509,
+      "step": 6337
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000132186081479175,
+      "loss": 0.2742,
+      "step": 6338
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013216605774325651,
+      "loss": 0.2572,
+      "step": 6339
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013214603256869758,
+      "loss": 0.2561,
+      "step": 6340
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013212600595639381,
+      "loss": 0.2688,
+      "step": 6341
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013210597790724093,
+      "loss": 0.2542,
+      "step": 6342
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013208594842213466,
+      "loss": 0.2626,
+      "step": 6343
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001320659175019709,
+      "loss": 0.244,
+      "step": 6344
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001320458851476455,
+      "loss": 0.2564,
+      "step": 6345
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001320258513600544,
+      "loss": 0.2561,
+      "step": 6346
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001320058161400936,
+      "loss": 0.2434,
+      "step": 6347
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013198577948865924,
+      "loss": 0.2555,
+      "step": 6348
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013196574140664745,
+      "loss": 0.2442,
+      "step": 6349
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013194570189495444,
+      "loss": 0.2416,
+      "step": 6350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001319256609544765,
+      "loss": 0.2491,
+      "step": 6351
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001319056185861099,
+      "loss": 0.2634,
+      "step": 6352
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013188557479075114,
+      "loss": 0.2632,
+      "step": 6353
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013186552956929664,
+      "loss": 0.2606,
+      "step": 6354
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00013184548292264294,
+      "loss": 0.2434,
+      "step": 6355
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001318254348516866,
+      "loss": 0.2398,
+      "step": 6356
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013180538535732431,
+      "loss": 0.2556,
+      "step": 6357
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001317853344404528,
+      "loss": 0.2484,
+      "step": 6358
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013176528210196887,
+      "loss": 0.2696,
+      "step": 6359
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001317452283427693,
+      "loss": 0.2575,
+      "step": 6360
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013172517316375107,
+      "loss": 0.2374,
+      "step": 6361
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001317051165658111,
+      "loss": 0.2615,
+      "step": 6362
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001316850585498465,
+      "loss": 0.2685,
+      "step": 6363
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013166499911675434,
+      "loss": 0.2442,
+      "step": 6364
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013164493826743176,
+      "loss": 0.257,
+      "step": 6365
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000131624876002776,
+      "loss": 0.2578,
+      "step": 6366
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001316048123236844,
+      "loss": 0.2515,
+      "step": 6367
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013158474723105426,
+      "loss": 0.2623,
+      "step": 6368
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013156468072578304,
+      "loss": 0.2589,
+      "step": 6369
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001315446128087682,
+      "loss": 0.2583,
+      "step": 6370
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013152454348090725,
+      "loss": 0.2398,
+      "step": 6371
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001315044727430979,
+      "loss": 0.2622,
+      "step": 6372
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013148440059623773,
+      "loss": 0.2401,
+      "step": 6373
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013146432704122453,
+      "loss": 0.2495,
+      "step": 6374
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013144425207895605,
+      "loss": 0.2468,
+      "step": 6375
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013142417571033018,
+      "loss": 0.2517,
+      "step": 6376
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013140409793624484,
+      "loss": 0.265,
+      "step": 6377
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013138401875759803,
+      "loss": 0.2719,
+      "step": 6378
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013136393817528776,
+      "loss": 0.2472,
+      "step": 6379
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001313438561902122,
+      "loss": 0.2584,
+      "step": 6380
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013132377280326946,
+      "loss": 0.2539,
+      "step": 6381
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013130368801535783,
+      "loss": 0.275,
+      "step": 6382
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013128360182737564,
+      "loss": 0.2468,
+      "step": 6383
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013126351424022116,
+      "loss": 0.2431,
+      "step": 6384
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013124342525479288,
+      "loss": 0.2635,
+      "step": 6385
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00013122333487198927,
+      "loss": 0.2448,
+      "step": 6386
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013120324309270884,
+      "loss": 0.2614,
+      "step": 6387
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001311831499178503,
+      "loss": 0.2715,
+      "step": 6388
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001311630553483122,
+      "loss": 0.2538,
+      "step": 6389
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001311429593849934,
+      "loss": 0.2422,
+      "step": 6390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013112286202879268,
+      "loss": 0.2625,
+      "step": 6391
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013110276328060882,
+      "loss": 0.2566,
+      "step": 6392
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001310826631413408,
+      "loss": 0.2564,
+      "step": 6393
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013106256161188762,
+      "loss": 0.27,
+      "step": 6394
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001310424586931483,
+      "loss": 0.2446,
+      "step": 6395
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013102235438602197,
+      "loss": 0.2613,
+      "step": 6396
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013100224869140782,
+      "loss": 0.2609,
+      "step": 6397
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013098214161020506,
+      "loss": 0.2606,
+      "step": 6398
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013096203314331297,
+      "loss": 0.2642,
+      "step": 6399
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013094192329163093,
+      "loss": 0.2658,
+      "step": 6400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013092181205605834,
+      "loss": 0.2587,
+      "step": 6401
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 0.2582,
+      "step": 6402
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013088158543683967,
+      "loss": 0.2589,
+      "step": 6403
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013086147005499267,
+      "loss": 0.2547,
+      "step": 6404
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013084135329285344,
+      "loss": 0.2494,
+      "step": 6405
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013082123515132174,
+      "loss": 0.2562,
+      "step": 6406
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013080111563129734,
+      "loss": 0.2691,
+      "step": 6407
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013078099473368007,
+      "loss": 0.2386,
+      "step": 6408
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001307608724593699,
+      "loss": 0.2596,
+      "step": 6409
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013074074880926675,
+      "loss": 0.2552,
+      "step": 6410
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013072062378427068,
+      "loss": 0.2628,
+      "step": 6411
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013070049738528184,
+      "loss": 0.251,
+      "step": 6412
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001306803696132003,
+      "loss": 0.2673,
+      "step": 6413
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013066024046892637,
+      "loss": 0.2517,
+      "step": 6414
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00013064010995336028,
+      "loss": 0.2321,
+      "step": 6415
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001306199780674024,
+      "loss": 0.2563,
+      "step": 6416
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001305998448119531,
+      "loss": 0.2579,
+      "step": 6417
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001305797101879129,
+      "loss": 0.2338,
+      "step": 6418
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001305595741961823,
+      "loss": 0.2536,
+      "step": 6419
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001305394368376619,
+      "loss": 0.2621,
+      "step": 6420
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013051929811325235,
+      "loss": 0.2601,
+      "step": 6421
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013049915802385435,
+      "loss": 0.241,
+      "step": 6422
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001304790165703687,
+      "loss": 0.2459,
+      "step": 6423
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001304588737536962,
+      "loss": 0.2473,
+      "step": 6424
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013043872957473778,
+      "loss": 0.2451,
+      "step": 6425
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001304185840343944,
+      "loss": 0.2472,
+      "step": 6426
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013039843713356702,
+      "loss": 0.2515,
+      "step": 6427
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001303782888731568,
+      "loss": 0.2662,
+      "step": 6428
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001303581392540648,
+      "loss": 0.2551,
+      "step": 6429
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013033798827719227,
+      "loss": 0.2486,
+      "step": 6430
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001303178359434405,
+      "loss": 0.2342,
+      "step": 6431
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013029768225371072,
+      "loss": 0.2295,
+      "step": 6432
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013027752720890442,
+      "loss": 0.2487,
+      "step": 6433
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013025737080992294,
+      "loss": 0.2423,
+      "step": 6434
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013023721305766784,
+      "loss": 0.2434,
+      "step": 6435
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001302170539530407,
+      "loss": 0.253,
+      "step": 6436
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013019689349694308,
+      "loss": 0.2684,
+      "step": 6437
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013017673169027672,
+      "loss": 0.2507,
+      "step": 6438
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013015656853394334,
+      "loss": 0.2547,
+      "step": 6439
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013013640402884477,
+      "loss": 0.2332,
+      "step": 6440
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013011623817588282,
+      "loss": 0.2452,
+      "step": 6441
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001300960709759595,
+      "loss": 0.246,
+      "step": 6442
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013007590242997678,
+      "loss": 0.2474,
+      "step": 6443
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013005573253883662,
+      "loss": 0.2568,
+      "step": 6444
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00013003556130344122,
+      "loss": 0.2499,
+      "step": 6445
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001300153887246927,
+      "loss": 0.2622,
+      "step": 6446
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00012999521480349332,
+      "loss": 0.2561,
+      "step": 6447
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001299750395407453,
+      "loss": 0.2508,
+      "step": 6448
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012995486293735108,
+      "loss": 0.2641,
+      "step": 6449
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000129934684994213,
+      "loss": 0.2467,
+      "step": 6450
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012991450571223355,
+      "loss": 0.2366,
+      "step": 6451
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012989432509231526,
+      "loss": 0.2422,
+      "step": 6452
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012987414313536075,
+      "loss": 0.2509,
+      "step": 6453
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001298539598422726,
+      "loss": 0.2607,
+      "step": 6454
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001298337752139535,
+      "loss": 0.2411,
+      "step": 6455
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012981358925130633,
+      "loss": 0.2535,
+      "step": 6456
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001297934019552338,
+      "loss": 0.255,
+      "step": 6457
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012977321332663884,
+      "loss": 0.2491,
+      "step": 6458
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012975302336642444,
+      "loss": 0.2529,
+      "step": 6459
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012973283207549355,
+      "loss": 0.2596,
+      "step": 6460
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001297126394547492,
+      "loss": 0.2563,
+      "step": 6461
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001296924455050946,
+      "loss": 0.2511,
+      "step": 6462
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012967225022743293,
+      "loss": 0.2434,
+      "step": 6463
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012965205362266733,
+      "loss": 0.2645,
+      "step": 6464
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012963185569170118,
+      "loss": 0.2623,
+      "step": 6465
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012961165643543784,
+      "loss": 0.2586,
+      "step": 6466
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001295914558547807,
+      "loss": 0.2568,
+      "step": 6467
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012957125395063326,
+      "loss": 0.2539,
+      "step": 6468
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001295510507238991,
+      "loss": 0.2537,
+      "step": 6469
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012953084617548172,
+      "loss": 0.2509,
+      "step": 6470
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012951064030628483,
+      "loss": 0.2421,
+      "step": 6471
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012949043311721216,
+      "loss": 0.2613,
+      "step": 6472
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012947022460916747,
+      "loss": 0.2411,
+      "step": 6473
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012945001478305458,
+      "loss": 0.2552,
+      "step": 6474
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012942980363977747,
+      "loss": 0.2524,
+      "step": 6475
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012940959118023994,
+      "loss": 0.2521,
+      "step": 6476
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00012938937740534611,
+      "loss": 0.2522,
+      "step": 6477
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000129369162316,
+      "loss": 0.2598,
+      "step": 6478
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001293489459131058,
+      "loss": 0.2483,
+      "step": 6479
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001293287281975676,
+      "loss": 0.2386,
+      "step": 6480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012930850917028974,
+      "loss": 0.2524,
+      "step": 6481
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012928828883217646,
+      "loss": 0.2793,
+      "step": 6482
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012926806718413218,
+      "loss": 0.2504,
+      "step": 6483
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012924784422706126,
+      "loss": 0.2612,
+      "step": 6484
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012922761996186823,
+      "loss": 0.2613,
+      "step": 6485
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001292073943894576,
+      "loss": 0.2533,
+      "step": 6486
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000129187167510734,
+      "loss": 0.2757,
+      "step": 6487
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012916693932660205,
+      "loss": 0.2616,
+      "step": 6488
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012914670983796647,
+      "loss": 0.2475,
+      "step": 6489
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012912647904573203,
+      "loss": 0.2544,
+      "step": 6490
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001291062469508036,
+      "loss": 0.2684,
+      "step": 6491
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012908601355408599,
+      "loss": 0.2593,
+      "step": 6492
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012906577885648426,
+      "loss": 0.2472,
+      "step": 6493
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001290455428589033,
+      "loss": 0.2531,
+      "step": 6494
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012902530556224825,
+      "loss": 0.2445,
+      "step": 6495
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001290050669674242,
+      "loss": 0.2605,
+      "step": 6496
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012898482707533632,
+      "loss": 0.2573,
+      "step": 6497
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001289645858868899,
+      "loss": 0.255,
+      "step": 6498
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012894434340299016,
+      "loss": 0.2491,
+      "step": 6499
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012892409962454252,
+      "loss": 0.248,
+      "step": 6500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012890385455245233,
+      "loss": 0.2514,
+      "step": 6501
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012888360818762512,
+      "loss": 0.2546,
+      "step": 6502
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001288633605309664,
+      "loss": 0.2564,
+      "step": 6503
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012884311158338173,
+      "loss": 0.255,
+      "step": 6504
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001288228613457768,
+      "loss": 0.2481,
+      "step": 6505
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012880260981905722,
+      "loss": 0.2635,
+      "step": 6506
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012878235700412884,
+      "loss": 0.2415,
+      "step": 6507
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00012876210290189746,
+      "loss": 0.2598,
+      "step": 6508
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001287418475132689,
+      "loss": 0.2499,
+      "step": 6509
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012872159083914916,
+      "loss": 0.2517,
+      "step": 6510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012870133288044416,
+      "loss": 0.2446,
+      "step": 6511
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012868107363806002,
+      "loss": 0.2731,
+      "step": 6512
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012866081311290278,
+      "loss": 0.2487,
+      "step": 6513
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012864055130587865,
+      "loss": 0.2551,
+      "step": 6514
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012862028821789378,
+      "loss": 0.2523,
+      "step": 6515
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012860002384985453,
+      "loss": 0.246,
+      "step": 6516
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012857975820266716,
+      "loss": 0.2462,
+      "step": 6517
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001285594912772381,
+      "loss": 0.2433,
+      "step": 6518
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012853922307447377,
+      "loss": 0.2608,
+      "step": 6519
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001285189535952807,
+      "loss": 0.2463,
+      "step": 6520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012849868284056545,
+      "loss": 0.2575,
+      "step": 6521
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012847841081123462,
+      "loss": 0.2446,
+      "step": 6522
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001284581375081949,
+      "loss": 0.2501,
+      "step": 6523
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012843786293235303,
+      "loss": 0.2431,
+      "step": 6524
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012841758708461579,
+      "loss": 0.2335,
+      "step": 6525
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012839730996588996,
+      "loss": 0.2563,
+      "step": 6526
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012837703157708255,
+      "loss": 0.2412,
+      "step": 6527
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012835675191910046,
+      "loss": 0.2382,
+      "step": 6528
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012833647099285073,
+      "loss": 0.2462,
+      "step": 6529
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001283161887992404,
+      "loss": 0.2541,
+      "step": 6530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012829590533917665,
+      "loss": 0.2528,
+      "step": 6531
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012827562061356663,
+      "loss": 0.242,
+      "step": 6532
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001282553346233176,
+      "loss": 0.253,
+      "step": 6533
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012823504736933684,
+      "loss": 0.2369,
+      "step": 6534
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001282147588525317,
+      "loss": 0.2509,
+      "step": 6535
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012819446907380962,
+      "loss": 0.2528,
+      "step": 6536
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012817417803407806,
+      "loss": 0.2574,
+      "step": 6537
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012815388573424457,
+      "loss": 0.2438,
+      "step": 6538
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00012813359217521666,
+      "loss": 0.2389,
+      "step": 6539
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012811329735790204,
+      "loss": 0.2429,
+      "step": 6540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012809300128320837,
+      "loss": 0.237,
+      "step": 6541
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001280727039520434,
+      "loss": 0.2555,
+      "step": 6542
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012805240536531496,
+      "loss": 0.244,
+      "step": 6543
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012803210552393089,
+      "loss": 0.2359,
+      "step": 6544
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001280118044287991,
+      "loss": 0.229,
+      "step": 6545
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012799150208082759,
+      "loss": 0.2429,
+      "step": 6546
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012797119848092438,
+      "loss": 0.2514,
+      "step": 6547
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012795089362999758,
+      "loss": 0.2511,
+      "step": 6548
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012793058752895529,
+      "loss": 0.2564,
+      "step": 6549
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012791028017870572,
+      "loss": 0.2489,
+      "step": 6550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012788997158015713,
+      "loss": 0.2501,
+      "step": 6551
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001278696617342179,
+      "loss": 0.2493,
+      "step": 6552
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012784935064179628,
+      "loss": 0.2427,
+      "step": 6553
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012782903830380076,
+      "loss": 0.2518,
+      "step": 6554
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001278087247211398,
+      "loss": 0.245,
+      "step": 6555
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012778840989472191,
+      "loss": 0.2402,
+      "step": 6556
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012776809382545573,
+      "loss": 0.2546,
+      "step": 6557
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012774777651424987,
+      "loss": 0.2622,
+      "step": 6558
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012772745796201304,
+      "loss": 0.2413,
+      "step": 6559
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000127707138169654,
+      "loss": 0.265,
+      "step": 6560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012768681713808155,
+      "loss": 0.2532,
+      "step": 6561
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012766649486820457,
+      "loss": 0.259,
+      "step": 6562
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012764617136093199,
+      "loss": 0.2465,
+      "step": 6563
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001276258466171728,
+      "loss": 0.3169,
+      "step": 6564
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012760552063783594,
+      "loss": 0.2478,
+      "step": 6565
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001275851934238306,
+      "loss": 0.2517,
+      "step": 6566
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012756486497606586,
+      "loss": 0.2675,
+      "step": 6567
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012754453529545097,
+      "loss": 0.2494,
+      "step": 6568
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012752420438289514,
+      "loss": 0.2712,
+      "step": 6569
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00012750387223930772,
+      "loss": 0.2645,
+      "step": 6570
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012748353886559802,
+      "loss": 0.2501,
+      "step": 6571
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012746320426267554,
+      "loss": 0.2665,
+      "step": 6572
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012744286843144968,
+      "loss": 0.2501,
+      "step": 6573
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012742253137283,
+      "loss": 0.2388,
+      "step": 6574
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012740219308772605,
+      "loss": 0.2517,
+      "step": 6575
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001273818535770475,
+      "loss": 0.2572,
+      "step": 6576
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012736151284170404,
+      "loss": 0.2594,
+      "step": 6577
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012734117088260543,
+      "loss": 0.266,
+      "step": 6578
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012732082770066145,
+      "loss": 0.2486,
+      "step": 6579
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012730048329678194,
+      "loss": 0.2686,
+      "step": 6580
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012728013767187687,
+      "loss": 0.2485,
+      "step": 6581
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012725979082685619,
+      "loss": 0.2539,
+      "step": 6582
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012723944276262986,
+      "loss": 0.253,
+      "step": 6583
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012721909348010804,
+      "loss": 0.2552,
+      "step": 6584
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001271987429802008,
+      "loss": 0.2494,
+      "step": 6585
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012717839126381834,
+      "loss": 0.2384,
+      "step": 6586
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001271580383318709,
+      "loss": 0.2516,
+      "step": 6587
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012713768418526878,
+      "loss": 0.2562,
+      "step": 6588
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012711732882492237,
+      "loss": 0.2516,
+      "step": 6589
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012709697225174197,
+      "loss": 0.2518,
+      "step": 6590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001270766144666381,
+      "loss": 0.2401,
+      "step": 6591
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012705625547052132,
+      "loss": 0.2489,
+      "step": 6592
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001270358952643021,
+      "loss": 0.2648,
+      "step": 6593
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001270155338488911,
+      "loss": 0.2526,
+      "step": 6594
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000126995171225199,
+      "loss": 0.2486,
+      "step": 6595
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001269748073941365,
+      "loss": 0.2467,
+      "step": 6596
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012695444235661442,
+      "loss": 0.2656,
+      "step": 6597
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012693407611354354,
+      "loss": 0.2514,
+      "step": 6598
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012691370866583482,
+      "loss": 0.2626,
+      "step": 6599
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00012689334001439914,
+      "loss": 0.2535,
+      "step": 6600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012687297016014754,
+      "loss": 0.2438,
+      "step": 6601
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012685259910399105,
+      "loss": 0.2467,
+      "step": 6602
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001268322268468408,
+      "loss": 0.2554,
+      "step": 6603
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012681185338960787,
+      "loss": 0.243,
+      "step": 6604
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012679147873320356,
+      "loss": 0.2438,
+      "step": 6605
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012677110287853907,
+      "loss": 0.2384,
+      "step": 6606
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012675072582652577,
+      "loss": 0.2524,
+      "step": 6607
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012673034757807502,
+      "loss": 0.2599,
+      "step": 6608
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012670996813409822,
+      "loss": 0.2529,
+      "step": 6609
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012668958749550687,
+      "loss": 0.2458,
+      "step": 6610
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012666920566321252,
+      "loss": 0.2436,
+      "step": 6611
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012664882263812674,
+      "loss": 0.2328,
+      "step": 6612
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012662843842116113,
+      "loss": 0.246,
+      "step": 6613
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001266080530132274,
+      "loss": 0.2562,
+      "step": 6614
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012658766641523736,
+      "loss": 0.2431,
+      "step": 6615
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012656727862810272,
+      "loss": 0.255,
+      "step": 6616
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001265468896527354,
+      "loss": 0.2401,
+      "step": 6617
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001265264994900473,
+      "loss": 0.2412,
+      "step": 6618
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012650610814095038,
+      "loss": 0.2427,
+      "step": 6619
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012648571560635655,
+      "loss": 0.2452,
+      "step": 6620
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.000126465321887178,
+      "loss": 0.2428,
+      "step": 6621
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001264449269843268,
+      "loss": 0.2393,
+      "step": 6622
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001264245308987151,
+      "loss": 0.2551,
+      "step": 6623
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001264041336312552,
+      "loss": 0.2557,
+      "step": 6624
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001263837351828593,
+      "loss": 0.2394,
+      "step": 6625
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001263633355544397,
+      "loss": 0.2594,
+      "step": 6626
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012634293474690888,
+      "loss": 0.2505,
+      "step": 6627
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012632253276117922,
+      "loss": 0.2532,
+      "step": 6628
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012630212959816324,
+      "loss": 0.2446,
+      "step": 6629
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012628172525877343,
+      "loss": 0.2401,
+      "step": 6630
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00012626131974392238,
+      "loss": 0.2409,
+      "step": 6631
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012624091305452281,
+      "loss": 0.2302,
+      "step": 6632
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012622050519148738,
+      "loss": 0.2502,
+      "step": 6633
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001262000961557288,
+      "loss": 0.2559,
+      "step": 6634
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012617968594815993,
+      "loss": 0.2493,
+      "step": 6635
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012615927456969357,
+      "loss": 0.2471,
+      "step": 6636
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012613886202124267,
+      "loss": 0.24,
+      "step": 6637
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012611844830372017,
+      "loss": 0.2407,
+      "step": 6638
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012609803341803909,
+      "loss": 0.2332,
+      "step": 6639
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012607761736511253,
+      "loss": 0.2466,
+      "step": 6640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012605720014585353,
+      "loss": 0.2407,
+      "step": 6641
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001260367817611753,
+      "loss": 0.2456,
+      "step": 6642
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012601636221199112,
+      "loss": 0.2495,
+      "step": 6643
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012599594149921415,
+      "loss": 0.2411,
+      "step": 6644
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012597551962375774,
+      "loss": 0.2509,
+      "step": 6645
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012595509658653534,
+      "loss": 0.2452,
+      "step": 6646
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001259346723884603,
+      "loss": 0.2381,
+      "step": 6647
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012591424703044618,
+      "loss": 0.255,
+      "step": 6648
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001258938205134064,
+      "loss": 0.2617,
+      "step": 6649
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001258733928382547,
+      "loss": 0.2537,
+      "step": 6650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012585296400590456,
+      "loss": 0.2464,
+      "step": 6651
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012583253401726976,
+      "loss": 0.237,
+      "step": 6652
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012581210287326405,
+      "loss": 0.2437,
+      "step": 6653
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012579167057480113,
+      "loss": 0.2276,
+      "step": 6654
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001257712371227949,
+      "loss": 0.2378,
+      "step": 6655
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001257508025181593,
+      "loss": 0.2427,
+      "step": 6656
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001257303667618082,
+      "loss": 0.2568,
+      "step": 6657
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012570992985465562,
+      "loss": 0.2576,
+      "step": 6658
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012568949179761563,
+      "loss": 0.25,
+      "step": 6659
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012566905259160234,
+      "loss": 0.2486,
+      "step": 6660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00012564861223752987,
+      "loss": 0.2443,
+      "step": 6661
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001256281707363124,
+      "loss": 0.2568,
+      "step": 6662
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012560772808886428,
+      "loss": 0.247,
+      "step": 6663
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012558728429609974,
+      "loss": 0.2601,
+      "step": 6664
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012556683935893313,
+      "loss": 0.2477,
+      "step": 6665
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001255463932782789,
+      "loss": 0.2622,
+      "step": 6666
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001255259460550515,
+      "loss": 0.2566,
+      "step": 6667
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012550549769016538,
+      "loss": 0.2435,
+      "step": 6668
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001254850481845352,
+      "loss": 0.249,
+      "step": 6669
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001254645975390755,
+      "loss": 0.2414,
+      "step": 6670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012544414575470098,
+      "loss": 0.2431,
+      "step": 6671
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012542369283232634,
+      "loss": 0.2549,
+      "step": 6672
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001254032387728664,
+      "loss": 0.2443,
+      "step": 6673
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012538278357723587,
+      "loss": 0.2421,
+      "step": 6674
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012536232724634968,
+      "loss": 0.2465,
+      "step": 6675
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012534186978112275,
+      "loss": 0.246,
+      "step": 6676
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012532141118247003,
+      "loss": 0.2407,
+      "step": 6677
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012530095145130656,
+      "loss": 0.2375,
+      "step": 6678
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012528049058854737,
+      "loss": 0.2463,
+      "step": 6679
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012526002859510765,
+      "loss": 0.2604,
+      "step": 6680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001252395654719025,
+      "loss": 0.2594,
+      "step": 6681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001252191012198472,
+      "loss": 0.259,
+      "step": 6682
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012519863583985703,
+      "loss": 0.2397,
+      "step": 6683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012517816933284724,
+      "loss": 0.2519,
+      "step": 6684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001251577016997332,
+      "loss": 0.2447,
+      "step": 6685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012513723294143043,
+      "loss": 0.2491,
+      "step": 6686
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012511676305885434,
+      "loss": 0.2523,
+      "step": 6687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012509629205292045,
+      "loss": 0.2279,
+      "step": 6688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012507581992454435,
+      "loss": 0.2569,
+      "step": 6689
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012505534667464166,
+      "loss": 0.2529,
+      "step": 6690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00012503487230412807,
+      "loss": 0.2569,
+      "step": 6691
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001250143968139193,
+      "loss": 0.2441,
+      "step": 6692
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012499392020493108,
+      "loss": 0.2645,
+      "step": 6693
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001249734424780793,
+      "loss": 0.2556,
+      "step": 6694
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001249529636342798,
+      "loss": 0.2402,
+      "step": 6695
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001249324836744485,
+      "loss": 0.2432,
+      "step": 6696
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001249120025995014,
+      "loss": 0.2381,
+      "step": 6697
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001248915204103545,
+      "loss": 0.2432,
+      "step": 6698
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001248710371079239,
+      "loss": 0.236,
+      "step": 6699
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001248505526931257,
+      "loss": 0.2483,
+      "step": 6700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012483006716687607,
+      "loss": 0.2488,
+      "step": 6701
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012480958053009126,
+      "loss": 0.2453,
+      "step": 6702
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012478909278368754,
+      "loss": 0.2449,
+      "step": 6703
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001247686039285812,
+      "loss": 0.2337,
+      "step": 6704
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012474811396568866,
+      "loss": 0.2308,
+      "step": 6705
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012472762289592629,
+      "loss": 0.2448,
+      "step": 6706
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012470713072021064,
+      "loss": 0.2463,
+      "step": 6707
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012468663743945814,
+      "loss": 0.2397,
+      "step": 6708
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001246661430545854,
+      "loss": 0.2424,
+      "step": 6709
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012464564756650907,
+      "loss": 0.2544,
+      "step": 6710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012462515097614576,
+      "loss": 0.2426,
+      "step": 6711
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012460465328441225,
+      "loss": 0.2403,
+      "step": 6712
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012458415449222526,
+      "loss": 0.2379,
+      "step": 6713
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012456365460050163,
+      "loss": 0.2461,
+      "step": 6714
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012454315361015822,
+      "loss": 0.2457,
+      "step": 6715
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012452265152211196,
+      "loss": 0.248,
+      "step": 6716
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001245021483372798,
+      "loss": 0.2394,
+      "step": 6717
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012448164405657873,
+      "loss": 0.2435,
+      "step": 6718
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012446113868092585,
+      "loss": 0.2462,
+      "step": 6719
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012444063221123824,
+      "loss": 0.2439,
+      "step": 6720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001244201246484331,
+      "loss": 0.2565,
+      "step": 6721
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012439961599342763,
+      "loss": 0.2372,
+      "step": 6722
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00012437910624713905,
+      "loss": 0.2516,
+      "step": 6723
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012435859541048473,
+      "loss": 0.2546,
+      "step": 6724
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012433808348438197,
+      "loss": 0.2522,
+      "step": 6725
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001243175704697482,
+      "loss": 0.2462,
+      "step": 6726
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012429705636750083,
+      "loss": 0.2431,
+      "step": 6727
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012427654117855744,
+      "loss": 0.2423,
+      "step": 6728
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012425602490383556,
+      "loss": 0.2389,
+      "step": 6729
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001242355075442527,
+      "loss": 0.2357,
+      "step": 6730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012421498910072663,
+      "loss": 0.2608,
+      "step": 6731
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012419446957417498,
+      "loss": 0.2562,
+      "step": 6732
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001241739489655155,
+      "loss": 0.2514,
+      "step": 6733
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012415342727566602,
+      "loss": 0.2372,
+      "step": 6734
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001241329045055443,
+      "loss": 0.2387,
+      "step": 6735
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001241123806560683,
+      "loss": 0.2474,
+      "step": 6736
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012409185572815593,
+      "loss": 0.2379,
+      "step": 6737
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001240713297227252,
+      "loss": 0.2406,
+      "step": 6738
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012405080264069412,
+      "loss": 0.2472,
+      "step": 6739
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001240302744829808,
+      "loss": 0.2174,
+      "step": 6740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012400974525050333,
+      "loss": 0.2447,
+      "step": 6741
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001239892149441799,
+      "loss": 0.2453,
+      "step": 6742
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001239686835649288,
+      "loss": 0.2281,
+      "step": 6743
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001239481511136682,
+      "loss": 0.2438,
+      "step": 6744
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012392761759131651,
+      "loss": 0.2541,
+      "step": 6745
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012390708299879203,
+      "loss": 0.2524,
+      "step": 6746
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012388654733701325,
+      "loss": 0.2471,
+      "step": 6747
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012386601060689856,
+      "loss": 0.2371,
+      "step": 6748
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001238454728093665,
+      "loss": 0.2265,
+      "step": 6749
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012382493394533567,
+      "loss": 0.246,
+      "step": 6750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012380439401572465,
+      "loss": 0.2417,
+      "step": 6751
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012378385302145212,
+      "loss": 0.2442,
+      "step": 6752
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012376331096343676,
+      "loss": 0.235,
+      "step": 6753
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001237427678425973,
+      "loss": 0.2294,
+      "step": 6754
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001237222236598526,
+      "loss": 0.257,
+      "step": 6755
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012370167841612144,
+      "loss": 0.2493,
+      "step": 6756
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012368113211232276,
+      "loss": 0.2676,
+      "step": 6757
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012366058474937549,
+      "loss": 0.2488,
+      "step": 6758
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012364003632819862,
+      "loss": 0.2672,
+      "step": 6759
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001236194868497112,
+      "loss": 0.2351,
+      "step": 6760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012359893631483227,
+      "loss": 0.2592,
+      "step": 6761
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012357838472448103,
+      "loss": 0.2564,
+      "step": 6762
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012355783207957662,
+      "loss": 0.2399,
+      "step": 6763
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012353727838103825,
+      "loss": 0.2413,
+      "step": 6764
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012351672362978522,
+      "loss": 0.251,
+      "step": 6765
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012349616782673684,
+      "loss": 0.2452,
+      "step": 6766
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012347561097281244,
+      "loss": 0.2506,
+      "step": 6767
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012345505306893154,
+      "loss": 0.2583,
+      "step": 6768
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012343449411601347,
+      "loss": 0.2437,
+      "step": 6769
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001234139341149778,
+      "loss": 0.2448,
+      "step": 6770
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001233933730667441,
+      "loss": 0.2553,
+      "step": 6771
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012337281097223195,
+      "loss": 0.2455,
+      "step": 6772
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012335224783236102,
+      "loss": 0.234,
+      "step": 6773
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.000123331683648051,
+      "loss": 0.2555,
+      "step": 6774
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012331111842022157,
+      "loss": 0.2383,
+      "step": 6775
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001232905521497926,
+      "loss": 0.2407,
+      "step": 6776
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012326998483768389,
+      "loss": 0.2348,
+      "step": 6777
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012324941648481533,
+      "loss": 0.2507,
+      "step": 6778
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012322884709210682,
+      "loss": 0.2373,
+      "step": 6779
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012320827666047836,
+      "loss": 0.2553,
+      "step": 6780
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012318770519085,
+      "loss": 0.2542,
+      "step": 6781
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012316713268414176,
+      "loss": 0.2491,
+      "step": 6782
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00012314655914127378,
+      "loss": 0.2396,
+      "step": 6783
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001231259845631662,
+      "loss": 0.2449,
+      "step": 6784
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012310540895073925,
+      "loss": 0.2439,
+      "step": 6785
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012308483230491314,
+      "loss": 0.2376,
+      "step": 6786
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012306425462660825,
+      "loss": 0.2436,
+      "step": 6787
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012304367591674484,
+      "loss": 0.261,
+      "step": 6788
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012302309617624334,
+      "loss": 0.248,
+      "step": 6789
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001230025154060242,
+      "loss": 0.2577,
+      "step": 6790
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012298193360700784,
+      "loss": 0.2483,
+      "step": 6791
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012296135078011488,
+      "loss": 0.2526,
+      "step": 6792
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001229407669262659,
+      "loss": 0.2421,
+      "step": 6793
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001229201820463814,
+      "loss": 0.2518,
+      "step": 6794
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012289959614138213,
+      "loss": 0.2414,
+      "step": 6795
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012287900921218882,
+      "loss": 0.2309,
+      "step": 6796
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001228584212597222,
+      "loss": 0.2516,
+      "step": 6797
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012283783228490307,
+      "loss": 0.2525,
+      "step": 6798
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001228172422886523,
+      "loss": 0.2634,
+      "step": 6799
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012279665127189077,
+      "loss": 0.2505,
+      "step": 6800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012277605923553944,
+      "loss": 0.2161,
+      "step": 6801
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012275546618051926,
+      "loss": 0.2375,
+      "step": 6802
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012273487210775133,
+      "loss": 0.2335,
+      "step": 6803
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012271427701815665,
+      "loss": 0.2423,
+      "step": 6804
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001226936809126564,
+      "loss": 0.2422,
+      "step": 6805
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001226730837921717,
+      "loss": 0.2366,
+      "step": 6806
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012265248565762382,
+      "loss": 0.2437,
+      "step": 6807
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000122631886509934,
+      "loss": 0.253,
+      "step": 6808
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012261128635002354,
+      "loss": 0.2399,
+      "step": 6809
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012259068517881378,
+      "loss": 0.234,
+      "step": 6810
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012257008299722613,
+      "loss": 0.2479,
+      "step": 6811
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012254947980618204,
+      "loss": 0.2524,
+      "step": 6812
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000122528875606603,
+      "loss": 0.2498,
+      "step": 6813
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012250827039941048,
+      "loss": 0.2412,
+      "step": 6814
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00012248766418552614,
+      "loss": 0.2466,
+      "step": 6815
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012246705696587155,
+      "loss": 0.2398,
+      "step": 6816
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001224464487413684,
+      "loss": 0.2488,
+      "step": 6817
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001224258395129384,
+      "loss": 0.2551,
+      "step": 6818
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012240522928150328,
+      "loss": 0.229,
+      "step": 6819
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001223846180479849,
+      "loss": 0.2447,
+      "step": 6820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000122364005813305,
+      "loss": 0.2411,
+      "step": 6821
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001223433925783856,
+      "loss": 0.2414,
+      "step": 6822
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012232277834414856,
+      "loss": 0.238,
+      "step": 6823
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012230216311151588,
+      "loss": 0.2516,
+      "step": 6824
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012228154688140956,
+      "loss": 0.2528,
+      "step": 6825
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012226092965475168,
+      "loss": 0.2644,
+      "step": 6826
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012224031143246437,
+      "loss": 0.2341,
+      "step": 6827
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012221969221546978,
+      "loss": 0.2455,
+      "step": 6828
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012219907200469013,
+      "loss": 0.2463,
+      "step": 6829
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012217845080104764,
+      "loss": 0.2182,
+      "step": 6830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012215782860546464,
+      "loss": 0.235,
+      "step": 6831
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001221372054188634,
+      "loss": 0.2419,
+      "step": 6832
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012211658124216637,
+      "loss": 0.2417,
+      "step": 6833
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012209595607629595,
+      "loss": 0.2437,
+      "step": 6834
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001220753299221746,
+      "loss": 0.2402,
+      "step": 6835
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012205470278072479,
+      "loss": 0.2436,
+      "step": 6836
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012203407465286918,
+      "loss": 0.2611,
+      "step": 6837
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012201344553953029,
+      "loss": 0.2562,
+      "step": 6838
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012199281544163081,
+      "loss": 0.2478,
+      "step": 6839
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012197218436009344,
+      "loss": 0.247,
+      "step": 6840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012195155229584085,
+      "loss": 0.251,
+      "step": 6841
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012193091924979585,
+      "loss": 0.2232,
+      "step": 6842
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012191028522288129,
+      "loss": 0.2268,
+      "step": 6843
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00012188965021602,
+      "loss": 0.2485,
+      "step": 6844
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001218690142301349,
+      "loss": 0.2665,
+      "step": 6845
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012184837726614897,
+      "loss": 0.256,
+      "step": 6846
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012182773932498514,
+      "loss": 0.2482,
+      "step": 6847
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001218071004075665,
+      "loss": 0.2431,
+      "step": 6848
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012178646051481614,
+      "loss": 0.2452,
+      "step": 6849
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012176581964765718,
+      "loss": 0.2388,
+      "step": 6850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012174517780701277,
+      "loss": 0.2445,
+      "step": 6851
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012172453499380616,
+      "loss": 0.2425,
+      "step": 6852
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001217038912089606,
+      "loss": 0.2435,
+      "step": 6853
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012168324645339935,
+      "loss": 0.2517,
+      "step": 6854
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001216626007280458,
+      "loss": 0.2395,
+      "step": 6855
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012164195403382332,
+      "loss": 0.2551,
+      "step": 6856
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012162130637165535,
+      "loss": 0.2492,
+      "step": 6857
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012160065774246537,
+      "loss": 0.2385,
+      "step": 6858
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012158000814717687,
+      "loss": 0.2411,
+      "step": 6859
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012155935758671345,
+      "loss": 0.2311,
+      "step": 6860
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012153870606199869,
+      "loss": 0.2418,
+      "step": 6861
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012151805357395625,
+      "loss": 0.241,
+      "step": 6862
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012149740012350982,
+      "loss": 0.2423,
+      "step": 6863
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012147674571158314,
+      "loss": 0.2303,
+      "step": 6864
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012145609033909997,
+      "loss": 0.2475,
+      "step": 6865
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012143543400698414,
+      "loss": 0.242,
+      "step": 6866
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001214147767161595,
+      "loss": 0.2397,
+      "step": 6867
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012139411846754998,
+      "loss": 0.2501,
+      "step": 6868
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012137345926207953,
+      "loss": 0.2451,
+      "step": 6869
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001213527991006721,
+      "loss": 0.2408,
+      "step": 6870
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012133213798425176,
+      "loss": 0.2307,
+      "step": 6871
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012131147591374258,
+      "loss": 0.2265,
+      "step": 6872
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012129081289006872,
+      "loss": 0.2419,
+      "step": 6873
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012127014891415429,
+      "loss": 0.2575,
+      "step": 6874
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012124948398692349,
+      "loss": 0.2305,
+      "step": 6875
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00012122881810930059,
+      "loss": 0.2375,
+      "step": 6876
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012120815128220986,
+      "loss": 0.2415,
+      "step": 6877
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012118748350657566,
+      "loss": 0.2491,
+      "step": 6878
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012116681478332235,
+      "loss": 0.2464,
+      "step": 6879
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012114614511337434,
+      "loss": 0.2365,
+      "step": 6880
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012112547449765612,
+      "loss": 0.2334,
+      "step": 6881
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012110480293709217,
+      "loss": 0.2361,
+      "step": 6882
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012108413043260705,
+      "loss": 0.2367,
+      "step": 6883
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001210634569851253,
+      "loss": 0.2412,
+      "step": 6884
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012104278259557158,
+      "loss": 0.2397,
+      "step": 6885
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012102210726487058,
+      "loss": 0.248,
+      "step": 6886
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012100143099394698,
+      "loss": 0.2392,
+      "step": 6887
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012098075378372558,
+      "loss": 0.2458,
+      "step": 6888
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012096007563513111,
+      "loss": 0.2416,
+      "step": 6889
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012093939654908846,
+      "loss": 0.2439,
+      "step": 6890
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012091871652652252,
+      "loss": 0.2347,
+      "step": 6891
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012089803556835815,
+      "loss": 0.2352,
+      "step": 6892
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001208773536755204,
+      "loss": 0.2392,
+      "step": 6893
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012085667084893419,
+      "loss": 0.2422,
+      "step": 6894
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012083598708952461,
+      "loss": 0.2388,
+      "step": 6895
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012081530239821677,
+      "loss": 0.2378,
+      "step": 6896
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012079461677593575,
+      "loss": 0.2353,
+      "step": 6897
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012077393022360677,
+      "loss": 0.251,
+      "step": 6898
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012075324274215503,
+      "loss": 0.2492,
+      "step": 6899
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012073255433250578,
+      "loss": 0.2409,
+      "step": 6900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001207118649955843,
+      "loss": 0.2453,
+      "step": 6901
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012069117473231599,
+      "loss": 0.227,
+      "step": 6902
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001206704835436262,
+      "loss": 0.224,
+      "step": 6903
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012064979143044031,
+      "loss": 0.2522,
+      "step": 6904
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012062909839368381,
+      "loss": 0.2487,
+      "step": 6905
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00012060840443428223,
+      "loss": 0.233,
+      "step": 6906
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001205877095531611,
+      "loss": 0.2425,
+      "step": 6907
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012056701375124599,
+      "loss": 0.2342,
+      "step": 6908
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012054631702946256,
+      "loss": 0.2392,
+      "step": 6909
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012052561938873646,
+      "loss": 3.3851,
+      "step": 6910
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012050492082999339,
+      "loss": 8.5248,
+      "step": 6911
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012048422135415913,
+      "loss": 16.4152,
+      "step": 6912
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012046352096215945,
+      "loss": 12.4589,
+      "step": 6913
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012044281965492019,
+      "loss": 7.7997,
+      "step": 6914
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012042211743336724,
+      "loss": 6.4504,
+      "step": 6915
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012040141429842649,
+      "loss": 7.4731,
+      "step": 6916
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012038071025102391,
+      "loss": 6.2519,
+      "step": 6917
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012036000529208553,
+      "loss": 5.3414,
+      "step": 6918
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012033929942253732,
+      "loss": 4.7702,
+      "step": 6919
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012031859264330538,
+      "loss": 4.083,
+      "step": 6920
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012029788495531586,
+      "loss": 3.7374,
+      "step": 6921
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012027717635949489,
+      "loss": 3.5441,
+      "step": 6922
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001202564668567687,
+      "loss": 3.1207,
+      "step": 6923
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012023575644806353,
+      "loss": 2.8158,
+      "step": 6924
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012021504513430561,
+      "loss": 2.8188,
+      "step": 6925
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001201943329164213,
+      "loss": 2.6706,
+      "step": 6926
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012017361979533696,
+      "loss": 2.4295,
+      "step": 6927
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012015290577197902,
+      "loss": 2.2879,
+      "step": 6928
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012013219084727388,
+      "loss": 2.242,
+      "step": 6929
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012011147502214803,
+      "loss": 2.1843,
+      "step": 6930
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012009075829752802,
+      "loss": 2.1269,
+      "step": 6931
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001200700406743404,
+      "loss": 2.0503,
+      "step": 6932
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001200493221535118,
+      "loss": 1.9934,
+      "step": 6933
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012002860273596878,
+      "loss": 1.9185,
+      "step": 6934
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00012000788242263811,
+      "loss": 1.8519,
+      "step": 6935
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00011998716121444648,
+      "loss": 1.8082,
+      "step": 6936
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00011996643911232067,
+      "loss": 1.7634,
+      "step": 6937
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011994571611718747,
+      "loss": 1.7485,
+      "step": 6938
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011992499222997373,
+      "loss": 1.6827,
+      "step": 6939
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011990426745160635,
+      "loss": 1.5976,
+      "step": 6940
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011988354178301222,
+      "loss": 1.6008,
+      "step": 6941
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011986281522511833,
+      "loss": 1.5507,
+      "step": 6942
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011984208777885171,
+      "loss": 1.5289,
+      "step": 6943
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011982135944513935,
+      "loss": 1.4768,
+      "step": 6944
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011980063022490835,
+      "loss": 1.4529,
+      "step": 6945
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011977990011908584,
+      "loss": 1.4498,
+      "step": 6946
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011975916912859899,
+      "loss": 1.3974,
+      "step": 6947
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011973843725437497,
+      "loss": 1.3763,
+      "step": 6948
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011971770449734107,
+      "loss": 1.3405,
+      "step": 6949
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011969697085842452,
+      "loss": 1.336,
+      "step": 6950
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011967623633855269,
+      "loss": 1.2995,
+      "step": 6951
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011965550093865292,
+      "loss": 1.2807,
+      "step": 6952
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011963476465965263,
+      "loss": 1.2533,
+      "step": 6953
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011961402750247921,
+      "loss": 1.2332,
+      "step": 6954
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011959328946806016,
+      "loss": 1.2458,
+      "step": 6955
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000119572550557323,
+      "loss": 1.2043,
+      "step": 6956
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011955181077119528,
+      "loss": 1.1629,
+      "step": 6957
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011953107011060462,
+      "loss": 1.1552,
+      "step": 6958
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011951032857647863,
+      "loss": 1.1694,
+      "step": 6959
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011948958616974499,
+      "loss": 1.1175,
+      "step": 6960
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011946884289133143,
+      "loss": 1.0862,
+      "step": 6961
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011944809874216567,
+      "loss": 1.0869,
+      "step": 6962
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011942735372317555,
+      "loss": 1.0616,
+      "step": 6963
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011940660783528886,
+      "loss": 1.0708,
+      "step": 6964
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011938586107943343,
+      "loss": 1.0284,
+      "step": 6965
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011936511345653725,
+      "loss": 1.0474,
+      "step": 6966
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011934436496752823,
+      "loss": 1.0464,
+      "step": 6967
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00011932361561333434,
+      "loss": 1.0034,
+      "step": 6968
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011930286539488362,
+      "loss": 1.019,
+      "step": 6969
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011928211431310412,
+      "loss": 1.0044,
+      "step": 6970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011926136236892398,
+      "loss": 0.9662,
+      "step": 6971
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001192406095632713,
+      "loss": 0.9712,
+      "step": 6972
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011921985589707427,
+      "loss": 0.9564,
+      "step": 6973
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011919910137126108,
+      "loss": 0.9621,
+      "step": 6974
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011917834598676005,
+      "loss": 0.9429,
+      "step": 6975
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001191575897444994,
+      "loss": 0.9228,
+      "step": 6976
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011913683264540752,
+      "loss": 0.9067,
+      "step": 6977
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011911607469041273,
+      "loss": 0.9333,
+      "step": 6978
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011909531588044348,
+      "loss": 0.9064,
+      "step": 6979
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011907455621642817,
+      "loss": 0.89,
+      "step": 6980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011905379569929535,
+      "loss": 0.8752,
+      "step": 6981
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001190330343299735,
+      "loss": 0.868,
+      "step": 6982
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001190122721093912,
+      "loss": 0.8462,
+      "step": 6983
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011899150903847702,
+      "loss": 0.8505,
+      "step": 6984
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011897074511815962,
+      "loss": 0.8519,
+      "step": 6985
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011894998034936766,
+      "loss": 0.834,
+      "step": 6986
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011892921473302987,
+      "loss": 0.8302,
+      "step": 6987
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011890844827007497,
+      "loss": 0.8491,
+      "step": 6988
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011888768096143179,
+      "loss": 0.8248,
+      "step": 6989
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011886691280802918,
+      "loss": 0.8281,
+      "step": 6990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001188461438107959,
+      "loss": 0.8104,
+      "step": 6991
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011882537397066093,
+      "loss": 0.8177,
+      "step": 6992
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011880460328855321,
+      "loss": 0.7951,
+      "step": 6993
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011878383176540169,
+      "loss": 0.7751,
+      "step": 6994
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001187630594021354,
+      "loss": 0.792,
+      "step": 6995
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011874228619968341,
+      "loss": 0.764,
+      "step": 6996
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011872151215897478,
+      "loss": 0.7634,
+      "step": 6997
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00011870073728093865,
+      "loss": 0.7665,
+      "step": 6998
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011867996156650416,
+      "loss": 0.7736,
+      "step": 6999
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011865918501660054,
+      "loss": 0.7607,
+      "step": 7000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011863840763215703,
+      "loss": 0.7529,
+      "step": 7001
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001186176294141029,
+      "loss": 0.7504,
+      "step": 7002
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011859685036336752,
+      "loss": 0.7685,
+      "step": 7003
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011857607048088014,
+      "loss": 0.7661,
+      "step": 7004
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001185552897675702,
+      "loss": 0.7461,
+      "step": 7005
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011853450822436713,
+      "loss": 0.7499,
+      "step": 7006
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011851372585220038,
+      "loss": 0.7226,
+      "step": 7007
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011849294265199946,
+      "loss": 0.7108,
+      "step": 7008
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011847215862469392,
+      "loss": 0.7249,
+      "step": 7009
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011845137377121329,
+      "loss": 0.7202,
+      "step": 7010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011843058809248724,
+      "loss": 0.7195,
+      "step": 7011
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011840980158944537,
+      "loss": 0.7005,
+      "step": 7012
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011838901426301741,
+      "loss": 0.7202,
+      "step": 7013
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011836822611413305,
+      "loss": 0.6823,
+      "step": 7014
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011834743714372203,
+      "loss": 0.6815,
+      "step": 7015
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011832664735271419,
+      "loss": 0.6952,
+      "step": 7016
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011830585674203933,
+      "loss": 0.6633,
+      "step": 7017
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011828506531262733,
+      "loss": 0.6564,
+      "step": 7018
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001182642730654081,
+      "loss": 0.6848,
+      "step": 7019
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001182434800013116,
+      "loss": 0.6724,
+      "step": 7020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011822268612126777,
+      "loss": 0.6648,
+      "step": 7021
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011820189142620663,
+      "loss": 0.6714,
+      "step": 7022
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011818109591705828,
+      "loss": 0.6514,
+      "step": 7023
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011816029959475274,
+      "loss": 0.6699,
+      "step": 7024
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011813950246022019,
+      "loss": 0.6638,
+      "step": 7025
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011811870451439075,
+      "loss": 0.6643,
+      "step": 7026
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011809790575819463,
+      "loss": 0.6269,
+      "step": 7027
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011807710619256207,
+      "loss": 0.657,
+      "step": 7028
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011805630581842333,
+      "loss": 0.6526,
+      "step": 7029
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011803550463670874,
+      "loss": 0.6296,
+      "step": 7030
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001180147026483486,
+      "loss": 0.6417,
+      "step": 7031
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011799389985427331,
+      "loss": 0.6356,
+      "step": 7032
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011797309625541333,
+      "loss": 0.6313,
+      "step": 7033
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011795229185269903,
+      "loss": 0.596,
+      "step": 7034
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011793148664706093,
+      "loss": 0.6058,
+      "step": 7035
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011791068063942954,
+      "loss": 0.6171,
+      "step": 7036
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011788987383073544,
+      "loss": 0.6214,
+      "step": 7037
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001178690662219092,
+      "loss": 0.6061,
+      "step": 7038
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011784825781388149,
+      "loss": 0.6082,
+      "step": 7039
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011782744860758292,
+      "loss": 0.6072,
+      "step": 7040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011780663860394422,
+      "loss": 0.5837,
+      "step": 7041
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011778582780389613,
+      "loss": 0.5822,
+      "step": 7042
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011776501620836944,
+      "loss": 0.6068,
+      "step": 7043
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001177442038182949,
+      "loss": 0.6004,
+      "step": 7044
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011772339063460338,
+      "loss": 0.5991,
+      "step": 7045
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011770257665822577,
+      "loss": 0.5888,
+      "step": 7046
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000117681761890093,
+      "loss": 0.6098,
+      "step": 7047
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011766094633113595,
+      "loss": 0.5779,
+      "step": 7048
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011764012998228567,
+      "loss": 0.5906,
+      "step": 7049
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011761931284447316,
+      "loss": 0.57,
+      "step": 7050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011759849491862948,
+      "loss": 0.5643,
+      "step": 7051
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011757767620568571,
+      "loss": 0.5819,
+      "step": 7052
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011755685670657302,
+      "loss": 0.5856,
+      "step": 7053
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001175360364222225,
+      "loss": 0.5904,
+      "step": 7054
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011751521535356537,
+      "loss": 0.5997,
+      "step": 7055
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011749439350153288,
+      "loss": 0.5725,
+      "step": 7056
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011747357086705628,
+      "loss": 0.5577,
+      "step": 7057
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011745274745106689,
+      "loss": 0.5424,
+      "step": 7058
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011743192325449603,
+      "loss": 0.5794,
+      "step": 7059
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00011741109827827505,
+      "loss": 0.5684,
+      "step": 7060
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011739027252333541,
+      "loss": 0.5538,
+      "step": 7061
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011736944599060853,
+      "loss": 0.562,
+      "step": 7062
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011734861868102585,
+      "loss": 0.5813,
+      "step": 7063
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011732779059551893,
+      "loss": 0.5421,
+      "step": 7064
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011730696173501925,
+      "loss": 0.5349,
+      "step": 7065
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011728613210045846,
+      "loss": 0.5547,
+      "step": 7066
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011726530169276815,
+      "loss": 0.5389,
+      "step": 7067
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011724447051287996,
+      "loss": 0.5355,
+      "step": 7068
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011722363856172556,
+      "loss": 0.5254,
+      "step": 7069
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011720280584023671,
+      "loss": 0.5468,
+      "step": 7070
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011718197234934511,
+      "loss": 0.5441,
+      "step": 7071
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011716113808998258,
+      "loss": 0.5312,
+      "step": 7072
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011714030306308093,
+      "loss": 0.5441,
+      "step": 7073
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011711946726957204,
+      "loss": 0.5217,
+      "step": 7074
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011709863071038776,
+      "loss": 0.5418,
+      "step": 7075
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011707779338646005,
+      "loss": 0.5399,
+      "step": 7076
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011705695529872082,
+      "loss": 0.5209,
+      "step": 7077
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011703611644810213,
+      "loss": 0.5284,
+      "step": 7078
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011701527683553594,
+      "loss": 0.5234,
+      "step": 7079
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011699443646195436,
+      "loss": 0.5021,
+      "step": 7080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011697359532828947,
+      "loss": 0.5224,
+      "step": 7081
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011695275343547337,
+      "loss": 0.5056,
+      "step": 7082
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011693191078443827,
+      "loss": 0.5194,
+      "step": 7083
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011691106737611633,
+      "loss": 0.5085,
+      "step": 7084
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011689022321143978,
+      "loss": 0.5121,
+      "step": 7085
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011686937829134093,
+      "loss": 0.5296,
+      "step": 7086
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011684853261675202,
+      "loss": 0.5152,
+      "step": 7087
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011682768618860538,
+      "loss": 0.5099,
+      "step": 7088
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011680683900783344,
+      "loss": 0.5082,
+      "step": 7089
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00011678599107536854,
+      "loss": 0.4922,
+      "step": 7090
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011676514239214313,
+      "loss": 0.5117,
+      "step": 7091
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011674429295908969,
+      "loss": 0.5062,
+      "step": 7092
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011672344277714073,
+      "loss": 0.4981,
+      "step": 7093
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011670259184722873,
+      "loss": 0.4991,
+      "step": 7094
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001166817401702863,
+      "loss": 0.4946,
+      "step": 7095
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011666088774724602,
+      "loss": 0.52,
+      "step": 7096
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011664003457904054,
+      "loss": 0.4926,
+      "step": 7097
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011661918066660252,
+      "loss": 0.4924,
+      "step": 7098
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011659832601086462,
+      "loss": 0.4938,
+      "step": 7099
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011657747061275965,
+      "loss": 0.4848,
+      "step": 7100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011655661447322033,
+      "loss": 0.4768,
+      "step": 7101
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011653575759317946,
+      "loss": 0.4818,
+      "step": 7102
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001165148999735699,
+      "loss": 0.4776,
+      "step": 7103
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001164940416153245,
+      "loss": 0.4964,
+      "step": 7104
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011647318251937611,
+      "loss": 0.4717,
+      "step": 7105
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011645232268665774,
+      "loss": 0.4727,
+      "step": 7106
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011643146211810231,
+      "loss": 0.4929,
+      "step": 7107
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011641060081464283,
+      "loss": 0.4995,
+      "step": 7108
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011638973877721232,
+      "loss": 0.4944,
+      "step": 7109
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011636887600674386,
+      "loss": 0.5016,
+      "step": 7110
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011634801250417056,
+      "loss": 0.4821,
+      "step": 7111
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001163271482704255,
+      "loss": 0.4805,
+      "step": 7112
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011630628330644191,
+      "loss": 0.4939,
+      "step": 7113
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001162854176131529,
+      "loss": 0.4777,
+      "step": 7114
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011626455119149175,
+      "loss": 0.4767,
+      "step": 7115
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011624368404239172,
+      "loss": 0.4657,
+      "step": 7116
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011622281616678609,
+      "loss": 0.4728,
+      "step": 7117
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011620194756560818,
+      "loss": 0.4584,
+      "step": 7118
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011618107823979136,
+      "loss": 0.4845,
+      "step": 7119
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000116160208190269,
+      "loss": 0.4698,
+      "step": 7120
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00011613933741797456,
+      "loss": 0.4602,
+      "step": 7121
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011611846592384146,
+      "loss": 0.4496,
+      "step": 7122
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001160975937088032,
+      "loss": 0.4722,
+      "step": 7123
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011607672077379328,
+      "loss": 0.457,
+      "step": 7124
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011605584711974527,
+      "loss": 0.4529,
+      "step": 7125
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011603497274759274,
+      "loss": 0.4444,
+      "step": 7126
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011601409765826932,
+      "loss": 0.4514,
+      "step": 7127
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011599322185270866,
+      "loss": 0.4579,
+      "step": 7128
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011597234533184443,
+      "loss": 0.4546,
+      "step": 7129
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001159514680966103,
+      "loss": 0.4645,
+      "step": 7130
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011593059014794007,
+      "loss": 0.4359,
+      "step": 7131
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011590971148676752,
+      "loss": 0.4648,
+      "step": 7132
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011588883211402646,
+      "loss": 0.4584,
+      "step": 7133
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011586795203065066,
+      "loss": 0.4631,
+      "step": 7134
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011584707123757403,
+      "loss": 0.4387,
+      "step": 7135
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011582618973573049,
+      "loss": 0.4615,
+      "step": 7136
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011580530752605396,
+      "loss": 0.4538,
+      "step": 7137
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011578442460947842,
+      "loss": 0.4434,
+      "step": 7138
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011576354098693785,
+      "loss": 0.4452,
+      "step": 7139
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011574265665936628,
+      "loss": 0.4409,
+      "step": 7140
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011572177162769777,
+      "loss": 0.4488,
+      "step": 7141
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011570088589286641,
+      "loss": 0.4297,
+      "step": 7142
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011567999945580633,
+      "loss": 0.4477,
+      "step": 7143
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011565911231745168,
+      "loss": 0.4401,
+      "step": 7144
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011563822447873664,
+      "loss": 0.4404,
+      "step": 7145
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011561733594059548,
+      "loss": 0.428,
+      "step": 7146
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011559644670396235,
+      "loss": 0.4344,
+      "step": 7147
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011557555676977161,
+      "loss": 0.4496,
+      "step": 7148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011555466613895754,
+      "loss": 0.4349,
+      "step": 7149
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011553377481245449,
+      "loss": 0.4426,
+      "step": 7150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011551288279119681,
+      "loss": 0.4233,
+      "step": 7151
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00011549199007611895,
+      "loss": 0.4047,
+      "step": 7152
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011547109666815533,
+      "loss": 0.4349,
+      "step": 7153
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001154502025682404,
+      "loss": 0.4282,
+      "step": 7154
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011542930777730865,
+      "loss": 0.4412,
+      "step": 7155
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011540841229629462,
+      "loss": 0.4054,
+      "step": 7156
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011538751612613288,
+      "loss": 0.4254,
+      "step": 7157
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000115366619267758,
+      "loss": 0.4287,
+      "step": 7158
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011534572172210465,
+      "loss": 0.4373,
+      "step": 7159
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001153248234901074,
+      "loss": 0.4076,
+      "step": 7160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011530392457270101,
+      "loss": 0.4118,
+      "step": 7161
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011528302497082016,
+      "loss": 0.424,
+      "step": 7162
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011526212468539959,
+      "loss": 0.4265,
+      "step": 7163
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011524122371737407,
+      "loss": 0.4338,
+      "step": 7164
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001152203220676784,
+      "loss": 0.4243,
+      "step": 7165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011519941973724743,
+      "loss": 0.4382,
+      "step": 7166
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011517851672701603,
+      "loss": 0.422,
+      "step": 7167
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001151576130379191,
+      "loss": 0.4072,
+      "step": 7168
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011513670867089152,
+      "loss": 0.414,
+      "step": 7169
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011511580362686831,
+      "loss": 0.4163,
+      "step": 7170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011509489790678443,
+      "loss": 0.3963,
+      "step": 7171
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011507399151157487,
+      "loss": 0.4343,
+      "step": 7172
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011505308444217475,
+      "loss": 0.4164,
+      "step": 7173
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011503217669951906,
+      "loss": 0.424,
+      "step": 7174
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011501126828454297,
+      "loss": 0.404,
+      "step": 7175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011499035919818157,
+      "loss": 0.4175,
+      "step": 7176
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011496944944137004,
+      "loss": 0.4129,
+      "step": 7177
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011494853901504363,
+      "loss": 0.4209,
+      "step": 7178
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011492762792013751,
+      "loss": 0.4164,
+      "step": 7179
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011490671615758697,
+      "loss": 0.4189,
+      "step": 7180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011488580372832727,
+      "loss": 0.4076,
+      "step": 7181
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00011486489063329374,
+      "loss": 0.4262,
+      "step": 7182
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011484397687342174,
+      "loss": 0.4068,
+      "step": 7183
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011482306244964662,
+      "loss": 0.4025,
+      "step": 7184
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001148021473629038,
+      "loss": 0.4165,
+      "step": 7185
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011478123161412873,
+      "loss": 0.4064,
+      "step": 7186
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011476031520425683,
+      "loss": 0.3942,
+      "step": 7187
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011473939813422366,
+      "loss": 0.4078,
+      "step": 7188
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011471848040496469,
+      "loss": 0.3948,
+      "step": 7189
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001146975620174155,
+      "loss": 0.4062,
+      "step": 7190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011467664297251168,
+      "loss": 0.4039,
+      "step": 7191
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011465572327118884,
+      "loss": 0.4075,
+      "step": 7192
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011463480291438262,
+      "loss": 0.4036,
+      "step": 7193
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011461388190302867,
+      "loss": 0.389,
+      "step": 7194
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011459296023806269,
+      "loss": 0.391,
+      "step": 7195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011457203792042047,
+      "loss": 0.3936,
+      "step": 7196
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001145511149510377,
+      "loss": 0.3863,
+      "step": 7197
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011453019133085019,
+      "loss": 0.3924,
+      "step": 7198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011450926706079379,
+      "loss": 0.4091,
+      "step": 7199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001144883421418043,
+      "loss": 0.3994,
+      "step": 7200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011446741657481764,
+      "loss": 0.3943,
+      "step": 7201
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011444649036076964,
+      "loss": 0.3795,
+      "step": 7202
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011442556350059637,
+      "loss": 0.3857,
+      "step": 7203
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011440463599523363,
+      "loss": 0.3867,
+      "step": 7204
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011438370784561753,
+      "loss": 0.3813,
+      "step": 7205
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011436277905268404,
+      "loss": 0.3796,
+      "step": 7206
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001143418496173692,
+      "loss": 0.3692,
+      "step": 7207
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011432091954060913,
+      "loss": 0.398,
+      "step": 7208
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001142999888233399,
+      "loss": 0.375,
+      "step": 7209
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011427905746649767,
+      "loss": 0.41,
+      "step": 7210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011425812547101861,
+      "loss": 0.3887,
+      "step": 7211
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011423719283783887,
+      "loss": 0.3815,
+      "step": 7212
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00011421625956789468,
+      "loss": 0.3836,
+      "step": 7213
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011419532566212232,
+      "loss": 0.3952,
+      "step": 7214
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011417439112145805,
+      "loss": 0.3912,
+      "step": 7215
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011415345594683819,
+      "loss": 0.387,
+      "step": 7216
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011413252013919906,
+      "loss": 0.3848,
+      "step": 7217
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011411158369947704,
+      "loss": 0.3794,
+      "step": 7218
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011409064662860846,
+      "loss": 0.3683,
+      "step": 7219
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011406970892752982,
+      "loss": 0.4026,
+      "step": 7220
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011404877059717752,
+      "loss": 0.3812,
+      "step": 7221
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011402783163848805,
+      "loss": 0.377,
+      "step": 7222
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011400689205239791,
+      "loss": 0.3722,
+      "step": 7223
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011398595183984367,
+      "loss": 0.3822,
+      "step": 7224
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001139650110017618,
+      "loss": 0.3892,
+      "step": 7225
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011394406953908898,
+      "loss": 0.3754,
+      "step": 7226
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011392312745276176,
+      "loss": 0.3782,
+      "step": 7227
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011390218474371683,
+      "loss": 0.3835,
+      "step": 7228
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011388124141289082,
+      "loss": 0.3813,
+      "step": 7229
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011386029746122049,
+      "loss": 0.4003,
+      "step": 7230
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001138393528896425,
+      "loss": 0.3793,
+      "step": 7231
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011381840769909362,
+      "loss": 0.3766,
+      "step": 7232
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001137974618905107,
+      "loss": 0.3856,
+      "step": 7233
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011377651546483046,
+      "loss": 0.377,
+      "step": 7234
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011375556842298977,
+      "loss": 0.3773,
+      "step": 7235
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011373462076592551,
+      "loss": 0.3775,
+      "step": 7236
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011371367249457457,
+      "loss": 0.3716,
+      "step": 7237
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011369272360987384,
+      "loss": 0.3738,
+      "step": 7238
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001136717741127603,
+      "loss": 0.3817,
+      "step": 7239
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011365082400417092,
+      "loss": 0.3752,
+      "step": 7240
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011362987328504271,
+      "loss": 0.3738,
+      "step": 7241
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011360892195631266,
+      "loss": 0.3701,
+      "step": 7242
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00011358797001891787,
+      "loss": 0.3745,
+      "step": 7243
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001135670174737954,
+      "loss": 0.3788,
+      "step": 7244
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001135460643218824,
+      "loss": 0.3714,
+      "step": 7245
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011352511056411595,
+      "loss": 0.3759,
+      "step": 7246
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011350415620143325,
+      "loss": 0.3629,
+      "step": 7247
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011348320123477148,
+      "loss": 0.3738,
+      "step": 7248
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011346224566506787,
+      "loss": 0.3668,
+      "step": 7249
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011344128949325965,
+      "loss": 0.3698,
+      "step": 7250
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011342033272028411,
+      "loss": 0.3693,
+      "step": 7251
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011339937534707856,
+      "loss": 0.3714,
+      "step": 7252
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011337841737458034,
+      "loss": 0.3742,
+      "step": 7253
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011335745880372675,
+      "loss": 0.3694,
+      "step": 7254
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001133364996354552,
+      "loss": 0.3715,
+      "step": 7255
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001133155398707031,
+      "loss": 0.3625,
+      "step": 7256
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011329457951040787,
+      "loss": 0.3752,
+      "step": 7257
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.000113273618555507,
+      "loss": 0.3955,
+      "step": 7258
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011325265700693794,
+      "loss": 0.3839,
+      "step": 7259
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011323169486563823,
+      "loss": 0.3547,
+      "step": 7260
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011321073213254541,
+      "loss": 0.3822,
+      "step": 7261
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011318976880859704,
+      "loss": 0.3436,
+      "step": 7262
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011316880489473074,
+      "loss": 0.3567,
+      "step": 7263
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011314784039188406,
+      "loss": 0.3694,
+      "step": 7264
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011312687530099473,
+      "loss": 0.3647,
+      "step": 7265
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011310590962300036,
+      "loss": 0.3727,
+      "step": 7266
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011308494335883865,
+      "loss": 0.3707,
+      "step": 7267
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001130639765094474,
+      "loss": 0.357,
+      "step": 7268
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011304300907576428,
+      "loss": 0.3791,
+      "step": 7269
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011302204105872709,
+      "loss": 0.3721,
+      "step": 7270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011300107245927365,
+      "loss": 0.3504,
+      "step": 7271
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011298010327834176,
+      "loss": 0.3604,
+      "step": 7272
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00011295913351686933,
+      "loss": 0.3566,
+      "step": 7273
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001129381631757942,
+      "loss": 0.356,
+      "step": 7274
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011291719225605426,
+      "loss": 0.3633,
+      "step": 7275
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011289622075858747,
+      "loss": 0.3715,
+      "step": 7276
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011287524868433179,
+      "loss": 0.3474,
+      "step": 7277
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011285427603422521,
+      "loss": 0.3516,
+      "step": 7278
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001128333028092057,
+      "loss": 0.3496,
+      "step": 7279
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011281232901021134,
+      "loss": 0.3583,
+      "step": 7280
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001127913546381802,
+      "loss": 0.3565,
+      "step": 7281
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011277037969405034,
+      "loss": 0.3661,
+      "step": 7282
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011274940417875988,
+      "loss": 0.3511,
+      "step": 7283
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011272842809324694,
+      "loss": 0.3564,
+      "step": 7284
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011270745143844972,
+      "loss": 0.3671,
+      "step": 7285
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011268647421530638,
+      "loss": 0.3491,
+      "step": 7286
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011266549642475515,
+      "loss": 0.3635,
+      "step": 7287
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011264451806773426,
+      "loss": 0.3616,
+      "step": 7288
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000112623539145182,
+      "loss": 0.3431,
+      "step": 7289
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011260255965803663,
+      "loss": 0.3627,
+      "step": 7290
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001125815796072365,
+      "loss": 0.3442,
+      "step": 7291
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001125605989937199,
+      "loss": 0.3651,
+      "step": 7292
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011253961781842526,
+      "loss": 0.3569,
+      "step": 7293
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011251863608229093,
+      "loss": 0.3641,
+      "step": 7294
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011249765378625536,
+      "loss": 0.3491,
+      "step": 7295
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011247667093125696,
+      "loss": 0.3619,
+      "step": 7296
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011245568751823419,
+      "loss": 0.3559,
+      "step": 7297
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011243470354812555,
+      "loss": 0.3769,
+      "step": 7298
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011241371902186959,
+      "loss": 0.3708,
+      "step": 7299
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011239273394040482,
+      "loss": 0.36,
+      "step": 7300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011237174830466981,
+      "loss": 0.3715,
+      "step": 7301
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011235076211560316,
+      "loss": 0.3589,
+      "step": 7302
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011232977537414351,
+      "loss": 0.3528,
+      "step": 7303
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011230878808122945,
+      "loss": 0.3613,
+      "step": 7304
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00011228780023779965,
+      "loss": 0.329,
+      "step": 7305
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011226681184479285,
+      "loss": 0.3527,
+      "step": 7306
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011224582290314769,
+      "loss": 0.3562,
+      "step": 7307
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011222483341380299,
+      "loss": 0.3732,
+      "step": 7308
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011220384337769747,
+      "loss": 0.3628,
+      "step": 7309
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001121828527957699,
+      "loss": 0.3549,
+      "step": 7310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011216186166895915,
+      "loss": 0.368,
+      "step": 7311
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011214086999820399,
+      "loss": 0.3518,
+      "step": 7312
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011211987778444337,
+      "loss": 0.3388,
+      "step": 7313
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011209888502861608,
+      "loss": 0.3442,
+      "step": 7314
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011207789173166108,
+      "loss": 0.3465,
+      "step": 7315
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011205689789451728,
+      "loss": 0.3518,
+      "step": 7316
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011203590351812367,
+      "loss": 0.3456,
+      "step": 7317
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011201490860341922,
+      "loss": 0.3564,
+      "step": 7318
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011199391315134291,
+      "loss": 0.3346,
+      "step": 7319
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001119729171628338,
+      "loss": 0.3457,
+      "step": 7320
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011195192063883093,
+      "loss": 0.3271,
+      "step": 7321
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001119309235802734,
+      "loss": 0.3558,
+      "step": 7322
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001119099259881003,
+      "loss": 0.3534,
+      "step": 7323
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011188892786325076,
+      "loss": 0.3489,
+      "step": 7324
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011186792920666389,
+      "loss": 0.355,
+      "step": 7325
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011184693001927893,
+      "loss": 0.3375,
+      "step": 7326
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011182593030203501,
+      "loss": 0.3495,
+      "step": 7327
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001118049300558714,
+      "loss": 0.3429,
+      "step": 7328
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011178392928172734,
+      "loss": 0.3363,
+      "step": 7329
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011176292798054209,
+      "loss": 0.3539,
+      "step": 7330
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011174192615325494,
+      "loss": 0.3479,
+      "step": 7331
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011172092380080519,
+      "loss": 0.3412,
+      "step": 7332
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011169992092413222,
+      "loss": 0.3293,
+      "step": 7333
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011167891752417537,
+      "loss": 0.3609,
+      "step": 7334
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00011165791360187398,
+      "loss": 0.3439,
+      "step": 7335
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011163690915816753,
+      "loss": 0.3358,
+      "step": 7336
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011161590419399544,
+      "loss": 0.3447,
+      "step": 7337
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011159489871029713,
+      "loss": 0.3491,
+      "step": 7338
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011157389270801211,
+      "loss": 0.3479,
+      "step": 7339
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011155288618807989,
+      "loss": 0.3303,
+      "step": 7340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011153187915143995,
+      "loss": 0.3399,
+      "step": 7341
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011151087159903186,
+      "loss": 0.3387,
+      "step": 7342
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011148986353179525,
+      "loss": 0.3312,
+      "step": 7343
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011146885495066964,
+      "loss": 0.3723,
+      "step": 7344
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011144784585659465,
+      "loss": 0.3374,
+      "step": 7345
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011142683625050993,
+      "loss": 0.3468,
+      "step": 7346
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001114058261333552,
+      "loss": 0.3512,
+      "step": 7347
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011138481550607008,
+      "loss": 0.3259,
+      "step": 7348
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011136380436959429,
+      "loss": 0.3283,
+      "step": 7349
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011134279272486757,
+      "loss": 0.3435,
+      "step": 7350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001113217805728297,
+      "loss": 0.3131,
+      "step": 7351
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011130076791442041,
+      "loss": 0.331,
+      "step": 7352
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011127975475057954,
+      "loss": 0.3598,
+      "step": 7353
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001112587410822469,
+      "loss": 0.342,
+      "step": 7354
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011123772691036231,
+      "loss": 0.3442,
+      "step": 7355
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011121671223586568,
+      "loss": 0.3268,
+      "step": 7356
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011119569705969688,
+      "loss": 0.3567,
+      "step": 7357
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011117468138279581,
+      "loss": 0.3409,
+      "step": 7358
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011115366520610243,
+      "loss": 0.3403,
+      "step": 7359
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011113264853055664,
+      "loss": 0.3355,
+      "step": 7360
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011111163135709855,
+      "loss": 0.342,
+      "step": 7361
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011109061368666802,
+      "loss": 0.3414,
+      "step": 7362
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011106959552020513,
+      "loss": 0.3411,
+      "step": 7363
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011104857685864993,
+      "loss": 0.3304,
+      "step": 7364
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00011102755770294248,
+      "loss": 0.3547,
+      "step": 7365
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001110065380540229,
+      "loss": 0.3396,
+      "step": 7366
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011098551791283127,
+      "loss": 0.3333,
+      "step": 7367
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011096449728030773,
+      "loss": 0.3491,
+      "step": 7368
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011094347615739244,
+      "loss": 0.344,
+      "step": 7369
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011092245454502557,
+      "loss": 0.3264,
+      "step": 7370
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011090143244414732,
+      "loss": 0.3258,
+      "step": 7371
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011088040985569795,
+      "loss": 0.3392,
+      "step": 7372
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011085938678061767,
+      "loss": 0.3453,
+      "step": 7373
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011083836321984674,
+      "loss": 0.3343,
+      "step": 7374
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011081733917432546,
+      "loss": 0.3235,
+      "step": 7375
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011079631464499414,
+      "loss": 0.3347,
+      "step": 7376
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011077528963279309,
+      "loss": 0.339,
+      "step": 7377
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001107542641386627,
+      "loss": 0.3376,
+      "step": 7378
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001107332381635433,
+      "loss": 0.3376,
+      "step": 7379
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011071221170837534,
+      "loss": 0.349,
+      "step": 7380
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011069118477409921,
+      "loss": 0.3481,
+      "step": 7381
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011067015736165533,
+      "loss": 0.3265,
+      "step": 7382
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001106491294719842,
+      "loss": 0.3365,
+      "step": 7383
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011062810110602626,
+      "loss": 0.3315,
+      "step": 7384
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011060707226472202,
+      "loss": 0.3561,
+      "step": 7385
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011058604294901202,
+      "loss": 0.336,
+      "step": 7386
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011056501315983681,
+      "loss": 0.3415,
+      "step": 7387
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011054398289813695,
+      "loss": 0.3305,
+      "step": 7388
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000110522952164853,
+      "loss": 0.339,
+      "step": 7389
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001105019209609256,
+      "loss": 0.3429,
+      "step": 7390
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011048088928729536,
+      "loss": 0.3188,
+      "step": 7391
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011045985714490294,
+      "loss": 0.3358,
+      "step": 7392
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011043882453468905,
+      "loss": 0.3391,
+      "step": 7393
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001104177914575943,
+      "loss": 0.3241,
+      "step": 7394
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011039675791455947,
+      "loss": 0.3311,
+      "step": 7395
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011037572390652526,
+      "loss": 0.3229,
+      "step": 7396
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00011035468943443243,
+      "loss": 0.3272,
+      "step": 7397
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011033365449922178,
+      "loss": 0.3219,
+      "step": 7398
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011031261910183409,
+      "loss": 0.3389,
+      "step": 7399
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011029158324321016,
+      "loss": 0.3417,
+      "step": 7400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011027054692429084,
+      "loss": 0.3313,
+      "step": 7401
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.000110249510146017,
+      "loss": 0.3256,
+      "step": 7402
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011022847290932954,
+      "loss": 0.3386,
+      "step": 7403
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011020743521516929,
+      "loss": 0.3341,
+      "step": 7404
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011018639706447722,
+      "loss": 0.3432,
+      "step": 7405
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011016535845819424,
+      "loss": 0.3261,
+      "step": 7406
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011014431939726135,
+      "loss": 0.3288,
+      "step": 7407
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001101232798826195,
+      "loss": 0.3363,
+      "step": 7408
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011010223991520971,
+      "loss": 0.3291,
+      "step": 7409
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.000110081199495973,
+      "loss": 0.3105,
+      "step": 7410
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001100601586258504,
+      "loss": 0.339,
+      "step": 7411
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011003911730578296,
+      "loss": 0.3294,
+      "step": 7412
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00011001807553671179,
+      "loss": 0.3272,
+      "step": 7413
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010999703331957798,
+      "loss": 0.3292,
+      "step": 7414
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010997599065532267,
+      "loss": 0.3238,
+      "step": 7415
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010995494754488695,
+      "loss": 0.3491,
+      "step": 7416
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010993390398921204,
+      "loss": 0.316,
+      "step": 7417
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010991285998923908,
+      "loss": 0.3118,
+      "step": 7418
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010989181554590931,
+      "loss": 0.3364,
+      "step": 7419
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010987077066016389,
+      "loss": 0.3147,
+      "step": 7420
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010984972533294414,
+      "loss": 0.3184,
+      "step": 7421
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010982867956519128,
+      "loss": 0.3216,
+      "step": 7422
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010980763335784661,
+      "loss": 0.3055,
+      "step": 7423
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010978658671185138,
+      "loss": 0.3374,
+      "step": 7424
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010976553962814698,
+      "loss": 0.3211,
+      "step": 7425
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010974449210767468,
+      "loss": 0.3295,
+      "step": 7426
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010972344415137588,
+      "loss": 0.3252,
+      "step": 7427
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010970239576019197,
+      "loss": 0.3219,
+      "step": 7428
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001096813469350643,
+      "loss": 0.31,
+      "step": 7429
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010966029767693435,
+      "loss": 0.3206,
+      "step": 7430
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010963924798674352,
+      "loss": 0.3387,
+      "step": 7431
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010961819786543326,
+      "loss": 0.3165,
+      "step": 7432
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010959714731394509,
+      "loss": 0.3444,
+      "step": 7433
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010957609633322044,
+      "loss": 0.3256,
+      "step": 7434
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010955504492420085,
+      "loss": 0.3269,
+      "step": 7435
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010953399308782786,
+      "loss": 0.3298,
+      "step": 7436
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010951294082504304,
+      "loss": 0.3109,
+      "step": 7437
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010949188813678792,
+      "loss": 0.3105,
+      "step": 7438
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010947083502400415,
+      "loss": 0.3254,
+      "step": 7439
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010944978148763327,
+      "loss": 0.3203,
+      "step": 7440
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010942872752861695,
+      "loss": 0.3238,
+      "step": 7441
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010940767314789682,
+      "loss": 0.3119,
+      "step": 7442
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010938661834641457,
+      "loss": 0.3231,
+      "step": 7443
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010936556312511186,
+      "loss": 0.3213,
+      "step": 7444
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010934450748493041,
+      "loss": 0.3131,
+      "step": 7445
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001093234514268119,
+      "loss": 0.3114,
+      "step": 7446
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010930239495169814,
+      "loss": 0.3209,
+      "step": 7447
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010928133806053085,
+      "loss": 0.3139,
+      "step": 7448
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001092602807542518,
+      "loss": 0.3168,
+      "step": 7449
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001092392230338028,
+      "loss": 0.3108,
+      "step": 7450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010921816490012567,
+      "loss": 0.3121,
+      "step": 7451
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010919710635416224,
+      "loss": 0.3039,
+      "step": 7452
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010917604739685439,
+      "loss": 0.3205,
+      "step": 7453
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010915498802914393,
+      "loss": 0.3125,
+      "step": 7454
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010913392825197279,
+      "loss": 0.3122,
+      "step": 7455
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010911286806628287,
+      "loss": 0.3192,
+      "step": 7456
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010909180747301608,
+      "loss": 0.3216,
+      "step": 7457
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001090707464731144,
+      "loss": 0.3257,
+      "step": 7458
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010904968506751974,
+      "loss": 0.3263,
+      "step": 7459
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010902862325717413,
+      "loss": 0.3229,
+      "step": 7460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010900756104301955,
+      "loss": 0.3182,
+      "step": 7461
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010898649842599801,
+      "loss": 0.3172,
+      "step": 7462
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010896543540705157,
+      "loss": 0.3208,
+      "step": 7463
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010894437198712226,
+      "loss": 0.3484,
+      "step": 7464
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010892330816715213,
+      "loss": 0.3214,
+      "step": 7465
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001089022439480833,
+      "loss": 0.3111,
+      "step": 7466
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010888117933085787,
+      "loss": 0.2989,
+      "step": 7467
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010886011431641796,
+      "loss": 0.3168,
+      "step": 7468
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001088390489057057,
+      "loss": 0.3221,
+      "step": 7469
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001088179830996633,
+      "loss": 0.3214,
+      "step": 7470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010879691689923286,
+      "loss": 0.3206,
+      "step": 7471
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010877585030535662,
+      "loss": 0.3107,
+      "step": 7472
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010875478331897682,
+      "loss": 0.3293,
+      "step": 7473
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010873371594103561,
+      "loss": 0.3186,
+      "step": 7474
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010871264817247531,
+      "loss": 0.3129,
+      "step": 7475
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010869158001423814,
+      "loss": 0.3223,
+      "step": 7476
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001086705114672664,
+      "loss": 0.3227,
+      "step": 7477
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001086494425325024,
+      "loss": 0.3123,
+      "step": 7478
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010862837321088844,
+      "loss": 0.3301,
+      "step": 7479
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010860730350336685,
+      "loss": 0.3283,
+      "step": 7480
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010858623341088001,
+      "loss": 0.3136,
+      "step": 7481
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010856516293437026,
+      "loss": 0.3051,
+      "step": 7482
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010854409207478002,
+      "loss": 0.3182,
+      "step": 7483
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010852302083305166,
+      "loss": 0.3232,
+      "step": 7484
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001085019492101276,
+      "loss": 0.3185,
+      "step": 7485
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010848087720695029,
+      "loss": 0.3157,
+      "step": 7486
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010845980482446219,
+      "loss": 0.3133,
+      "step": 7487
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010843873206360575,
+      "loss": 0.321,
+      "step": 7488
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010841765892532347,
+      "loss": 0.3114,
+      "step": 7489
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010839658541055786,
+      "loss": 0.3337,
+      "step": 7490
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010837551152025143,
+      "loss": 0.3033,
+      "step": 7491
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010835443725534676,
+      "loss": 0.3104,
+      "step": 7492
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010833336261678635,
+      "loss": 0.3171,
+      "step": 7493
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001083122876055128,
+      "loss": 0.3309,
+      "step": 7494
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010829121222246868,
+      "loss": 0.3106,
+      "step": 7495
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001082701364685966,
+      "loss": 0.3071,
+      "step": 7496
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010824906034483919,
+      "loss": 0.3159,
+      "step": 7497
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001082279838521391,
+      "loss": 0.3182,
+      "step": 7498
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010820690699143896,
+      "loss": 0.305,
+      "step": 7499
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010818582976368148,
+      "loss": 0.3015,
+      "step": 7500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001081647521698093,
+      "loss": 0.3207,
+      "step": 7501
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010814367421076515,
+      "loss": 0.3232,
+      "step": 7502
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010812259588749178,
+      "loss": 0.3209,
+      "step": 7503
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010810151720093186,
+      "loss": 0.3152,
+      "step": 7504
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010808043815202818,
+      "loss": 0.3237,
+      "step": 7505
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001080593587417235,
+      "loss": 0.3177,
+      "step": 7506
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010803827897096064,
+      "loss": 0.293,
+      "step": 7507
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010801719884068236,
+      "loss": 0.308,
+      "step": 7508
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010799611835183148,
+      "loss": 0.3249,
+      "step": 7509
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010797503750535086,
+      "loss": 0.3185,
+      "step": 7510
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010795395630218332,
+      "loss": 0.3031,
+      "step": 7511
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010793287474327176,
+      "loss": 0.3158,
+      "step": 7512
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010791179282955903,
+      "loss": 0.3126,
+      "step": 7513
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010789071056198803,
+      "loss": 0.3234,
+      "step": 7514
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001078696279415017,
+      "loss": 0.3159,
+      "step": 7515
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010784854496904293,
+      "loss": 0.3222,
+      "step": 7516
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010782746164555473,
+      "loss": 0.3218,
+      "step": 7517
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010780637797197999,
+      "loss": 0.3105,
+      "step": 7518
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010778529394926171,
+      "loss": 0.3206,
+      "step": 7519
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010776420957834287,
+      "loss": 0.3108,
+      "step": 7520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001077431248601665,
+      "loss": 0.3152,
+      "step": 7521
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010772203979567564,
+      "loss": 0.2954,
+      "step": 7522
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010770095438581331,
+      "loss": 0.2984,
+      "step": 7523
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010767986863152253,
+      "loss": 0.3019,
+      "step": 7524
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010765878253374641,
+      "loss": 0.2968,
+      "step": 7525
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000107637696093428,
+      "loss": 0.3009,
+      "step": 7526
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010761660931151045,
+      "loss": 0.3102,
+      "step": 7527
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010759552218893683,
+      "loss": 0.3058,
+      "step": 7528
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010757443472665028,
+      "loss": 0.3057,
+      "step": 7529
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010755334692559396,
+      "loss": 0.3002,
+      "step": 7530
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010753225878671103,
+      "loss": 0.311,
+      "step": 7531
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010751117031094464,
+      "loss": 0.3217,
+      "step": 7532
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010749008149923805,
+      "loss": 0.3026,
+      "step": 7533
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010746899235253436,
+      "loss": 0.3036,
+      "step": 7534
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010744790287177688,
+      "loss": 0.304,
+      "step": 7535
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010742681305790878,
+      "loss": 0.3075,
+      "step": 7536
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010740572291187337,
+      "loss": 0.3044,
+      "step": 7537
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010738463243461388,
+      "loss": 0.3144,
+      "step": 7538
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001073635416270736,
+      "loss": 0.3033,
+      "step": 7539
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010734245049019582,
+      "loss": 0.3236,
+      "step": 7540
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010732135902492388,
+      "loss": 0.3137,
+      "step": 7541
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010730026723220105,
+      "loss": 0.3081,
+      "step": 7542
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010727917511297074,
+      "loss": 0.3142,
+      "step": 7543
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010725808266817623,
+      "loss": 0.3243,
+      "step": 7544
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010723698989876092,
+      "loss": 0.2938,
+      "step": 7545
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010721589680566819,
+      "loss": 0.3239,
+      "step": 7546
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010719480338984147,
+      "loss": 0.3027,
+      "step": 7547
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010717370965222411,
+      "loss": 0.3047,
+      "step": 7548
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001071526155937596,
+      "loss": 0.3183,
+      "step": 7549
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010713152121539132,
+      "loss": 0.3045,
+      "step": 7550
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010711042651806278,
+      "loss": 0.315,
+      "step": 7551
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010708933150271741,
+      "loss": 0.3163,
+      "step": 7552
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010706823617029873,
+      "loss": 0.3124,
+      "step": 7553
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001070471405217502,
+      "loss": 0.2916,
+      "step": 7554
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010702604455801535,
+      "loss": 0.3109,
+      "step": 7555
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010700494828003769,
+      "loss": 0.3225,
+      "step": 7556
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010698385168876079,
+      "loss": 0.2873,
+      "step": 7557
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010696275478512818,
+      "loss": 0.2957,
+      "step": 7558
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010694165757008343,
+      "loss": 0.3237,
+      "step": 7559
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010692056004457014,
+      "loss": 0.3242,
+      "step": 7560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010689946220953189,
+      "loss": 0.3038,
+      "step": 7561
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010687836406591227,
+      "loss": 0.3029,
+      "step": 7562
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010685726561465497,
+      "loss": 0.3033,
+      "step": 7563
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010683616685670355,
+      "loss": 0.2956,
+      "step": 7564
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001068150677930017,
+      "loss": 0.2935,
+      "step": 7565
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010679396842449307,
+      "loss": 0.3088,
+      "step": 7566
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010677286875212136,
+      "loss": 0.2801,
+      "step": 7567
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010675176877683024,
+      "loss": 0.3192,
+      "step": 7568
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010673066849956341,
+      "loss": 0.3188,
+      "step": 7569
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010670956792126463,
+      "loss": 0.3067,
+      "step": 7570
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010668846704287758,
+      "loss": 0.3281,
+      "step": 7571
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010666736586534605,
+      "loss": 0.2928,
+      "step": 7572
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010664626438961378,
+      "loss": 0.3024,
+      "step": 7573
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010662516261662452,
+      "loss": 0.3015,
+      "step": 7574
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010660406054732208,
+      "loss": 0.3131,
+      "step": 7575
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010658295818265027,
+      "loss": 0.3194,
+      "step": 7576
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010656185552355286,
+      "loss": 0.3203,
+      "step": 7577
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010654075257097374,
+      "loss": 0.3115,
+      "step": 7578
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001065196493258567,
+      "loss": 0.3181,
+      "step": 7579
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010649854578914558,
+      "loss": 0.3087,
+      "step": 7580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010647744196178429,
+      "loss": 0.299,
+      "step": 7581
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010645633784471669,
+      "loss": 0.2943,
+      "step": 7582
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010643523343888666,
+      "loss": 0.3205,
+      "step": 7583
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001064141287452381,
+      "loss": 0.3044,
+      "step": 7584
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010639302376471494,
+      "loss": 0.3211,
+      "step": 7585
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010637191849826112,
+      "loss": 0.3187,
+      "step": 7586
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010635081294682052,
+      "loss": 0.3032,
+      "step": 7587
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010632970711133718,
+      "loss": 0.3138,
+      "step": 7588
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010630860099275503,
+      "loss": 0.2961,
+      "step": 7589
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010628749459201805,
+      "loss": 0.2944,
+      "step": 7590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010626638791007022,
+      "loss": 0.3109,
+      "step": 7591
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010624528094785555,
+      "loss": 0.3075,
+      "step": 7592
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010622417370631808,
+      "loss": 0.3045,
+      "step": 7593
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010620306618640182,
+      "loss": 0.3383,
+      "step": 7594
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010618195838905081,
+      "loss": 0.313,
+      "step": 7595
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001061608503152091,
+      "loss": 0.2802,
+      "step": 7596
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001061397419658208,
+      "loss": 0.3014,
+      "step": 7597
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010611863334182994,
+      "loss": 0.3049,
+      "step": 7598
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010609752444418061,
+      "loss": 0.3031,
+      "step": 7599
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010607641527381698,
+      "loss": 0.3048,
+      "step": 7600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001060553058316831,
+      "loss": 0.3271,
+      "step": 7601
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010603419611872312,
+      "loss": 0.3141,
+      "step": 7602
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010601308613588119,
+      "loss": 0.2996,
+      "step": 7603
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010599197588410146,
+      "loss": 0.2957,
+      "step": 7604
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010597086536432808,
+      "loss": 0.3127,
+      "step": 7605
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010594975457750523,
+      "loss": 0.2946,
+      "step": 7606
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001059286435245771,
+      "loss": 0.3076,
+      "step": 7607
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001059075322064879,
+      "loss": 0.3136,
+      "step": 7608
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010588642062418186,
+      "loss": 0.2989,
+      "step": 7609
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010586530877860317,
+      "loss": 0.3044,
+      "step": 7610
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010584419667069607,
+      "loss": 0.3069,
+      "step": 7611
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010582308430140481,
+      "loss": 0.3045,
+      "step": 7612
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010580197167167369,
+      "loss": 0.3014,
+      "step": 7613
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010578085878244695,
+      "loss": 0.3066,
+      "step": 7614
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010575974563466884,
+      "loss": 0.2927,
+      "step": 7615
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010573863222928368,
+      "loss": 0.3223,
+      "step": 7616
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001057175185672358,
+      "loss": 0.3109,
+      "step": 7617
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010569640464946949,
+      "loss": 0.3019,
+      "step": 7618
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010567529047692908,
+      "loss": 0.2922,
+      "step": 7619
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010565417605055894,
+      "loss": 0.3096,
+      "step": 7620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010563306137130338,
+      "loss": 0.3013,
+      "step": 7621
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010561194644010679,
+      "loss": 0.3144,
+      "step": 7622
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010559083125791356,
+      "loss": 0.3066,
+      "step": 7623
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010556971582566802,
+      "loss": 0.2937,
+      "step": 7624
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001055486001443146,
+      "loss": 0.3175,
+      "step": 7625
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001055274842147977,
+      "loss": 0.3119,
+      "step": 7626
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010550636803806175,
+      "loss": 0.302,
+      "step": 7627
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010548525161505118,
+      "loss": 0.2983,
+      "step": 7628
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010546413494671044,
+      "loss": 0.3044,
+      "step": 7629
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010544301803398395,
+      "loss": 0.3161,
+      "step": 7630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010542190087781619,
+      "loss": 0.3063,
+      "step": 7631
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010540078347915163,
+      "loss": 0.2915,
+      "step": 7632
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010537966583893479,
+      "loss": 0.3088,
+      "step": 7633
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010535854795811011,
+      "loss": 0.307,
+      "step": 7634
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010533742983762212,
+      "loss": 0.2934,
+      "step": 7635
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010531631147841532,
+      "loss": 0.2827,
+      "step": 7636
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010529519288143428,
+      "loss": 0.3005,
+      "step": 7637
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010527407404762351,
+      "loss": 0.3078,
+      "step": 7638
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010525295497792753,
+      "loss": 0.3037,
+      "step": 7639
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010523183567329095,
+      "loss": 0.3054,
+      "step": 7640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010521071613465833,
+      "loss": 0.327,
+      "step": 7641
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010518959636297423,
+      "loss": 0.2941,
+      "step": 7642
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010516847635918328,
+      "loss": 0.2942,
+      "step": 7643
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010514735612423001,
+      "loss": 0.3211,
+      "step": 7644
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001051262356590591,
+      "loss": 0.2898,
+      "step": 7645
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010510511496461513,
+      "loss": 0.2947,
+      "step": 7646
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010508399404184275,
+      "loss": 0.2915,
+      "step": 7647
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010506287289168659,
+      "loss": 0.3033,
+      "step": 7648
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010504175151509133,
+      "loss": 0.2848,
+      "step": 7649
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010502062991300161,
+      "loss": 0.2884,
+      "step": 7650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001049995080863621,
+      "loss": 0.2955,
+      "step": 7651
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001049783860361175,
+      "loss": 0.297,
+      "step": 7652
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010495726376321249,
+      "loss": 0.2871,
+      "step": 7653
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010493614126859177,
+      "loss": 0.3009,
+      "step": 7654
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010491501855320005,
+      "loss": 0.2931,
+      "step": 7655
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010489389561798207,
+      "loss": 0.2973,
+      "step": 7656
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010487277246388254,
+      "loss": 0.3031,
+      "step": 7657
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010485164909184622,
+      "loss": 0.3004,
+      "step": 7658
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010483052550281784,
+      "loss": 0.2802,
+      "step": 7659
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001048094016977422,
+      "loss": 0.2835,
+      "step": 7660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010478827767756405,
+      "loss": 0.2983,
+      "step": 7661
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010476715344322814,
+      "loss": 0.2797,
+      "step": 7662
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010474602899567931,
+      "loss": 0.2989,
+      "step": 7663
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010472490433586233,
+      "loss": 0.2856,
+      "step": 7664
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000104703779464722,
+      "loss": 0.305,
+      "step": 7665
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010468265438320317,
+      "loss": 0.2998,
+      "step": 7666
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010466152909225069,
+      "loss": 0.2828,
+      "step": 7667
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001046404035928093,
+      "loss": 0.3067,
+      "step": 7668
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010461927788582395,
+      "loss": 0.2971,
+      "step": 7669
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010459815197223943,
+      "loss": 0.3006,
+      "step": 7670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010457702585300065,
+      "loss": 0.3085,
+      "step": 7671
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010455589952905245,
+      "loss": 0.2892,
+      "step": 7672
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010453477300133976,
+      "loss": 0.3092,
+      "step": 7673
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010451364627080743,
+      "loss": 0.3112,
+      "step": 7674
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010449251933840034,
+      "loss": 0.2898,
+      "step": 7675
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010447139220506345,
+      "loss": 0.2866,
+      "step": 7676
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010445026487174165,
+      "loss": 0.3134,
+      "step": 7677
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010442913733937991,
+      "loss": 0.3045,
+      "step": 7678
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010440800960892312,
+      "loss": 0.3016,
+      "step": 7679
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010438688168131625,
+      "loss": 0.2992,
+      "step": 7680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010436575355750426,
+      "loss": 0.3014,
+      "step": 7681
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010434462523843212,
+      "loss": 0.315,
+      "step": 7682
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010432349672504477,
+      "loss": 0.3068,
+      "step": 7683
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010430236801828724,
+      "loss": 0.2995,
+      "step": 7684
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010428123911910445,
+      "loss": 0.2866,
+      "step": 7685
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010426011002844144,
+      "loss": 0.3303,
+      "step": 7686
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010423898074724325,
+      "loss": 0.2874,
+      "step": 7687
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010421785127645485,
+      "loss": 0.3011,
+      "step": 7688
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010419672161702128,
+      "loss": 0.3017,
+      "step": 7689
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010417559176988757,
+      "loss": 0.2923,
+      "step": 7690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010415446173599877,
+      "loss": 0.3011,
+      "step": 7691
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010413333151629989,
+      "loss": 0.2971,
+      "step": 7692
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010411220111173608,
+      "loss": 0.3087,
+      "step": 7693
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010409107052325231,
+      "loss": 0.3105,
+      "step": 7694
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010406993975179369,
+      "loss": 0.29,
+      "step": 7695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001040488087983053,
+      "loss": 0.3043,
+      "step": 7696
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010402767766373224,
+      "loss": 0.2731,
+      "step": 7697
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010400654634901961,
+      "loss": 0.3022,
+      "step": 7698
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001039854148551125,
+      "loss": 0.294,
+      "step": 7699
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010396428318295603,
+      "loss": 0.2903,
+      "step": 7700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010394315133349535,
+      "loss": 0.3032,
+      "step": 7701
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010392201930767553,
+      "loss": 0.3055,
+      "step": 7702
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00010390088710644182,
+      "loss": 0.2914,
+      "step": 7703
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010387975473073924,
+      "loss": 0.2927,
+      "step": 7704
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010385862218151298,
+      "loss": 0.306,
+      "step": 7705
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010383748945970825,
+      "loss": 0.2944,
+      "step": 7706
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001038163565662702,
+      "loss": 0.2852,
+      "step": 7707
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010379522350214399,
+      "loss": 0.2958,
+      "step": 7708
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001037740902682748,
+      "loss": 0.2987,
+      "step": 7709
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010375295686560784,
+      "loss": 0.2912,
+      "step": 7710
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001037318232950883,
+      "loss": 0.2952,
+      "step": 7711
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010371068955766141,
+      "loss": 0.2928,
+      "step": 7712
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010368955565427238,
+      "loss": 0.2856,
+      "step": 7713
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010366842158586642,
+      "loss": 0.2951,
+      "step": 7714
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010364728735338876,
+      "loss": 0.2922,
+      "step": 7715
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010362615295778462,
+      "loss": 0.2867,
+      "step": 7716
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010360501839999928,
+      "loss": 0.2915,
+      "step": 7717
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010358388368097796,
+      "loss": 0.296,
+      "step": 7718
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010356274880166595,
+      "loss": 0.2801,
+      "step": 7719
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010354161376300848,
+      "loss": 0.3051,
+      "step": 7720
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010352047856595089,
+      "loss": 0.2806,
+      "step": 7721
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010349934321143837,
+      "loss": 0.297,
+      "step": 7722
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001034782077004163,
+      "loss": 0.3033,
+      "step": 7723
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001034570720338299,
+      "loss": 0.3053,
+      "step": 7724
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001034359362126245,
+      "loss": 0.2992,
+      "step": 7725
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010341480023774541,
+      "loss": 0.2916,
+      "step": 7726
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010339366411013794,
+      "loss": 0.2739,
+      "step": 7727
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010337252783074742,
+      "loss": 0.3051,
+      "step": 7728
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010335139140051917,
+      "loss": 0.2959,
+      "step": 7729
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010333025482039853,
+      "loss": 0.2975,
+      "step": 7730
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010330911809133086,
+      "loss": 0.3106,
+      "step": 7731
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010328798121426151,
+      "loss": 0.3011,
+      "step": 7732
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010326684419013578,
+      "loss": 0.3017,
+      "step": 7733
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00010324570701989908,
+      "loss": 0.2964,
+      "step": 7734
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010322456970449677,
+      "loss": 0.3005,
+      "step": 7735
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001032034322448742,
+      "loss": 0.2881,
+      "step": 7736
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010318229464197682,
+      "loss": 0.2853,
+      "step": 7737
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010316115689674997,
+      "loss": 0.295,
+      "step": 7738
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010314001901013904,
+      "loss": 0.3061,
+      "step": 7739
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010311888098308942,
+      "loss": 0.2751,
+      "step": 7740
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010309774281654653,
+      "loss": 0.3001,
+      "step": 7741
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010307660451145583,
+      "loss": 0.2887,
+      "step": 7742
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010305546606876267,
+      "loss": 0.2981,
+      "step": 7743
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010303432748941253,
+      "loss": 0.3287,
+      "step": 7744
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001030131887743508,
+      "loss": 0.2741,
+      "step": 7745
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010299204992452294,
+      "loss": 0.3073,
+      "step": 7746
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001029709109408744,
+      "loss": 0.2904,
+      "step": 7747
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001029497718243506,
+      "loss": 0.3017,
+      "step": 7748
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010292863257589699,
+      "loss": 0.3006,
+      "step": 7749
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010290749319645909,
+      "loss": 0.2912,
+      "step": 7750
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010288635368698233,
+      "loss": 0.2987,
+      "step": 7751
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010286521404841221,
+      "loss": 0.2961,
+      "step": 7752
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010284407428169418,
+      "loss": 0.2906,
+      "step": 7753
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010282293438777371,
+      "loss": 0.2731,
+      "step": 7754
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010280179436759631,
+      "loss": 0.3128,
+      "step": 7755
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010278065422210749,
+      "loss": 0.2991,
+      "step": 7756
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010275951395225274,
+      "loss": 0.2924,
+      "step": 7757
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010273837355897758,
+      "loss": 0.2963,
+      "step": 7758
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001027172330432275,
+      "loss": 0.2926,
+      "step": 7759
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010269609240594801,
+      "loss": 0.2925,
+      "step": 7760
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010267495164808469,
+      "loss": 0.2815,
+      "step": 7761
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010265381077058303,
+      "loss": 0.2871,
+      "step": 7762
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010263266977438857,
+      "loss": 0.2939,
+      "step": 7763
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00010261152866044684,
+      "loss": 0.3039,
+      "step": 7764
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010259038742970337,
+      "loss": 0.2841,
+      "step": 7765
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010256924608310374,
+      "loss": 0.2981,
+      "step": 7766
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001025481046215935,
+      "loss": 0.2868,
+      "step": 7767
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010252696304611822,
+      "loss": 0.2892,
+      "step": 7768
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010250582135762346,
+      "loss": 0.2879,
+      "step": 7769
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010248467955705477,
+      "loss": 0.2894,
+      "step": 7770
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010246353764535774,
+      "loss": 0.2876,
+      "step": 7771
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010244239562347797,
+      "loss": 0.3006,
+      "step": 7772
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010242125349236104,
+      "loss": 0.2876,
+      "step": 7773
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010240011125295252,
+      "loss": 0.2883,
+      "step": 7774
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000102378968906198,
+      "loss": 0.2662,
+      "step": 7775
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010235782645304311,
+      "loss": 0.2881,
+      "step": 7776
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010233668389443343,
+      "loss": 0.2921,
+      "step": 7777
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010231554123131459,
+      "loss": 0.2772,
+      "step": 7778
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010229439846463217,
+      "loss": 0.2791,
+      "step": 7779
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010227325559533185,
+      "loss": 0.2755,
+      "step": 7780
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010225211262435919,
+      "loss": 0.2817,
+      "step": 7781
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010223096955265985,
+      "loss": 0.2949,
+      "step": 7782
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001022098263811795,
+      "loss": 0.299,
+      "step": 7783
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001021886831108637,
+      "loss": 0.2935,
+      "step": 7784
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010216753974265813,
+      "loss": 0.2979,
+      "step": 7785
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010214639627750844,
+      "loss": 0.3056,
+      "step": 7786
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010212525271636026,
+      "loss": 0.297,
+      "step": 7787
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010210410906015926,
+      "loss": 0.2818,
+      "step": 7788
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010208296530985111,
+      "loss": 0.3024,
+      "step": 7789
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010206182146638146,
+      "loss": 0.2938,
+      "step": 7790
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010204067753069595,
+      "loss": 0.2793,
+      "step": 7791
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010201953350374032,
+      "loss": 0.282,
+      "step": 7792
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001019983893864602,
+      "loss": 0.2793,
+      "step": 7793
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00010197724517980126,
+      "loss": 0.2798,
+      "step": 7794
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001019561008847092,
+      "loss": 0.2896,
+      "step": 7795
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001019349565021297,
+      "loss": 0.2893,
+      "step": 7796
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010191381203300847,
+      "loss": 0.2816,
+      "step": 7797
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010189266747829118,
+      "loss": 0.2983,
+      "step": 7798
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010187152283892354,
+      "loss": 0.302,
+      "step": 7799
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010185037811585127,
+      "loss": 0.2825,
+      "step": 7800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010182923331002006,
+      "loss": 0.2879,
+      "step": 7801
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001018080884223756,
+      "loss": 0.3036,
+      "step": 7802
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010178694345386366,
+      "loss": 0.293,
+      "step": 7803
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001017657984054299,
+      "loss": 0.2888,
+      "step": 7804
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010174465327802005,
+      "loss": 0.2943,
+      "step": 7805
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010172350807257986,
+      "loss": 0.2841,
+      "step": 7806
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010170236279005507,
+      "loss": 0.3022,
+      "step": 7807
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010168121743139136,
+      "loss": 0.2816,
+      "step": 7808
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001016600719975345,
+      "loss": 0.2898,
+      "step": 7809
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010163892648943025,
+      "loss": 0.2949,
+      "step": 7810
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001016177809080243,
+      "loss": 0.2959,
+      "step": 7811
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010159663525426242,
+      "loss": 0.2874,
+      "step": 7812
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010157548952909035,
+      "loss": 0.2711,
+      "step": 7813
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010155434373345385,
+      "loss": 0.3037,
+      "step": 7814
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010153319786829867,
+      "loss": 0.2889,
+      "step": 7815
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010151205193457059,
+      "loss": 0.2842,
+      "step": 7816
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010149090593321535,
+      "loss": 0.3012,
+      "step": 7817
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001014697598651787,
+      "loss": 0.2794,
+      "step": 7818
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010144861373140642,
+      "loss": 0.2793,
+      "step": 7819
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010142746753284428,
+      "loss": 0.2987,
+      "step": 7820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010140632127043805,
+      "loss": 0.2928,
+      "step": 7821
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00010138517494513353,
+      "loss": 0.281,
+      "step": 7822
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001013640285578765,
+      "loss": 0.2807,
+      "step": 7823
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001013428821096127,
+      "loss": 0.2866,
+      "step": 7824
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001013217356012879,
+      "loss": 0.2817,
+      "step": 7825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010130058903384794,
+      "loss": 0.2888,
+      "step": 7826
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010127944240823859,
+      "loss": 0.2887,
+      "step": 7827
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010125829572540564,
+      "loss": 0.2762,
+      "step": 7828
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010123714898629488,
+      "loss": 0.2997,
+      "step": 7829
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010121600219185212,
+      "loss": 0.2879,
+      "step": 7830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010119485534302316,
+      "loss": 0.3041,
+      "step": 7831
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010117370844075378,
+      "loss": 0.29,
+      "step": 7832
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010115256148598982,
+      "loss": 0.2882,
+      "step": 7833
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010113141447967703,
+      "loss": 0.2904,
+      "step": 7834
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010111026742276126,
+      "loss": 0.2739,
+      "step": 7835
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010108912031618831,
+      "loss": 0.2751,
+      "step": 7836
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010106797316090399,
+      "loss": 0.2837,
+      "step": 7837
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010104682595785413,
+      "loss": 0.2926,
+      "step": 7838
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010102567870798454,
+      "loss": 0.2743,
+      "step": 7839
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010100453141224105,
+      "loss": 0.2711,
+      "step": 7840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010098338407156946,
+      "loss": 0.2926,
+      "step": 7841
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001009622366869156,
+      "loss": 0.3042,
+      "step": 7842
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010094108925922532,
+      "loss": 0.3041,
+      "step": 7843
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010091994178944442,
+      "loss": 0.2842,
+      "step": 7844
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010089879427851871,
+      "loss": 0.279,
+      "step": 7845
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010087764672739408,
+      "loss": 0.2822,
+      "step": 7846
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010085649913701632,
+      "loss": 0.2835,
+      "step": 7847
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010083535150833125,
+      "loss": 0.2907,
+      "step": 7848
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010081420384228477,
+      "loss": 0.2907,
+      "step": 7849
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010079305613982266,
+      "loss": 0.2868,
+      "step": 7850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010077190840189079,
+      "loss": 0.285,
+      "step": 7851
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010075076062943498,
+      "loss": 0.2875,
+      "step": 7852
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010072961282340112,
+      "loss": 0.2792,
+      "step": 7853
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010070846498473499,
+      "loss": 0.2718,
+      "step": 7854
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010068731711438246,
+      "loss": 0.2764,
+      "step": 7855
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00010066616921328939,
+      "loss": 0.2898,
+      "step": 7856
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010064502128240163,
+      "loss": 0.3004,
+      "step": 7857
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010062387332266501,
+      "loss": 0.2881,
+      "step": 7858
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001006027253350254,
+      "loss": 0.2895,
+      "step": 7859
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010058157732042865,
+      "loss": 0.2646,
+      "step": 7860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001005604292798206,
+      "loss": 0.2973,
+      "step": 7861
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010053928121414712,
+      "loss": 0.3024,
+      "step": 7862
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010051813312435407,
+      "loss": 0.2905,
+      "step": 7863
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010049698501138731,
+      "loss": 0.2848,
+      "step": 7864
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010047583687619267,
+      "loss": 0.2894,
+      "step": 7865
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010045468871971602,
+      "loss": 0.2838,
+      "step": 7866
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010043354054290321,
+      "loss": 0.297,
+      "step": 7867
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010041239234670015,
+      "loss": 0.279,
+      "step": 7868
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010039124413205264,
+      "loss": 0.2909,
+      "step": 7869
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010037009589990658,
+      "loss": 0.286,
+      "step": 7870
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010034894765120784,
+      "loss": 0.2727,
+      "step": 7871
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010032779938690224,
+      "loss": 0.2808,
+      "step": 7872
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010030665110793574,
+      "loss": 0.2736,
+      "step": 7873
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010028550281525408,
+      "loss": 0.2991,
+      "step": 7874
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001002643545098032,
+      "loss": 0.2855,
+      "step": 7875
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010024320619252896,
+      "loss": 0.273,
+      "step": 7876
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010022205786437722,
+      "loss": 0.2994,
+      "step": 7877
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010020090952629383,
+      "loss": 0.2928,
+      "step": 7878
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010017976117922469,
+      "loss": 0.3015,
+      "step": 7879
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010015861282411567,
+      "loss": 0.2853,
+      "step": 7880
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010013746446191261,
+      "loss": 0.2804,
+      "step": 7881
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001001163160935614,
+      "loss": 0.2684,
+      "step": 7882
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010009516772000791,
+      "loss": 0.2941,
+      "step": 7883
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000100074019342198,
+      "loss": 0.2784,
+      "step": 7884
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010005287096107752,
+      "loss": 0.279,
+      "step": 7885
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001000317225775924,
+      "loss": 0.2682,
+      "step": 7886
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00010001057419268845,
+      "loss": 0.2685,
+      "step": 7887
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.998942580731158e-05,
+      "loss": 0.2934,
+      "step": 7888
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.996827742240764e-05,
+      "loss": 0.2982,
+      "step": 7889
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.994712903892249e-05,
+      "loss": 0.2734,
+      "step": 7890
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.992598065780204e-05,
+      "loss": 0.2748,
+      "step": 7891
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.990483227999211e-05,
+      "loss": 0.2862,
+      "step": 7892
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.988368390643862e-05,
+      "loss": 0.2875,
+      "step": 7893
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.986253553808741e-05,
+      "loss": 0.2846,
+      "step": 7894
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.984138717588437e-05,
+      "loss": 0.2757,
+      "step": 7895
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.982023882077532e-05,
+      "loss": 0.2912,
+      "step": 7896
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.979909047370619e-05,
+      "loss": 0.2895,
+      "step": 7897
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.977794213562282e-05,
+      "loss": 0.287,
+      "step": 7898
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.975679380747106e-05,
+      "loss": 0.2871,
+      "step": 7899
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.973564549019682e-05,
+      "loss": 0.2921,
+      "step": 7900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.971449718474593e-05,
+      "loss": 0.2825,
+      "step": 7901
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.96933488920643e-05,
+      "loss": 0.295,
+      "step": 7902
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.967220061309774e-05,
+      "loss": 0.2764,
+      "step": 7903
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.965105234879216e-05,
+      "loss": 0.2974,
+      "step": 7904
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.962990410009342e-05,
+      "loss": 0.2868,
+      "step": 7905
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.960875586794736e-05,
+      "loss": 0.2847,
+      "step": 7906
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.958760765329986e-05,
+      "loss": 0.2677,
+      "step": 7907
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.956645945709683e-05,
+      "loss": 0.2946,
+      "step": 7908
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.954531128028404e-05,
+      "loss": 0.2867,
+      "step": 7909
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95241631238074e-05,
+      "loss": 0.286,
+      "step": 7910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.950301498861274e-05,
+      "loss": 0.2786,
+      "step": 7911
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.948186687564595e-05,
+      "loss": 0.2929,
+      "step": 7912
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.94607187858529e-05,
+      "loss": 0.2787,
+      "step": 7913
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943957072017942e-05,
+      "loss": 0.276,
+      "step": 7914
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.941842267957137e-05,
+      "loss": 0.2828,
+      "step": 7915
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.939727466497461e-05,
+      "loss": 0.2981,
+      "step": 7916
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.9376126677335e-05,
+      "loss": 0.2919,
+      "step": 7917
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.935497871759839e-05,
+      "loss": 0.2862,
+      "step": 7918
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.933383078671062e-05,
+      "loss": 0.2907,
+      "step": 7919
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.931268288561756e-05,
+      "loss": 0.2937,
+      "step": 7920
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.929153501526504e-05,
+      "loss": 0.2785,
+      "step": 7921
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.927038717659892e-05,
+      "loss": 0.2926,
+      "step": 7922
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.924923937056502e-05,
+      "loss": 0.2802,
+      "step": 7923
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.922809159810922e-05,
+      "loss": 0.3098,
+      "step": 7924
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.920694386017735e-05,
+      "loss": 0.2776,
+      "step": 7925
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.918579615771524e-05,
+      "loss": 0.2905,
+      "step": 7926
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.916464849166875e-05,
+      "loss": 0.2961,
+      "step": 7927
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.914350086298375e-05,
+      "loss": 0.2846,
+      "step": 7928
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.912235327260597e-05,
+      "loss": 0.282,
+      "step": 7929
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.910120572148133e-05,
+      "loss": 0.2868,
+      "step": 7930
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.908005821055563e-05,
+      "loss": 0.2703,
+      "step": 7931
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.905891074077472e-05,
+      "loss": 0.3053,
+      "step": 7932
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.903776331308442e-05,
+      "loss": 0.2872,
+      "step": 7933
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.901661592843056e-05,
+      "loss": 0.2824,
+      "step": 7934
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.899546858775898e-05,
+      "loss": 0.2831,
+      "step": 7935
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.897432129201547e-05,
+      "loss": 0.2737,
+      "step": 7936
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.895317404214587e-05,
+      "loss": 0.298,
+      "step": 7937
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.893202683909603e-05,
+      "loss": 0.2895,
+      "step": 7938
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.89108796838117e-05,
+      "loss": 0.2937,
+      "step": 7939
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.888973257723877e-05,
+      "loss": 0.2983,
+      "step": 7940
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.8868585520323e-05,
+      "loss": 0.2758,
+      "step": 7941
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.884743851401022e-05,
+      "loss": 0.2736,
+      "step": 7942
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.882629155924623e-05,
+      "loss": 0.2772,
+      "step": 7943
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.880514465697684e-05,
+      "loss": 0.2987,
+      "step": 7944
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.878399780814788e-05,
+      "loss": 0.2708,
+      "step": 7945
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.876285101370512e-05,
+      "loss": 0.2738,
+      "step": 7946
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.874170427459438e-05,
+      "loss": 0.2883,
+      "step": 7947
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.872055759176145e-05,
+      "loss": 0.2748,
+      "step": 7948
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.86994109661521e-05,
+      "loss": 0.2889,
+      "step": 7949
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.867826439871215e-05,
+      "loss": 0.2884,
+      "step": 7950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.865711789038737e-05,
+      "loss": 0.2884,
+      "step": 7951
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.863597144212355e-05,
+      "loss": 0.2811,
+      "step": 7952
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.861482505486648e-05,
+      "loss": 0.2858,
+      "step": 7953
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.859367872956196e-05,
+      "loss": 0.2827,
+      "step": 7954
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.857253246715574e-05,
+      "loss": 0.2914,
+      "step": 7955
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.855138626859361e-05,
+      "loss": 0.2836,
+      "step": 7956
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.853024013482132e-05,
+      "loss": 0.2816,
+      "step": 7957
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.850909406678468e-05,
+      "loss": 0.275,
+      "step": 7958
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.848794806542942e-05,
+      "loss": 0.2634,
+      "step": 7959
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.846680213170135e-05,
+      "loss": 0.2829,
+      "step": 7960
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.844565626654618e-05,
+      "loss": 0.2765,
+      "step": 7961
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.842451047090967e-05,
+      "loss": 0.3022,
+      "step": 7962
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.840336474573761e-05,
+      "loss": 0.2915,
+      "step": 7963
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.838221909197573e-05,
+      "loss": 0.285,
+      "step": 7964
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.836107351056979e-05,
+      "loss": 0.2801,
+      "step": 7965
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.83399280024655e-05,
+      "loss": 0.2725,
+      "step": 7966
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.831878256860863e-05,
+      "loss": 0.2849,
+      "step": 7967
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.829763720994496e-05,
+      "loss": 0.2714,
+      "step": 7968
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.827649192742015e-05,
+      "loss": 0.2823,
+      "step": 7969
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.825534672197996e-05,
+      "loss": 0.2686,
+      "step": 7970
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.823420159457013e-05,
+      "loss": 0.2795,
+      "step": 7971
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.821305654613639e-05,
+      "loss": 0.2989,
+      "step": 7972
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.819191157762442e-05,
+      "loss": 0.2823,
+      "step": 7973
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.817076668997998e-05,
+      "loss": 0.2752,
+      "step": 7974
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.814962188414876e-05,
+      "loss": 0.2839,
+      "step": 7975
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.812847716107647e-05,
+      "loss": 0.2728,
+      "step": 7976
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.810733252170885e-05,
+      "loss": 0.2672,
+      "step": 7977
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.808618796699155e-05,
+      "loss": 0.2955,
+      "step": 7978
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.806504349787031e-05,
+      "loss": 0.2912,
+      "step": 7979
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.804389911529082e-05,
+      "loss": 0.2687,
+      "step": 7980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.802275482019875e-05,
+      "loss": 0.2583,
+      "step": 7981
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.800161061353982e-05,
+      "loss": 0.2738,
+      "step": 7982
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.798046649625969e-05,
+      "loss": 0.2705,
+      "step": 7983
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.795932246930404e-05,
+      "loss": 0.2806,
+      "step": 7984
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.793817853361855e-05,
+      "loss": 0.2808,
+      "step": 7985
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.79170346901489e-05,
+      "loss": 0.2945,
+      "step": 7986
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.789589093984074e-05,
+      "loss": 0.2767,
+      "step": 7987
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.787474728363979e-05,
+      "loss": 0.2723,
+      "step": 7988
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.785360372249162e-05,
+      "loss": 0.2869,
+      "step": 7989
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.783246025734192e-05,
+      "loss": 0.2837,
+      "step": 7990
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.781131688913635e-05,
+      "loss": 0.2835,
+      "step": 7991
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.779017361882054e-05,
+      "loss": 0.2978,
+      "step": 7992
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.776903044734016e-05,
+      "loss": 0.2844,
+      "step": 7993
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.774788737564082e-05,
+      "loss": 0.2803,
+      "step": 7994
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.772674440466818e-05,
+      "loss": 0.3003,
+      "step": 7995
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.770560153536784e-05,
+      "loss": 0.2749,
+      "step": 7996
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.768445876868544e-05,
+      "loss": 0.2752,
+      "step": 7997
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.76633161055666e-05,
+      "loss": 0.2814,
+      "step": 7998
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.764217354695691e-05,
+      "loss": 0.2759,
+      "step": 7999
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.762103109380202e-05,
+      "loss": 0.2722,
+      "step": 8000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.759988874704751e-05,
+      "loss": 0.2883,
+      "step": 8001
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.757874650763897e-05,
+      "loss": 0.2589,
+      "step": 8002
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.755760437652203e-05,
+      "loss": 0.2677,
+      "step": 8003
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.753646235464224e-05,
+      "loss": 0.2815,
+      "step": 8004
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.751532044294523e-05,
+      "loss": 0.2817,
+      "step": 8005
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.749417864237654e-05,
+      "loss": 0.2837,
+      "step": 8006
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.747303695388177e-05,
+      "loss": 0.277,
+      "step": 8007
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.745189537840654e-05,
+      "loss": 0.2761,
+      "step": 8008
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.74307539168963e-05,
+      "loss": 0.271,
+      "step": 8009
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.740961257029668e-05,
+      "loss": 0.2907,
+      "step": 8010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.738847133955323e-05,
+      "loss": 0.2733,
+      "step": 8011
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.736733022561148e-05,
+      "loss": 0.2719,
+      "step": 8012
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.7346189229417e-05,
+      "loss": 0.2701,
+      "step": 8013
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.732504835191533e-05,
+      "loss": 0.2809,
+      "step": 8014
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.7303907594052e-05,
+      "loss": 0.2778,
+      "step": 8015
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.728276695677251e-05,
+      "loss": 0.2913,
+      "step": 8016
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.726162644102244e-05,
+      "loss": 0.2714,
+      "step": 8017
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.724048604774727e-05,
+      "loss": 0.2825,
+      "step": 8018
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.721934577789252e-05,
+      "loss": 0.2748,
+      "step": 8019
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.719820563240371e-05,
+      "loss": 0.2729,
+      "step": 8020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.717706561222631e-05,
+      "loss": 0.2701,
+      "step": 8021
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.715592571830584e-05,
+      "loss": 0.2584,
+      "step": 8022
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.71347859515878e-05,
+      "loss": 0.2691,
+      "step": 8023
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.711364631301766e-05,
+      "loss": 0.2812,
+      "step": 8024
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.709250680354093e-05,
+      "loss": 0.2769,
+      "step": 8025
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.707136742410302e-05,
+      "loss": 0.2911,
+      "step": 8026
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.705022817564944e-05,
+      "loss": 0.3006,
+      "step": 8027
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.702908905912567e-05,
+      "loss": 0.2923,
+      "step": 8028
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.70079500754771e-05,
+      "loss": 0.2843,
+      "step": 8029
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.698681122564925e-05,
+      "loss": 0.2708,
+      "step": 8030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.696567251058752e-05,
+      "loss": 0.2748,
+      "step": 8031
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.694453393123736e-05,
+      "loss": 0.2888,
+      "step": 8032
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.69233954885442e-05,
+      "loss": 0.2855,
+      "step": 8033
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.690225718345348e-05,
+      "loss": 0.2724,
+      "step": 8034
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.688111901691061e-05,
+      "loss": 0.2707,
+      "step": 8035
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.6859980989861e-05,
+      "loss": 0.2878,
+      "step": 8036
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.683884310325005e-05,
+      "loss": 0.2739,
+      "step": 8037
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.681770535802322e-05,
+      "loss": 0.2679,
+      "step": 8038
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.679656775512582e-05,
+      "loss": 0.2688,
+      "step": 8039
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.677543029550326e-05,
+      "loss": 0.2842,
+      "step": 8040
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.675429298010094e-05,
+      "loss": 0.2784,
+      "step": 8041
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.673315580986423e-05,
+      "loss": 0.2834,
+      "step": 8042
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.671201878573853e-05,
+      "loss": 0.272,
+      "step": 8043
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.669088190866914e-05,
+      "loss": 0.28,
+      "step": 8044
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.666974517960146e-05,
+      "loss": 0.2877,
+      "step": 8045
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.664860859948083e-05,
+      "loss": 0.2841,
+      "step": 8046
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.662747216925257e-05,
+      "loss": 0.2603,
+      "step": 8047
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.660633588986207e-05,
+      "loss": 0.2759,
+      "step": 8048
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.658519976225461e-05,
+      "loss": 0.2693,
+      "step": 8049
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.656406378737555e-05,
+      "loss": 0.2947,
+      "step": 8050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.654292796617014e-05,
+      "loss": 0.2831,
+      "step": 8051
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.652179229958375e-05,
+      "loss": 0.2804,
+      "step": 8052
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.650065678856165e-05,
+      "loss": 0.2819,
+      "step": 8053
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.647952143404915e-05,
+      "loss": 0.2674,
+      "step": 8054
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.645838623699153e-05,
+      "loss": 0.2644,
+      "step": 8055
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.643725119833408e-05,
+      "loss": 0.2868,
+      "step": 8056
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.641611631902206e-05,
+      "loss": 0.2692,
+      "step": 8057
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.639498160000076e-05,
+      "loss": 0.2767,
+      "step": 8058
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63738470422154e-05,
+      "loss": 0.285,
+      "step": 8059
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.635271264661128e-05,
+      "loss": 0.2755,
+      "step": 8060
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63315784141336e-05,
+      "loss": 0.2755,
+      "step": 8061
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.631044434572763e-05,
+      "loss": 0.2859,
+      "step": 8062
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.628931044233857e-05,
+      "loss": 0.2823,
+      "step": 8063
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.626817670491168e-05,
+      "loss": 0.2856,
+      "step": 8064
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.624704313439215e-05,
+      "loss": 0.2852,
+      "step": 8065
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.62259097317252e-05,
+      "loss": 0.2683,
+      "step": 8066
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.620477649785602e-05,
+      "loss": 0.2909,
+      "step": 8067
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.618364343372984e-05,
+      "loss": 0.2729,
+      "step": 8068
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.616251054029178e-05,
+      "loss": 0.281,
+      "step": 8069
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.614137781848704e-05,
+      "loss": 0.2877,
+      "step": 8070
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.612024526926081e-05,
+      "loss": 0.2833,
+      "step": 8071
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.609911289355825e-05,
+      "loss": 0.2821,
+      "step": 8072
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.607798069232448e-05,
+      "loss": 0.2706,
+      "step": 8073
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.605684866650468e-05,
+      "loss": 0.2874,
+      "step": 8074
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.603571681704398e-05,
+      "loss": 0.2834,
+      "step": 8075
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.601458514488752e-05,
+      "loss": 0.2829,
+      "step": 8076
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.599345365098042e-05,
+      "loss": 0.285,
+      "step": 8077
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.597232233626778e-05,
+      "loss": 0.2853,
+      "step": 8078
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.595119120169473e-05,
+      "loss": 0.2809,
+      "step": 8079
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.593006024820634e-05,
+      "loss": 0.2741,
+      "step": 8080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.590892947674772e-05,
+      "loss": 0.2833,
+      "step": 8081
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.588779888826394e-05,
+      "loss": 0.281,
+      "step": 8082
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.58666684837001e-05,
+      "loss": 0.2826,
+      "step": 8083
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.584553826400124e-05,
+      "loss": 0.2792,
+      "step": 8084
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.582440823011243e-05,
+      "loss": 0.2602,
+      "step": 8085
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.580327838297873e-05,
+      "loss": 0.2692,
+      "step": 8086
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.578214872354516e-05,
+      "loss": 0.2606,
+      "step": 8087
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.576101925275679e-05,
+      "loss": 0.2791,
+      "step": 8088
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.573988997155858e-05,
+      "loss": 0.2806,
+      "step": 8089
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57187608808956e-05,
+      "loss": 0.2726,
+      "step": 8090
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.569763198171283e-05,
+      "loss": 0.2874,
+      "step": 8091
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.567650327495526e-05,
+      "loss": 0.2802,
+      "step": 8092
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.565537476156792e-05,
+      "loss": 0.275,
+      "step": 8093
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.563424644249575e-05,
+      "loss": 0.2758,
+      "step": 8094
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.561311831868376e-05,
+      "loss": 0.2925,
+      "step": 8095
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55919903910769e-05,
+      "loss": 0.2827,
+      "step": 8096
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.557086266062012e-05,
+      "loss": 0.2836,
+      "step": 8097
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.554973512825837e-05,
+      "loss": 0.2935,
+      "step": 8098
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.552860779493657e-05,
+      "loss": 0.2835,
+      "step": 8099
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55074806615997e-05,
+      "loss": 0.2898,
+      "step": 8100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.548635372919261e-05,
+      "loss": 0.2892,
+      "step": 8101
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.546522699866026e-05,
+      "loss": 0.2733,
+      "step": 8102
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.544410047094758e-05,
+      "loss": 0.2798,
+      "step": 8103
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.542297414699936e-05,
+      "loss": 0.2848,
+      "step": 8104
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.540184802776058e-05,
+      "loss": 0.291,
+      "step": 8105
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.538072211417607e-05,
+      "loss": 0.2814,
+      "step": 8106
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.53595964071907e-05,
+      "loss": 0.2889,
+      "step": 8107
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.533847090774936e-05,
+      "loss": 0.2609,
+      "step": 8108
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.531734561679686e-05,
+      "loss": 0.2794,
+      "step": 8109
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.529622053527802e-05,
+      "loss": 0.2774,
+      "step": 8110
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.52750956641377e-05,
+      "loss": 0.278,
+      "step": 8111
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.525397100432072e-05,
+      "loss": 0.2818,
+      "step": 8112
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.523284655677188e-05,
+      "loss": 0.2827,
+      "step": 8113
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.521172232243598e-05,
+      "loss": 0.2685,
+      "step": 8114
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.519059830225782e-05,
+      "loss": 0.2998,
+      "step": 8115
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.516947449718218e-05,
+      "loss": 0.2862,
+      "step": 8116
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.514835090815382e-05,
+      "loss": 0.2713,
+      "step": 8117
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.51272275361175e-05,
+      "loss": 0.2808,
+      "step": 8118
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.510610438201796e-05,
+      "loss": 0.265,
+      "step": 8119
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.508498144679997e-05,
+      "loss": 0.2828,
+      "step": 8120
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.506385873140826e-05,
+      "loss": 0.2694,
+      "step": 8121
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.504273623678753e-05,
+      "loss": 0.2879,
+      "step": 8122
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.502161396388251e-05,
+      "loss": 0.2739,
+      "step": 8123
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.500049191363791e-05,
+      "loss": 0.2867,
+      "step": 8124
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.497937008699839e-05,
+      "loss": 0.2851,
+      "step": 8125
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.495824848490867e-05,
+      "loss": 0.2911,
+      "step": 8126
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.493712710831339e-05,
+      "loss": 0.2719,
+      "step": 8127
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.491600595815729e-05,
+      "loss": 0.2648,
+      "step": 8128
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.489488503538491e-05,
+      "loss": 0.2789,
+      "step": 8129
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.487376434094095e-05,
+      "loss": 0.2808,
+      "step": 8130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.485264387577002e-05,
+      "loss": 0.2693,
+      "step": 8131
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.483152364081677e-05,
+      "loss": 0.2926,
+      "step": 8132
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.481040363702579e-05,
+      "loss": 0.2767,
+      "step": 8133
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.478928386534169e-05,
+      "loss": 0.2772,
+      "step": 8134
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.476816432670906e-05,
+      "loss": 0.2721,
+      "step": 8135
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.474704502207247e-05,
+      "loss": 0.2844,
+      "step": 8136
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.472592595237653e-05,
+      "loss": 0.2831,
+      "step": 8137
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.470480711856574e-05,
+      "loss": 0.2682,
+      "step": 8138
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468368852158469e-05,
+      "loss": 0.2742,
+      "step": 8139
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.46625701623779e-05,
+      "loss": 0.2829,
+      "step": 8140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.464145204188991e-05,
+      "loss": 0.2868,
+      "step": 8141
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.462033416106523e-05,
+      "loss": 0.2749,
+      "step": 8142
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.459921652084836e-05,
+      "loss": 0.2606,
+      "step": 8143
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.457809912218381e-05,
+      "loss": 0.2877,
+      "step": 8144
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.455698196601606e-05,
+      "loss": 0.2685,
+      "step": 8145
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.453586505328957e-05,
+      "loss": 0.2829,
+      "step": 8146
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.45147483849488e-05,
+      "loss": 0.27,
+      "step": 8147
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.449363196193828e-05,
+      "loss": 0.2707,
+      "step": 8148
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.447251578520234e-05,
+      "loss": 0.2889,
+      "step": 8149
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.445139985568544e-05,
+      "loss": 0.2828,
+      "step": 8150
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.443028417433203e-05,
+      "loss": 0.2743,
+      "step": 8151
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.440916874208649e-05,
+      "loss": 0.2829,
+      "step": 8152
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.438805355989323e-05,
+      "loss": 0.2747,
+      "step": 8153
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.436693862869664e-05,
+      "loss": 0.2814,
+      "step": 8154
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.43458239494411e-05,
+      "loss": 0.2876,
+      "step": 8155
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.432470952307093e-05,
+      "loss": 0.2691,
+      "step": 8156
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.430359535053052e-05,
+      "loss": 0.2676,
+      "step": 8157
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.428248143276422e-05,
+      "loss": 0.2768,
+      "step": 8158
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.426136777071633e-05,
+      "loss": 0.2868,
+      "step": 8159
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.42402543653312e-05,
+      "loss": 0.2771,
+      "step": 8160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.421914121755309e-05,
+      "loss": 0.2604,
+      "step": 8161
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.419802832832632e-05,
+      "loss": 0.2879,
+      "step": 8162
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.417691569859518e-05,
+      "loss": 0.2858,
+      "step": 8163
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.415580332930394e-05,
+      "loss": 0.3005,
+      "step": 8164
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.413469122139684e-05,
+      "loss": 0.2709,
+      "step": 8165
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.411357937581814e-05,
+      "loss": 0.2899,
+      "step": 8166
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.409246779351209e-05,
+      "loss": 0.2913,
+      "step": 8167
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.407135647542294e-05,
+      "loss": 0.2695,
+      "step": 8168
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.405024542249481e-05,
+      "loss": 0.2782,
+      "step": 8169
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.402913463567197e-05,
+      "loss": 0.2668,
+      "step": 8170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.400802411589858e-05,
+      "loss": 0.2732,
+      "step": 8171
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.398691386411884e-05,
+      "loss": 0.2817,
+      "step": 8172
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.39658038812769e-05,
+      "loss": 0.2595,
+      "step": 8173
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.394469416831692e-05,
+      "loss": 0.2841,
+      "step": 8174
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.392358472618304e-05,
+      "loss": 0.2733,
+      "step": 8175
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.39024755558194e-05,
+      "loss": 0.2862,
+      "step": 8176
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.388136665817008e-05,
+      "loss": 0.2791,
+      "step": 8177
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.386025803417923e-05,
+      "loss": 0.2824,
+      "step": 8178
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.383914968479091e-05,
+      "loss": 0.2643,
+      "step": 8179
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.381804161094921e-05,
+      "loss": 0.2925,
+      "step": 8180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.379693381359822e-05,
+      "loss": 0.2897,
+      "step": 8181
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.377582629368196e-05,
+      "loss": 0.2857,
+      "step": 8182
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.375471905214447e-05,
+      "loss": 0.2955,
+      "step": 8183
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37336120899298e-05,
+      "loss": 0.3016,
+      "step": 8184
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.371250540798197e-05,
+      "loss": 0.2773,
+      "step": 8185
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.369139900724498e-05,
+      "loss": 0.302,
+      "step": 8186
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.367029288866282e-05,
+      "loss": 0.275,
+      "step": 8187
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.364918705317949e-05,
+      "loss": 0.2842,
+      "step": 8188
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.362808150173892e-05,
+      "loss": 0.2834,
+      "step": 8189
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.360697623528509e-05,
+      "loss": 0.2581,
+      "step": 8190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.358587125476192e-05,
+      "loss": 0.2725,
+      "step": 8191
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.356476656111336e-05,
+      "loss": 0.2796,
+      "step": 8192
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.354366215528332e-05,
+      "loss": 0.2758,
+      "step": 8193
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.352255803821574e-05,
+      "loss": 0.2691,
+      "step": 8194
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.350145421085443e-05,
+      "loss": 0.2648,
+      "step": 8195
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.348035067414333e-05,
+      "loss": 0.2778,
+      "step": 8196
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.345924742902628e-05,
+      "loss": 0.2639,
+      "step": 8197
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.343814447644715e-05,
+      "loss": 0.2847,
+      "step": 8198
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.341704181734974e-05,
+      "loss": 0.2563,
+      "step": 8199
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.339593945267793e-05,
+      "loss": 0.2594,
+      "step": 8200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.33748373833755e-05,
+      "loss": 0.2565,
+      "step": 8201
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.335373561038624e-05,
+      "loss": 0.2715,
+      "step": 8202
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.333263413465396e-05,
+      "loss": 0.2702,
+      "step": 8203
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.331153295712242e-05,
+      "loss": 0.2677,
+      "step": 8204
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.329043207873538e-05,
+      "loss": 0.2858,
+      "step": 8205
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.326933150043658e-05,
+      "loss": 0.2669,
+      "step": 8206
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.324823122316976e-05,
+      "loss": 0.2642,
+      "step": 8207
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.322713124787869e-05,
+      "loss": 0.2615,
+      "step": 8208
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.320603157550697e-05,
+      "loss": 0.2704,
+      "step": 8209
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.318493220699834e-05,
+      "loss": 0.2589,
+      "step": 8210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.316383314329649e-05,
+      "loss": 0.2641,
+      "step": 8211
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.314273438534508e-05,
+      "loss": 0.2791,
+      "step": 8212
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.312163593408774e-05,
+      "loss": 0.2681,
+      "step": 8213
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.310053779046815e-05,
+      "loss": 0.2864,
+      "step": 8214
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.30794399554299e-05,
+      "loss": 0.2656,
+      "step": 8215
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.305834242991659e-05,
+      "loss": 0.2893,
+      "step": 8216
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.303724521487184e-05,
+      "loss": 0.2691,
+      "step": 8217
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.301614831123923e-05,
+      "loss": 0.2899,
+      "step": 8218
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.299505171996232e-05,
+      "loss": 0.2721,
+      "step": 8219
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.297395544198468e-05,
+      "loss": 0.2715,
+      "step": 8220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.295285947824982e-05,
+      "loss": 0.2751,
+      "step": 8221
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.29317638297013e-05,
+      "loss": 0.2686,
+      "step": 8222
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.29106684972826e-05,
+      "loss": 0.2683,
+      "step": 8223
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.288957348193723e-05,
+      "loss": 0.2624,
+      "step": 8224
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.286847878460869e-05,
+      "loss": 0.273,
+      "step": 8225
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.284738440624041e-05,
+      "loss": 0.272,
+      "step": 8226
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.282629034777587e-05,
+      "loss": 0.2783,
+      "step": 8227
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.280519661015859e-05,
+      "loss": 0.2637,
+      "step": 8228
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.278410319433183e-05,
+      "loss": 0.2703,
+      "step": 8229
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.276301010123912e-05,
+      "loss": 0.2841,
+      "step": 8230
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.274191733182381e-05,
+      "loss": 0.2747,
+      "step": 8231
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.272082488702931e-05,
+      "loss": 0.2742,
+      "step": 8232
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.269973276779897e-05,
+      "loss": 0.2829,
+      "step": 8233
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.267864097507615e-05,
+      "loss": 0.2511,
+      "step": 8234
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.265754950980419e-05,
+      "loss": 0.2658,
+      "step": 8235
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.263645837292642e-05,
+      "loss": 0.2688,
+      "step": 8236
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.261536756538613e-05,
+      "loss": 0.2463,
+      "step": 8237
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.259427708812664e-05,
+      "loss": 0.2741,
+      "step": 8238
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.257318694209122e-05,
+      "loss": 0.2796,
+      "step": 8239
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255209712822315e-05,
+      "loss": 0.2742,
+      "step": 8240
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.253100764746565e-05,
+      "loss": 0.2722,
+      "step": 8241
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.250991850076198e-05,
+      "loss": 0.2849,
+      "step": 8242
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.248882968905536e-05,
+      "loss": 0.2569,
+      "step": 8243
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.246774121328898e-05,
+      "loss": 0.2555,
+      "step": 8244
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244665307440604e-05,
+      "loss": 0.2677,
+      "step": 8245
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.242556527334972e-05,
+      "loss": 0.289,
+      "step": 8246
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.24044778110632e-05,
+      "loss": 0.2657,
+      "step": 8247
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.23833906884896e-05,
+      "loss": 0.2544,
+      "step": 8248
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.236230390657203e-05,
+      "loss": 0.2745,
+      "step": 8249
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234121746625364e-05,
+      "loss": 0.2686,
+      "step": 8250
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.232013136847751e-05,
+      "loss": 0.2815,
+      "step": 8251
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.229904561418674e-05,
+      "loss": 0.2745,
+      "step": 8252
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.227796020432437e-05,
+      "loss": 0.2831,
+      "step": 8253
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.22568751398335e-05,
+      "loss": 0.2841,
+      "step": 8254
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.223579042165714e-05,
+      "loss": 0.2824,
+      "step": 8255
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.221470605073832e-05,
+      "loss": 0.2724,
+      "step": 8256
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.219362202802004e-05,
+      "loss": 0.2755,
+      "step": 8257
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.217253835444529e-05,
+      "loss": 0.2803,
+      "step": 8258
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.215145503095708e-05,
+      "loss": 0.2705,
+      "step": 8259
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.213037205849832e-05,
+      "loss": 0.2713,
+      "step": 8260
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.210928943801198e-05,
+      "loss": 0.2521,
+      "step": 8261
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.2088207170441e-05,
+      "loss": 0.2602,
+      "step": 8262
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.206712525672826e-05,
+      "loss": 0.2729,
+      "step": 8263
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.204604369781669e-05,
+      "loss": 0.2601,
+      "step": 8264
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.202496249464915e-05,
+      "loss": 0.2808,
+      "step": 8265
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.200388164816853e-05,
+      "loss": 0.2751,
+      "step": 8266
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.198280115931765e-05,
+      "loss": 0.2607,
+      "step": 8267
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.196172102903938e-05,
+      "loss": 0.2615,
+      "step": 8268
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19406412582765e-05,
+      "loss": 0.2654,
+      "step": 8269
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.191956184797183e-05,
+      "loss": 0.2569,
+      "step": 8270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.189848279906819e-05,
+      "loss": 0.2548,
+      "step": 8271
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.187740411250827e-05,
+      "loss": 0.2906,
+      "step": 8272
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.185632578923486e-05,
+      "loss": 0.2893,
+      "step": 8273
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.183524783019071e-05,
+      "loss": 0.2652,
+      "step": 8274
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.181417023631854e-05,
+      "loss": 0.2707,
+      "step": 8275
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.179309300856105e-05,
+      "loss": 0.266,
+      "step": 8276
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.177201614786092e-05,
+      "loss": 0.2569,
+      "step": 8277
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.175093965516083e-05,
+      "loss": 0.2773,
+      "step": 8278
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.172986353140342e-05,
+      "loss": 0.2612,
+      "step": 8279
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.170878777753136e-05,
+      "loss": 0.2749,
+      "step": 8280
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.168771239448724e-05,
+      "loss": 0.2616,
+      "step": 8281
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.166663738321367e-05,
+      "loss": 0.2678,
+      "step": 8282
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.164556274465325e-05,
+      "loss": 0.2606,
+      "step": 8283
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.162448847974856e-05,
+      "loss": 0.2778,
+      "step": 8284
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.160341458944213e-05,
+      "loss": 0.2624,
+      "step": 8285
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.158234107467652e-05,
+      "loss": 0.2538,
+      "step": 8286
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.156126793639426e-05,
+      "loss": 0.2824,
+      "step": 8287
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.154019517553786e-05,
+      "loss": 0.286,
+      "step": 8288
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.151912279304975e-05,
+      "loss": 0.275,
+      "step": 8289
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.149805078987244e-05,
+      "loss": 0.2715,
+      "step": 8290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.147697916694839e-05,
+      "loss": 0.2939,
+      "step": 8291
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.145590792522002e-05,
+      "loss": 0.2696,
+      "step": 8292
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.143483706562975e-05,
+      "loss": 0.2802,
+      "step": 8293
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.141376658912001e-05,
+      "loss": 0.2786,
+      "step": 8294
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.139269649663316e-05,
+      "loss": 0.2692,
+      "step": 8295
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.137162678911158e-05,
+      "loss": 0.283,
+      "step": 8296
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.135055746749763e-05,
+      "loss": 0.2819,
+      "step": 8297
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.132948853273361e-05,
+      "loss": 0.2793,
+      "step": 8298
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.130841998576188e-05,
+      "loss": 0.2707,
+      "step": 8299
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.128735182752472e-05,
+      "loss": 0.2948,
+      "step": 8300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.126628405896441e-05,
+      "loss": 0.2798,
+      "step": 8301
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.124521668102322e-05,
+      "loss": 0.2647,
+      "step": 8302
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.122414969464338e-05,
+      "loss": 0.2699,
+      "step": 8303
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.120308310076715e-05,
+      "loss": 0.2763,
+      "step": 8304
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.118201690033672e-05,
+      "loss": 0.2835,
+      "step": 8305
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.116095109429429e-05,
+      "loss": 0.2678,
+      "step": 8306
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.113988568358204e-05,
+      "loss": 0.2755,
+      "step": 8307
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.111882066914217e-05,
+      "loss": 0.2676,
+      "step": 8308
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.109775605191674e-05,
+      "loss": 0.2712,
+      "step": 8309
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.107669183284791e-05,
+      "loss": 0.2784,
+      "step": 8310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.105562801287779e-05,
+      "loss": 0.2665,
+      "step": 8311
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.103456459294845e-05,
+      "loss": 0.2681,
+      "step": 8312
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.101350157400201e-05,
+      "loss": 0.255,
+      "step": 8313
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.099243895698046e-05,
+      "loss": 0.2684,
+      "step": 8314
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.097137674282588e-05,
+      "loss": 0.2595,
+      "step": 8315
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.095031493248027e-05,
+      "loss": 0.2692,
+      "step": 8316
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.092925352688564e-05,
+      "loss": 0.273,
+      "step": 8317
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.090819252698394e-05,
+      "loss": 0.2642,
+      "step": 8318
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.088713193371716e-05,
+      "loss": 0.2693,
+      "step": 8319
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.086607174802723e-05,
+      "loss": 0.2556,
+      "step": 8320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.084501197085608e-05,
+      "loss": 0.2655,
+      "step": 8321
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.082395260314564e-05,
+      "loss": 0.275,
+      "step": 8322
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.080289364583777e-05,
+      "loss": 0.2715,
+      "step": 8323
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.078183509987433e-05,
+      "loss": 0.2739,
+      "step": 8324
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.076077696619721e-05,
+      "loss": 0.2667,
+      "step": 8325
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.073971924574822e-05,
+      "loss": 0.2668,
+      "step": 8326
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.071866193946917e-05,
+      "loss": 0.2627,
+      "step": 8327
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.069760504830191e-05,
+      "loss": 0.267,
+      "step": 8328
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.067654857318812e-05,
+      "loss": 0.2709,
+      "step": 8329
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.065549251506964e-05,
+      "loss": 0.2586,
+      "step": 8330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.063443687488817e-05,
+      "loss": 0.2813,
+      "step": 8331
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.061338165358547e-05,
+      "loss": 0.2634,
+      "step": 8332
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.059232685210319e-05,
+      "loss": 0.2577,
+      "step": 8333
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.057127247138307e-05,
+      "loss": 0.2614,
+      "step": 8334
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.055021851236675e-05,
+      "loss": 0.2741,
+      "step": 8335
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.052916497599588e-05,
+      "loss": 0.2978,
+      "step": 8336
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.05081118632121e-05,
+      "loss": 0.251,
+      "step": 8337
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.048705917495698e-05,
+      "loss": 0.2685,
+      "step": 8338
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.046600691217215e-05,
+      "loss": 0.2827,
+      "step": 8339
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.044495507579917e-05,
+      "loss": 0.2862,
+      "step": 8340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.042390366677959e-05,
+      "loss": 0.2697,
+      "step": 8341
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.040285268605494e-05,
+      "loss": 0.2692,
+      "step": 8342
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.038180213456674e-05,
+      "loss": 0.2686,
+      "step": 8343
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.036075201325649e-05,
+      "loss": 0.2636,
+      "step": 8344
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.033970232306566e-05,
+      "loss": 0.2745,
+      "step": 8345
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.031865306493569e-05,
+      "loss": 0.2668,
+      "step": 8346
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.029760423980804e-05,
+      "loss": 0.2787,
+      "step": 8347
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.027655584862413e-05,
+      "loss": 0.2671,
+      "step": 8348
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.025550789232536e-05,
+      "loss": 0.2602,
+      "step": 8349
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.023446037185308e-05,
+      "loss": 0.2669,
+      "step": 8350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.021341328814866e-05,
+      "loss": 0.2757,
+      "step": 8351
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.019236664215344e-05,
+      "loss": 0.2653,
+      "step": 8352
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.017132043480875e-05,
+      "loss": 0.272,
+      "step": 8353
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.015027466705587e-05,
+      "loss": 0.261,
+      "step": 8354
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.012922933983612e-05,
+      "loss": 0.2697,
+      "step": 8355
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.010818445409073e-05,
+      "loss": 0.2656,
+      "step": 8356
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.008714001076094e-05,
+      "loss": 0.2674,
+      "step": 8357
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.0066096010788e-05,
+      "loss": 0.2621,
+      "step": 8358
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.004505245511307e-05,
+      "loss": 0.2651,
+      "step": 8359
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.002400934467737e-05,
+      "loss": 0.2729,
+      "step": 8360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.000296668042203e-05,
+      "loss": 0.2714,
+      "step": 8361
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.998192446328822e-05,
+      "loss": 0.2643,
+      "step": 8362
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.996088269421703e-05,
+      "loss": 0.2493,
+      "step": 8363
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.993984137414961e-05,
+      "loss": 0.2664,
+      "step": 8364
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.991880050402701e-05,
+      "loss": 0.2594,
+      "step": 8365
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.989776008479028e-05,
+      "loss": 0.261,
+      "step": 8366
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.987672011738049e-05,
+      "loss": 0.2613,
+      "step": 8367
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.985568060273869e-05,
+      "loss": 0.2718,
+      "step": 8368
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.98346415418058e-05,
+      "loss": 0.2574,
+      "step": 8369
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.981360293552283e-05,
+      "loss": 0.2707,
+      "step": 8370
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.979256478483076e-05,
+      "loss": 0.2577,
+      "step": 8371
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.977152709067051e-05,
+      "loss": 0.2666,
+      "step": 8372
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.975048985398302e-05,
+      "loss": 0.2748,
+      "step": 8373
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.972945307570917e-05,
+      "loss": 0.2498,
+      "step": 8374
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.970841675678987e-05,
+      "loss": 0.2704,
+      "step": 8375
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.968738089816594e-05,
+      "loss": 0.2661,
+      "step": 8376
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.966634550077823e-05,
+      "loss": 0.2767,
+      "step": 8377
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.964531056556758e-05,
+      "loss": 0.2685,
+      "step": 8378
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.962427609347475e-05,
+      "loss": 0.2553,
+      "step": 8379
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.960324208544055e-05,
+      "loss": 0.2628,
+      "step": 8380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.958220854240571e-05,
+      "loss": 0.263,
+      "step": 8381
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.956117546531097e-05,
+      "loss": 0.2838,
+      "step": 8382
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.954014285509705e-05,
+      "loss": 0.2744,
+      "step": 8383
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.951911071270465e-05,
+      "loss": 0.2782,
+      "step": 8384
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.949807903907441e-05,
+      "loss": 0.2768,
+      "step": 8385
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.947704783514701e-05,
+      "loss": 0.2624,
+      "step": 8386
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.945601710186306e-05,
+      "loss": 0.2646,
+      "step": 8387
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.943498684016324e-05,
+      "loss": 0.2824,
+      "step": 8388
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.941395705098801e-05,
+      "loss": 0.2762,
+      "step": 8389
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.939292773527802e-05,
+      "loss": 0.279,
+      "step": 8390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.937189889397378e-05,
+      "loss": 0.2565,
+      "step": 8391
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.935087052801585e-05,
+      "loss": 0.2439,
+      "step": 8392
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.932984263834469e-05,
+      "loss": 0.2766,
+      "step": 8393
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.930881522590082e-05,
+      "loss": 0.2717,
+      "step": 8394
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.928778829162467e-05,
+      "loss": 0.2691,
+      "step": 8395
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.926676183645671e-05,
+      "loss": 0.2643,
+      "step": 8396
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.924573586133732e-05,
+      "loss": 0.2642,
+      "step": 8397
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.922471036720694e-05,
+      "loss": 0.2814,
+      "step": 8398
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92036853550059e-05,
+      "loss": 0.2771,
+      "step": 8399
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.918266082567456e-05,
+      "loss": 0.2756,
+      "step": 8400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.916163678015327e-05,
+      "loss": 0.2625,
+      "step": 8401
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.914061321938235e-05,
+      "loss": 0.2698,
+      "step": 8402
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.911959014430208e-05,
+      "loss": 0.2597,
+      "step": 8403
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.909856755585269e-05,
+      "loss": 0.2674,
+      "step": 8404
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.907754545497446e-05,
+      "loss": 0.2534,
+      "step": 8405
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.905652384260759e-05,
+      "loss": 0.2619,
+      "step": 8406
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.903550271969228e-05,
+      "loss": 0.2629,
+      "step": 8407
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.901448208716875e-05,
+      "loss": 0.2589,
+      "step": 8408
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.899346194597714e-05,
+      "loss": 0.2569,
+      "step": 8409
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.897244229705753e-05,
+      "loss": 0.2598,
+      "step": 8410
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.89514231413501e-05,
+      "loss": 0.2673,
+      "step": 8411
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.893040447979489e-05,
+      "loss": 0.2681,
+      "step": 8412
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.890938631333202e-05,
+      "loss": 0.2586,
+      "step": 8413
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.888836864290149e-05,
+      "loss": 0.2647,
+      "step": 8414
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.886735146944337e-05,
+      "loss": 0.2377,
+      "step": 8415
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.884633479389761e-05,
+      "loss": 0.2586,
+      "step": 8416
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.882531861720423e-05,
+      "loss": 0.2652,
+      "step": 8417
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.880430294030316e-05,
+      "loss": 0.2561,
+      "step": 8418
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.878328776413433e-05,
+      "loss": 0.2764,
+      "step": 8419
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.876227308963771e-05,
+      "loss": 0.2708,
+      "step": 8420
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.874125891775312e-05,
+      "loss": 0.2707,
+      "step": 8421
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.872024524942048e-05,
+      "loss": 0.2659,
+      "step": 8422
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.86992320855796e-05,
+      "loss": 0.2563,
+      "step": 8423
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.867821942717032e-05,
+      "loss": 0.2701,
+      "step": 8424
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.865720727513243e-05,
+      "loss": 0.263,
+      "step": 8425
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.863619563040571e-05,
+      "loss": 0.2802,
+      "step": 8426
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.861518449392993e-05,
+      "loss": 0.2876,
+      "step": 8427
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.859417386664486e-05,
+      "loss": 0.2621,
+      "step": 8428
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.857316374949009e-05,
+      "loss": 0.27,
+      "step": 8429
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.85521541434054e-05,
+      "loss": 0.2511,
+      "step": 8430
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.853114504933043e-05,
+      "loss": 0.2582,
+      "step": 8431
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.85101364682048e-05,
+      "loss": 0.2679,
+      "step": 8432
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.848912840096815e-05,
+      "loss": 0.2599,
+      "step": 8433
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.846812084856008e-05,
+      "loss": 0.2749,
+      "step": 8434
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.844711381192015e-05,
+      "loss": 0.278,
+      "step": 8435
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.84261072919879e-05,
+      "loss": 0.2644,
+      "step": 8436
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.840510128970288e-05,
+      "loss": 0.2627,
+      "step": 8437
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.838409580600457e-05,
+      "loss": 0.2799,
+      "step": 8438
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.836309084183248e-05,
+      "loss": 0.2707,
+      "step": 8439
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.834208639812603e-05,
+      "loss": 0.2669,
+      "step": 8440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.832108247582467e-05,
+      "loss": 0.262,
+      "step": 8441
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.83000790758678e-05,
+      "loss": 0.258,
+      "step": 8442
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.827907619919481e-05,
+      "loss": 0.2662,
+      "step": 8443
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.825807384674507e-05,
+      "loss": 0.2604,
+      "step": 8444
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.823707201945792e-05,
+      "loss": 0.2646,
+      "step": 8445
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.821607071827265e-05,
+      "loss": 0.269,
+      "step": 8446
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.819506994412858e-05,
+      "loss": 0.272,
+      "step": 8447
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.817406969796502e-05,
+      "loss": 0.2581,
+      "step": 8448
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.815306998072112e-05,
+      "loss": 0.2438,
+      "step": 8449
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.813207079333613e-05,
+      "loss": 0.2383,
+      "step": 8450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.811107213674929e-05,
+      "loss": 0.2783,
+      "step": 8451
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.809007401189974e-05,
+      "loss": 0.2708,
+      "step": 8452
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.806907641972663e-05,
+      "loss": 0.2551,
+      "step": 8453
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.804807936116909e-05,
+      "loss": 0.2818,
+      "step": 8454
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.802708283716623e-05,
+      "loss": 0.2857,
+      "step": 8455
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.800608684865712e-05,
+      "loss": 0.279,
+      "step": 8456
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.798509139658082e-05,
+      "loss": 0.2577,
+      "step": 8457
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.796409648187636e-05,
+      "loss": 0.2722,
+      "step": 8458
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.794310210548273e-05,
+      "loss": 0.2687,
+      "step": 8459
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.792210826833894e-05,
+      "loss": 0.2755,
+      "step": 8460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.790111497138395e-05,
+      "loss": 0.2704,
+      "step": 8461
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.788012221555666e-05,
+      "loss": 0.2723,
+      "step": 8462
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.7859130001796e-05,
+      "loss": 0.2575,
+      "step": 8463
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.783813833104086e-05,
+      "loss": 0.2677,
+      "step": 8464
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.781714720423009e-05,
+      "loss": 0.253,
+      "step": 8465
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.779615662230254e-05,
+      "loss": 0.2638,
+      "step": 8466
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.7775166586197e-05,
+      "loss": 0.264,
+      "step": 8467
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.775417709685235e-05,
+      "loss": 0.2654,
+      "step": 8468
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.773318815520722e-05,
+      "loss": 0.2483,
+      "step": 8469
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.771219976220038e-05,
+      "loss": 0.2466,
+      "step": 8470
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.76912119187706e-05,
+      "loss": 0.2619,
+      "step": 8471
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.767022462585654e-05,
+      "loss": 0.262,
+      "step": 8472
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.764923788439686e-05,
+      "loss": 0.2744,
+      "step": 8473
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.762825169533021e-05,
+      "loss": 0.2706,
+      "step": 8474
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.76072660595952e-05,
+      "loss": 0.2742,
+      "step": 8475
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.758628097813044e-05,
+      "loss": 0.2674,
+      "step": 8476
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.756529645187446e-05,
+      "loss": 0.2705,
+      "step": 8477
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.754431248176583e-05,
+      "loss": 0.2495,
+      "step": 8478
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.752332906874308e-05,
+      "loss": 0.2683,
+      "step": 8479
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.750234621374466e-05,
+      "loss": 0.2666,
+      "step": 8480
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.748136391770909e-05,
+      "loss": 0.2721,
+      "step": 8481
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.746038218157475e-05,
+      "loss": 0.2574,
+      "step": 8482
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.74394010062801e-05,
+      "loss": 0.2673,
+      "step": 8483
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.741842039276352e-05,
+      "loss": 0.2745,
+      "step": 8484
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.739744034196338e-05,
+      "loss": 0.2741,
+      "step": 8485
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.737646085481801e-05,
+      "loss": 0.2553,
+      "step": 8486
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.735548193226575e-05,
+      "loss": 0.2685,
+      "step": 8487
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.733450357524489e-05,
+      "loss": 0.28,
+      "step": 8488
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.731352578469365e-05,
+      "loss": 0.2615,
+      "step": 8489
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.729254856155031e-05,
+      "loss": 0.2743,
+      "step": 8490
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.727157190675307e-05,
+      "loss": 0.2698,
+      "step": 8491
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.725059582124016e-05,
+      "loss": 0.2675,
+      "step": 8492
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.72296203059497e-05,
+      "loss": 0.2622,
+      "step": 8493
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.720864536181983e-05,
+      "loss": 0.2652,
+      "step": 8494
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.718767098978868e-05,
+      "loss": 0.2485,
+      "step": 8495
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.716669719079431e-05,
+      "loss": 0.2639,
+      "step": 8496
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.714572396577483e-05,
+      "loss": 0.2713,
+      "step": 8497
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.712475131566822e-05,
+      "loss": 0.2688,
+      "step": 8498
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.710377924141255e-05,
+      "loss": 0.2763,
+      "step": 8499
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.708280774394576e-05,
+      "loss": 0.2753,
+      "step": 8500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.706183682420584e-05,
+      "loss": 0.2498,
+      "step": 8501
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.704086648313068e-05,
+      "loss": 0.2718,
+      "step": 8502
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.701989672165823e-05,
+      "loss": 0.2613,
+      "step": 8503
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.699892754072636e-05,
+      "loss": 0.2643,
+      "step": 8504
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.697795894127291e-05,
+      "loss": 0.2499,
+      "step": 8505
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.695699092423573e-05,
+      "loss": 0.2613,
+      "step": 8506
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.693602349055261e-05,
+      "loss": 0.2772,
+      "step": 8507
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.691505664116137e-05,
+      "loss": 0.2878,
+      "step": 8508
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.689409037699969e-05,
+      "loss": 0.27,
+      "step": 8509
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.687312469900532e-05,
+      "loss": 0.2744,
+      "step": 8510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.685215960811596e-05,
+      "loss": 0.2688,
+      "step": 8511
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.683119510526931e-05,
+      "loss": 0.2801,
+      "step": 8512
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.6810231191403e-05,
+      "loss": 0.2708,
+      "step": 8513
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.678926786745462e-05,
+      "loss": 0.2803,
+      "step": 8514
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.67683051343618e-05,
+      "loss": 0.2823,
+      "step": 8515
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.674734299306208e-05,
+      "loss": 0.262,
+      "step": 8516
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.672638144449303e-05,
+      "loss": 0.291,
+      "step": 8517
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.670542048959215e-05,
+      "loss": 0.27,
+      "step": 8518
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.668446012929693e-05,
+      "loss": 0.2695,
+      "step": 8519
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.666350036454482e-05,
+      "loss": 0.2621,
+      "step": 8520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.664254119627328e-05,
+      "loss": 0.2715,
+      "step": 8521
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.662158262541968e-05,
+      "loss": 0.2632,
+      "step": 8522
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.660062465292142e-05,
+      "loss": 0.2548,
+      "step": 8523
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.657966727971587e-05,
+      "loss": 0.2545,
+      "step": 8524
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.655871050674035e-05,
+      "loss": 0.2614,
+      "step": 8525
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.653775433493214e-05,
+      "loss": 0.2713,
+      "step": 8526
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.651679876522853e-05,
+      "loss": 0.2582,
+      "step": 8527
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.64958437985668e-05,
+      "loss": 0.2704,
+      "step": 8528
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.64748894358841e-05,
+      "loss": 0.2521,
+      "step": 8529
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.645393567811765e-05,
+      "loss": 0.2447,
+      "step": 8530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.643298252620462e-05,
+      "loss": 0.2633,
+      "step": 8531
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.641202998108215e-05,
+      "loss": 0.2688,
+      "step": 8532
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.639107804368735e-05,
+      "loss": 0.2789,
+      "step": 8533
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.637012671495732e-05,
+      "loss": 0.2649,
+      "step": 8534
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.63491759958291e-05,
+      "loss": 0.2683,
+      "step": 8535
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.632822588723972e-05,
+      "loss": 0.2615,
+      "step": 8536
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.630727639012617e-05,
+      "loss": 0.2442,
+      "step": 8537
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.628632750542546e-05,
+      "loss": 0.2665,
+      "step": 8538
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.626537923407451e-05,
+      "loss": 0.2608,
+      "step": 8539
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.624443157701024e-05,
+      "loss": 0.2676,
+      "step": 8540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.622348453516955e-05,
+      "loss": 0.265,
+      "step": 8541
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620253810948931e-05,
+      "loss": 0.2542,
+      "step": 8542
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.618159230090637e-05,
+      "loss": 0.2542,
+      "step": 8543
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.61606471103575e-05,
+      "loss": 0.2715,
+      "step": 8544
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.613970253877952e-05,
+      "loss": 0.2614,
+      "step": 8545
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.611875858710919e-05,
+      "loss": 0.2794,
+      "step": 8546
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.60978152562832e-05,
+      "loss": 0.2612,
+      "step": 8547
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.607687254723829e-05,
+      "loss": 0.2701,
+      "step": 8548
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.605593046091107e-05,
+      "loss": 0.2619,
+      "step": 8549
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.603498899823823e-05,
+      "loss": 0.2793,
+      "step": 8550
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.601404816015638e-05,
+      "loss": 0.2711,
+      "step": 8551
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.599310794760211e-05,
+      "loss": 0.2494,
+      "step": 8552
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.597216836151198e-05,
+      "loss": 0.2579,
+      "step": 8553
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.595122940282251e-05,
+      "loss": 0.2771,
+      "step": 8554
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.593029107247022e-05,
+      "loss": 0.2563,
+      "step": 8555
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.590935337139155e-05,
+      "loss": 0.2553,
+      "step": 8556
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.5888416300523e-05,
+      "loss": 0.2627,
+      "step": 8557
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.586747986080095e-05,
+      "loss": 0.2754,
+      "step": 8558
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.584654405316183e-05,
+      "loss": 0.2553,
+      "step": 8559
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.582560887854196e-05,
+      "loss": 0.2802,
+      "step": 8560
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.580467433787769e-05,
+      "loss": 0.258,
+      "step": 8561
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.578374043210533e-05,
+      "loss": 0.2598,
+      "step": 8562
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.576280716216116e-05,
+      "loss": 0.2571,
+      "step": 8563
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.574187452898141e-05,
+      "loss": 0.2721,
+      "step": 8564
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.572094253350234e-05,
+      "loss": 0.2646,
+      "step": 8565
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.570001117666008e-05,
+      "loss": 0.2772,
+      "step": 8566
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.567908045939088e-05,
+      "loss": 0.2564,
+      "step": 8567
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.565815038263081e-05,
+      "loss": 0.2603,
+      "step": 8568
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.563722094731599e-05,
+      "loss": 0.2635,
+      "step": 8569
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.561629215438249e-05,
+      "loss": 0.2583,
+      "step": 8570
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55953640047664e-05,
+      "loss": 0.2517,
+      "step": 8571
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.557443649940368e-05,
+      "loss": 0.2649,
+      "step": 8572
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.555350963923037e-05,
+      "loss": 0.258,
+      "step": 8573
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55325834251824e-05,
+      "loss": 0.2525,
+      "step": 8574
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.551165785819572e-05,
+      "loss": 0.2571,
+      "step": 8575
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.549073293920624e-05,
+      "loss": 0.2774,
+      "step": 8576
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.546980866914982e-05,
+      "loss": 0.2744,
+      "step": 8577
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.544888504896233e-05,
+      "loss": 0.2849,
+      "step": 8578
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.542796207957956e-05,
+      "loss": 0.265,
+      "step": 8579
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.540703976193732e-05,
+      "loss": 0.2568,
+      "step": 8580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.538611809697136e-05,
+      "loss": 0.267,
+      "step": 8581
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.53651970856174e-05,
+      "loss": 0.2699,
+      "step": 8582
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.534427672881117e-05,
+      "loss": 0.2547,
+      "step": 8583
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.532335702748831e-05,
+      "loss": 0.2517,
+      "step": 8584
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.530243798258449e-05,
+      "loss": 0.2652,
+      "step": 8585
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.52815195950353e-05,
+      "loss": 0.2683,
+      "step": 8586
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.526060186577634e-05,
+      "loss": 0.273,
+      "step": 8587
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.52396847957432e-05,
+      "loss": 0.2603,
+      "step": 8588
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.521876838587132e-05,
+      "loss": 0.2538,
+      "step": 8589
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.519785263709624e-05,
+      "loss": 0.2587,
+      "step": 8590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.517693755035343e-05,
+      "loss": 0.2696,
+      "step": 8591
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.515602312657831e-05,
+      "loss": 0.263,
+      "step": 8592
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.513510936670629e-05,
+      "loss": 0.2627,
+      "step": 8593
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.511419627167277e-05,
+      "loss": 0.2764,
+      "step": 8594
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.509328384241307e-05,
+      "loss": 0.2559,
+      "step": 8595
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.507237207986251e-05,
+      "loss": 0.2715,
+      "step": 8596
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.50514609849564e-05,
+      "loss": 0.2539,
+      "step": 8597
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.503055055862997e-05,
+      "loss": 0.2535,
+      "step": 8598
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.500964080181845e-05,
+      "loss": 0.2562,
+      "step": 8599
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.498873171545707e-05,
+      "loss": 0.2565,
+      "step": 8600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.496782330048096e-05,
+      "loss": 0.2622,
+      "step": 8601
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.494691555782528e-05,
+      "loss": 0.2494,
+      "step": 8602
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.492600848842512e-05,
+      "loss": 0.2597,
+      "step": 8603
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.490510209321558e-05,
+      "loss": 0.2536,
+      "step": 8604
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.48841963731317e-05,
+      "loss": 0.2655,
+      "step": 8605
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.486329132910848e-05,
+      "loss": 0.272,
+      "step": 8606
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.484238696208091e-05,
+      "loss": 0.2863,
+      "step": 8607
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.482148327298402e-05,
+      "loss": 0.3106,
+      "step": 8608
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.480058026275262e-05,
+      "loss": 0.284,
+      "step": 8609
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.477967793232164e-05,
+      "loss": 0.2616,
+      "step": 8610
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.475877628262598e-05,
+      "loss": 0.2792,
+      "step": 8611
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.473787531460046e-05,
+      "loss": 0.28,
+      "step": 8612
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.471697502917988e-05,
+      "loss": 0.2847,
+      "step": 8613
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.469607542729901e-05,
+      "loss": 0.2718,
+      "step": 8614
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.467517650989262e-05,
+      "loss": 0.2709,
+      "step": 8615
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.465427827789538e-05,
+      "loss": 0.2733,
+      "step": 8616
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.463338073224201e-05,
+      "loss": 0.2602,
+      "step": 8617
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.461248387386714e-05,
+      "loss": 0.2532,
+      "step": 8618
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.45915877037054e-05,
+      "loss": 0.2729,
+      "step": 8619
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.457069222269138e-05,
+      "loss": 0.2513,
+      "step": 8620
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.454979743175964e-05,
+      "loss": 0.2685,
+      "step": 8621
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.452890333184468e-05,
+      "loss": 0.2507,
+      "step": 8622
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.450800992388106e-05,
+      "loss": 0.2539,
+      "step": 8623
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.448711720880322e-05,
+      "loss": 0.2659,
+      "step": 8624
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.446622518754555e-05,
+      "loss": 0.2626,
+      "step": 8625
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.44453338610425e-05,
+      "loss": 0.2657,
+      "step": 8626
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.442444323022843e-05,
+      "loss": 0.2495,
+      "step": 8627
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.44035532960377e-05,
+      "loss": 0.2736,
+      "step": 8628
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.438266405940458e-05,
+      "loss": 0.2638,
+      "step": 8629
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.436177552126339e-05,
+      "loss": 0.2618,
+      "step": 8630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.434088768254835e-05,
+      "loss": 0.2688,
+      "step": 8631
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43200005441937e-05,
+      "loss": 0.2892,
+      "step": 8632
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.429911410713361e-05,
+      "loss": 0.2533,
+      "step": 8633
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.427822837230227e-05,
+      "loss": 0.2644,
+      "step": 8634
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.425734334063375e-05,
+      "loss": 0.2426,
+      "step": 8635
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.423645901306219e-05,
+      "loss": 0.2704,
+      "step": 8636
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.421557539052161e-05,
+      "loss": 0.2682,
+      "step": 8637
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.419469247394605e-05,
+      "loss": 0.2739,
+      "step": 8638
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.417381026426952e-05,
+      "loss": 0.2636,
+      "step": 8639
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.415292876242598e-05,
+      "loss": 0.2604,
+      "step": 8640
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.413204796934936e-05,
+      "loss": 0.2609,
+      "step": 8641
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.411116788597358e-05,
+      "loss": 0.2637,
+      "step": 8642
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.409028851323247e-05,
+      "loss": 0.2745,
+      "step": 8643
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.406940985205991e-05,
+      "loss": 0.2591,
+      "step": 8644
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.404853190338969e-05,
+      "loss": 0.2532,
+      "step": 8645
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.402765466815558e-05,
+      "loss": 0.2559,
+      "step": 8646
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.400677814729135e-05,
+      "loss": 0.264,
+      "step": 8647
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.398590234173069e-05,
+      "loss": 0.2721,
+      "step": 8648
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.39650272524073e-05,
+      "loss": 0.268,
+      "step": 8649
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.394415288025477e-05,
+      "loss": 0.2511,
+      "step": 8650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.392327922620675e-05,
+      "loss": 0.2688,
+      "step": 8651
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.390240629119685e-05,
+      "loss": 0.2596,
+      "step": 8652
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.388153407615857e-05,
+      "loss": 0.2625,
+      "step": 8653
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.386066258202546e-05,
+      "loss": 0.2617,
+      "step": 8654
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.3839791809731e-05,
+      "loss": 0.256,
+      "step": 8655
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.381892176020866e-05,
+      "loss": 0.2463,
+      "step": 8656
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.379805243439184e-05,
+      "loss": 0.2622,
+      "step": 8657
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.377718383321393e-05,
+      "loss": 0.2628,
+      "step": 8658
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.37563159576083e-05,
+      "loss": 0.2616,
+      "step": 8659
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.373544880850826e-05,
+      "loss": 0.2682,
+      "step": 8660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.371458238684711e-05,
+      "loss": 0.2598,
+      "step": 8661
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.369371669355812e-05,
+      "loss": 0.2588,
+      "step": 8662
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.367285172957449e-05,
+      "loss": 0.2618,
+      "step": 8663
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.365198749582945e-05,
+      "loss": 0.2598,
+      "step": 8664
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.363112399325613e-05,
+      "loss": 0.2488,
+      "step": 8665
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.361026122278768e-05,
+      "loss": 0.244,
+      "step": 8666
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.358939918535718e-05,
+      "loss": 0.2523,
+      "step": 8667
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.356853788189774e-05,
+      "loss": 0.2555,
+      "step": 8668
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.354767731334231e-05,
+      "loss": 0.2561,
+      "step": 8669
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.352681748062394e-05,
+      "loss": 0.2644,
+      "step": 8670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.350595838467557e-05,
+      "loss": 0.2583,
+      "step": 8671
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.348510002643014e-05,
+      "loss": 0.2559,
+      "step": 8672
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.346424240682055e-05,
+      "loss": 0.2689,
+      "step": 8673
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.344338552677971e-05,
+      "loss": 0.2699,
+      "step": 8674
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.342252938724037e-05,
+      "loss": 0.253,
+      "step": 8675
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.340167398913539e-05,
+      "loss": 0.2709,
+      "step": 8676
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.338081933339752e-05,
+      "loss": 0.2658,
+      "step": 8677
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.33599654209595e-05,
+      "loss": 0.2581,
+      "step": 8678
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.3339112252754e-05,
+      "loss": 0.2583,
+      "step": 8679
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.331825982971372e-05,
+      "loss": 0.2576,
+      "step": 8680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.329740815277128e-05,
+      "loss": 0.2419,
+      "step": 8681
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.327655722285929e-05,
+      "loss": 0.2637,
+      "step": 8682
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.32557070409103e-05,
+      "loss": 0.2674,
+      "step": 8683
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.323485760785686e-05,
+      "loss": 0.2627,
+      "step": 8684
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.321400892463145e-05,
+      "loss": 0.2479,
+      "step": 8685
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.319316099216657e-05,
+      "loss": 0.2571,
+      "step": 8686
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.31723138113946e-05,
+      "loss": 0.2557,
+      "step": 8687
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.315146738324803e-05,
+      "loss": 0.2512,
+      "step": 8688
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.313062170865914e-05,
+      "loss": 0.273,
+      "step": 8689
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.310977678856025e-05,
+      "loss": 0.2559,
+      "step": 8690
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.308893262388371e-05,
+      "loss": 0.2691,
+      "step": 8691
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.306808921556178e-05,
+      "loss": 0.2661,
+      "step": 8692
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.304724656452664e-05,
+      "loss": 0.2611,
+      "step": 8693
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.302640467171057e-05,
+      "loss": 0.2428,
+      "step": 8694
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.300556353804566e-05,
+      "loss": 0.2421,
+      "step": 8695
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.298472316446408e-05,
+      "loss": 0.237,
+      "step": 8696
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.296388355189789e-05,
+      "loss": 0.2563,
+      "step": 8697
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.294304470127919e-05,
+      "loss": 0.2688,
+      "step": 8698
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.292220661353997e-05,
+      "loss": 0.2622,
+      "step": 8699
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.290136928961225e-05,
+      "loss": 0.2664,
+      "step": 8700
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.288053273042797e-05,
+      "loss": 0.2571,
+      "step": 8701
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.285969693691909e-05,
+      "loss": 0.2656,
+      "step": 8702
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.283886191001745e-05,
+      "loss": 0.2793,
+      "step": 8703
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.281802765065491e-05,
+      "loss": 0.2661,
+      "step": 8704
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.279719415976333e-05,
+      "loss": 0.2595,
+      "step": 8705
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.277636143827445e-05,
+      "loss": 0.2702,
+      "step": 8706
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.275552948712006e-05,
+      "loss": 0.2513,
+      "step": 8707
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.273469830723189e-05,
+      "loss": 0.2533,
+      "step": 8708
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.271386789954156e-05,
+      "loss": 0.2878,
+      "step": 8709
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.269303826498076e-05,
+      "loss": 0.2882,
+      "step": 8710
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.267220940448111e-05,
+      "loss": 0.2664,
+      "step": 8711
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.265138131897416e-05,
+      "loss": 0.2696,
+      "step": 8712
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.263055400939149e-05,
+      "loss": 0.2609,
+      "step": 8713
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.260972747666461e-05,
+      "loss": 0.2607,
+      "step": 8714
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.258890172172496e-05,
+      "loss": 0.269,
+      "step": 8715
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2568076745504e-05,
+      "loss": 0.2723,
+      "step": 8716
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.254725254893314e-05,
+      "loss": 0.2668,
+      "step": 8717
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.252642913294373e-05,
+      "loss": 0.2595,
+      "step": 8718
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.250560649846714e-05,
+      "loss": 0.2609,
+      "step": 8719
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.248478464643464e-05,
+      "loss": 0.2562,
+      "step": 8720
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.246396357777753e-05,
+      "loss": 0.2474,
+      "step": 8721
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2443143293427e-05,
+      "loss": 0.2592,
+      "step": 8722
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.242232379431428e-05,
+      "loss": 0.2809,
+      "step": 8723
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.240150508137051e-05,
+      "loss": 0.2547,
+      "step": 8724
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.238068715552684e-05,
+      "loss": 0.2629,
+      "step": 8725
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.235987001771433e-05,
+      "loss": 0.2601,
+      "step": 8726
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.233905366886406e-05,
+      "loss": 0.259,
+      "step": 8727
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.231823810990707e-05,
+      "loss": 0.2664,
+      "step": 8728
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.229742334177428e-05,
+      "loss": 0.2613,
+      "step": 8729
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.227660936539667e-05,
+      "loss": 0.2618,
+      "step": 8730
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.225579618170515e-05,
+      "loss": 0.2658,
+      "step": 8731
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.223498379163061e-05,
+      "loss": 0.2544,
+      "step": 8732
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.221417219610389e-05,
+      "loss": 0.2572,
+      "step": 8733
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.219336139605579e-05,
+      "loss": 0.2552,
+      "step": 8734
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.21725513924171e-05,
+      "loss": 0.259,
+      "step": 8735
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.215174218611853e-05,
+      "loss": 0.2535,
+      "step": 8736
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.213093377809081e-05,
+      "loss": 0.2464,
+      "step": 8737
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.211012616926458e-05,
+      "loss": 0.2612,
+      "step": 8738
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.208931936057047e-05,
+      "loss": 0.2645,
+      "step": 8739
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.20685133529391e-05,
+      "loss": 0.2488,
+      "step": 8740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2047708147301e-05,
+      "loss": 0.2626,
+      "step": 8741
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.20269037445867e-05,
+      "loss": 0.253,
+      "step": 8742
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.200610014572667e-05,
+      "loss": 0.2631,
+      "step": 8743
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.19852973516514e-05,
+      "loss": 0.2547,
+      "step": 8744
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.196449536329127e-05,
+      "loss": 0.2639,
+      "step": 8745
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.194369418157665e-05,
+      "loss": 0.2534,
+      "step": 8746
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.192289380743793e-05,
+      "loss": 0.2579,
+      "step": 8747
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.190209424180542e-05,
+      "loss": 0.2452,
+      "step": 8748
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18812954856093e-05,
+      "loss": 0.2519,
+      "step": 8749
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.186049753977986e-05,
+      "loss": 0.2565,
+      "step": 8750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.183970040524728e-05,
+      "loss": 0.2605,
+      "step": 8751
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.181890408294177e-05,
+      "loss": 0.2505,
+      "step": 8752
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.179810857379338e-05,
+      "loss": 0.2746,
+      "step": 8753
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.177731387873225e-05,
+      "loss": 0.2636,
+      "step": 8754
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.175651999868843e-05,
+      "loss": 0.2516,
+      "step": 8755
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.173572693459191e-05,
+      "loss": 0.2469,
+      "step": 8756
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.171493468737268e-05,
+      "loss": 0.253,
+      "step": 8757
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.169414325796069e-05,
+      "loss": 0.2602,
+      "step": 8758
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.167335264728583e-05,
+      "loss": 0.2614,
+      "step": 8759
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.1652562856278e-05,
+      "loss": 0.2709,
+      "step": 8760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.163177388586697e-05,
+      "loss": 0.2585,
+      "step": 8761
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16109857369826e-05,
+      "loss": 0.2487,
+      "step": 8762
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.159019841055462e-05,
+      "loss": 0.2532,
+      "step": 8763
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.156941190751277e-05,
+      "loss": 0.2666,
+      "step": 8764
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.154862622878671e-05,
+      "loss": 0.251,
+      "step": 8765
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.152784137530609e-05,
+      "loss": 0.271,
+      "step": 8766
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.150705734800054e-05,
+      "loss": 0.2481,
+      "step": 8767
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.148627414779966e-05,
+      "loss": 0.243,
+      "step": 8768
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.146549177563292e-05,
+      "loss": 0.2555,
+      "step": 8769
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.144471023242985e-05,
+      "loss": 0.261,
+      "step": 8770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.14239295191199e-05,
+      "loss": 0.2496,
+      "step": 8771
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.140314963663253e-05,
+      "loss": 0.2638,
+      "step": 8772
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.138237058589711e-05,
+      "loss": 0.262,
+      "step": 8773
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.136159236784298e-05,
+      "loss": 0.2747,
+      "step": 8774
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.134081498339947e-05,
+      "loss": 0.282,
+      "step": 8775
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.132003843349586e-05,
+      "loss": 0.253,
+      "step": 8776
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.129926271906139e-05,
+      "loss": 0.2488,
+      "step": 8777
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.127848784102524e-05,
+      "loss": 0.2537,
+      "step": 8778
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.125771380031661e-05,
+      "loss": 0.2499,
+      "step": 8779
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.123694059786462e-05,
+      "loss": 0.2577,
+      "step": 8780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.121616823459832e-05,
+      "loss": 0.2498,
+      "step": 8781
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.119539671144681e-05,
+      "loss": 0.2753,
+      "step": 8782
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.117462602933908e-05,
+      "loss": 0.2663,
+      "step": 8783
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.115385618920412e-05,
+      "loss": 0.2559,
+      "step": 8784
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.113308719197086e-05,
+      "loss": 0.2568,
+      "step": 8785
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.11123190385682e-05,
+      "loss": 0.2579,
+      "step": 8786
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.109155172992503e-05,
+      "loss": 0.2627,
+      "step": 8787
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.107078526697016e-05,
+      "loss": 0.2547,
+      "step": 8788
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.105001965063236e-05,
+      "loss": 0.2449,
+      "step": 8789
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.10292548818404e-05,
+      "loss": 0.2548,
+      "step": 8790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.1008490961523e-05,
+      "loss": 0.2683,
+      "step": 8791
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.098772789060885e-05,
+      "loss": 0.2605,
+      "step": 8792
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.096696567002652e-05,
+      "loss": 0.2572,
+      "step": 8793
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.094620430070467e-05,
+      "loss": 0.2629,
+      "step": 8794
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.092544378357184e-05,
+      "loss": 0.2591,
+      "step": 8795
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.090468411955653e-05,
+      "loss": 0.2721,
+      "step": 8796
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.088392530958728e-05,
+      "loss": 0.2677,
+      "step": 8797
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.08631673545925e-05,
+      "loss": 0.2641,
+      "step": 8798
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.084241025550062e-05,
+      "loss": 0.2608,
+      "step": 8799
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.082165401323997e-05,
+      "loss": 0.2561,
+      "step": 8800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.080089862873892e-05,
+      "loss": 0.2799,
+      "step": 8801
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.078014410292574e-05,
+      "loss": 0.2503,
+      "step": 8802
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.075939043672871e-05,
+      "loss": 0.2629,
+      "step": 8803
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.073863763107603e-05,
+      "loss": 0.2595,
+      "step": 8804
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.071788568689586e-05,
+      "loss": 0.2547,
+      "step": 8805
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.069713460511639e-05,
+      "loss": 0.2485,
+      "step": 8806
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.067638438666567e-05,
+      "loss": 0.2451,
+      "step": 8807
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.065563503247182e-05,
+      "loss": 0.258,
+      "step": 8808
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.06348865434628e-05,
+      "loss": 0.2609,
+      "step": 8809
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.06141389205666e-05,
+      "loss": 0.2624,
+      "step": 8810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.05933921647112e-05,
+      "loss": 0.2678,
+      "step": 8811
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.05726462768245e-05,
+      "loss": 0.2723,
+      "step": 8812
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.055190125783435e-05,
+      "loss": 0.2624,
+      "step": 8813
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053115710866858e-05,
+      "loss": 0.2647,
+      "step": 8814
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.051041383025502e-05,
+      "loss": 0.2453,
+      "step": 8815
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.048967142352139e-05,
+      "loss": 0.2585,
+      "step": 8816
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.04689298893954e-05,
+      "loss": 0.249,
+      "step": 8817
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.044818922880473e-05,
+      "loss": 0.2477,
+      "step": 8818
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.042744944267703e-05,
+      "loss": 0.2545,
+      "step": 8819
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.040671053193986e-05,
+      "loss": 0.257,
+      "step": 8820
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.038597249752082e-05,
+      "loss": 0.2551,
+      "step": 8821
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.036523534034738e-05,
+      "loss": 0.2561,
+      "step": 8822
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.034449906134707e-05,
+      "loss": 0.2622,
+      "step": 8823
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03237636614473e-05,
+      "loss": 0.2517,
+      "step": 8824
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.030302914157546e-05,
+      "loss": 0.2478,
+      "step": 8825
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.028229550265894e-05,
+      "loss": 0.2712,
+      "step": 8826
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.026156274562502e-05,
+      "loss": 0.2537,
+      "step": 8827
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.024083087140106e-05,
+      "loss": 0.2481,
+      "step": 8828
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02200998809142e-05,
+      "loss": 0.2659,
+      "step": 8829
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.019936977509169e-05,
+      "loss": 0.2719,
+      "step": 8830
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.017864055486069e-05,
+      "loss": 0.2591,
+      "step": 8831
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.015791222114834e-05,
+      "loss": 0.2671,
+      "step": 8832
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.013718477488168e-05,
+      "loss": 0.2494,
+      "step": 8833
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.01164582169878e-05,
+      "loss": 0.2459,
+      "step": 8834
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.009573254839367e-05,
+      "loss": 0.2527,
+      "step": 8835
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.007500777002629e-05,
+      "loss": 0.2676,
+      "step": 8836
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.005428388281255e-05,
+      "loss": 0.2684,
+      "step": 8837
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.003356088767934e-05,
+      "loss": 0.263,
+      "step": 8838
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.001283878555354e-05,
+      "loss": 0.2573,
+      "step": 8839
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.999211757736191e-05,
+      "loss": 0.2695,
+      "step": 8840
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.997139726403124e-05,
+      "loss": 0.2492,
+      "step": 8841
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.995067784648824e-05,
+      "loss": 0.2538,
+      "step": 8842
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.992995932565961e-05,
+      "loss": 0.2536,
+      "step": 8843
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.990924170247198e-05,
+      "loss": 0.2485,
+      "step": 8844
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.988852497785197e-05,
+      "loss": 0.253,
+      "step": 8845
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.986780915272614e-05,
+      "loss": 0.2527,
+      "step": 8846
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.9847094228021e-05,
+      "loss": 0.243,
+      "step": 8847
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.982638020466306e-05,
+      "loss": 0.262,
+      "step": 8848
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.980566708357872e-05,
+      "loss": 0.2459,
+      "step": 8849
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.978495486569443e-05,
+      "loss": 0.2423,
+      "step": 8850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.976424355193653e-05,
+      "loss": 0.2636,
+      "step": 8851
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.974353314323132e-05,
+      "loss": 0.2611,
+      "step": 8852
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.972282364050512e-05,
+      "loss": 0.2572,
+      "step": 8853
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.970211504468415e-05,
+      "loss": 0.2601,
+      "step": 8854
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.968140735669464e-05,
+      "loss": 0.2667,
+      "step": 8855
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.966070057746271e-05,
+      "loss": 0.2526,
+      "step": 8856
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96399947079145e-05,
+      "loss": 0.2683,
+      "step": 8857
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96192897489761e-05,
+      "loss": 0.2751,
+      "step": 8858
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.959858570157352e-05,
+      "loss": 0.2563,
+      "step": 8859
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.957788256663277e-05,
+      "loss": 0.2718,
+      "step": 8860
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.955718034507982e-05,
+      "loss": 0.2665,
+      "step": 8861
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.953647903784056e-05,
+      "loss": 0.2675,
+      "step": 8862
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.951577864584088e-05,
+      "loss": 0.2637,
+      "step": 8863
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.949507917000662e-05,
+      "loss": 0.2503,
+      "step": 8864
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.947438061126355e-05,
+      "loss": 0.258,
+      "step": 8865
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.945368297053745e-05,
+      "loss": 0.2758,
+      "step": 8866
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.9432986248754e-05,
+      "loss": 0.256,
+      "step": 8867
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.941229044683892e-05,
+      "loss": 0.2545,
+      "step": 8868
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.939159556571778e-05,
+      "loss": 0.2695,
+      "step": 8869
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.937090160631621e-05,
+      "loss": 0.2507,
+      "step": 8870
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.935020856955974e-05,
+      "loss": 0.251,
+      "step": 8871
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.932951645637385e-05,
+      "loss": 0.2535,
+      "step": 8872
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.930882526768403e-05,
+      "loss": 0.2484,
+      "step": 8873
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92881350044157e-05,
+      "loss": 0.2596,
+      "step": 8874
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.926744566749424e-05,
+      "loss": 0.2554,
+      "step": 8875
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.924675725784499e-05,
+      "loss": 0.2492,
+      "step": 8876
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.922606977639324e-05,
+      "loss": 0.2514,
+      "step": 8877
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.920538322406426e-05,
+      "loss": 0.2692,
+      "step": 8878
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.918469760178326e-05,
+      "loss": 0.2586,
+      "step": 8879
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91640129104754e-05,
+      "loss": 0.261,
+      "step": 8880
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.914332915106584e-05,
+      "loss": 0.2618,
+      "step": 8881
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.912264632447963e-05,
+      "loss": 0.256,
+      "step": 8882
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.910196443164185e-05,
+      "loss": 0.2559,
+      "step": 8883
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.908128347347749e-05,
+      "loss": 0.2475,
+      "step": 8884
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.906060345091153e-05,
+      "loss": 0.2505,
+      "step": 8885
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.90399243648689e-05,
+      "loss": 0.2522,
+      "step": 8886
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.901924621627443e-05,
+      "loss": 0.2587,
+      "step": 8887
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.899856900605304e-05,
+      "loss": 0.2415,
+      "step": 8888
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.897789273512946e-05,
+      "loss": 0.265,
+      "step": 8889
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.895721740442845e-05,
+      "loss": 0.2676,
+      "step": 8890
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.893654301487474e-05,
+      "loss": 0.2464,
+      "step": 8891
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.891586956739301e-05,
+      "loss": 0.2613,
+      "step": 8892
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.889519706290787e-05,
+      "loss": 0.2518,
+      "step": 8893
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.887452550234391e-05,
+      "loss": 0.249,
+      "step": 8894
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.885385488662568e-05,
+      "loss": 0.2591,
+      "step": 8895
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.883318521667768e-05,
+      "loss": 0.2539,
+      "step": 8896
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.881251649342437e-05,
+      "loss": 0.2567,
+      "step": 8897
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.879184871779018e-05,
+      "loss": 0.2477,
+      "step": 8898
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.877118189069945e-05,
+      "loss": 0.2568,
+      "step": 8899
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.875051601307655e-05,
+      "loss": 0.2616,
+      "step": 8900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.872985108584574e-05,
+      "loss": 0.2519,
+      "step": 8901
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.870918710993129e-05,
+      "loss": 0.2617,
+      "step": 8902
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.86885240862574e-05,
+      "loss": 0.243,
+      "step": 8903
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.866786201574822e-05,
+      "loss": 0.2516,
+      "step": 8904
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.86472008993279e-05,
+      "loss": 0.2589,
+      "step": 8905
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.862654073792048e-05,
+      "loss": 0.2425,
+      "step": 8906
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.860588153245002e-05,
+      "loss": 0.2616,
+      "step": 8907
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.858522328384053e-05,
+      "loss": 0.25,
+      "step": 8908
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.856456599301591e-05,
+      "loss": 0.2647,
+      "step": 8909
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.854390966090007e-05,
+      "loss": 0.2645,
+      "step": 8910
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.852325428841691e-05,
+      "loss": 0.2656,
+      "step": 8911
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.850259987649021e-05,
+      "loss": 0.2569,
+      "step": 8912
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.848194642604379e-05,
+      "loss": 0.259,
+      "step": 8913
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.846129393800133e-05,
+      "loss": 0.2636,
+      "step": 8914
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.844064241328658e-05,
+      "loss": 0.2615,
+      "step": 8915
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.841999185282316e-05,
+      "loss": 0.2472,
+      "step": 8916
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.839934225753467e-05,
+      "loss": 0.256,
+      "step": 8917
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.837869362834468e-05,
+      "loss": 0.262,
+      "step": 8918
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.83580459661767e-05,
+      "loss": 0.2488,
+      "step": 8919
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.833739927195422e-05,
+      "loss": 0.2566,
+      "step": 8920
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.831675354660066e-05,
+      "loss": 0.259,
+      "step": 8921
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.829610879103942e-05,
+      "loss": 0.2528,
+      "step": 8922
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.827546500619383e-05,
+      "loss": 0.2556,
+      "step": 8923
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.825482219298724e-05,
+      "loss": 0.2801,
+      "step": 8924
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.823418035234284e-05,
+      "loss": 0.3003,
+      "step": 8925
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.821353948518387e-05,
+      "loss": 0.2629,
+      "step": 8926
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.81928995924335e-05,
+      "loss": 0.269,
+      "step": 8927
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.81722606750149e-05,
+      "loss": 0.2907,
+      "step": 8928
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.815162273385108e-05,
+      "loss": 0.247,
+      "step": 8929
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.813098576986512e-05,
+      "loss": 0.2472,
+      "step": 8930
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.811034978398003e-05,
+      "loss": 0.2538,
+      "step": 8931
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.808971477711874e-05,
+      "loss": 0.2653,
+      "step": 8932
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.806908075020416e-05,
+      "loss": 0.2711,
+      "step": 8933
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.804844770415918e-05,
+      "loss": 0.2621,
+      "step": 8934
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.802781563990658e-05,
+      "loss": 0.2727,
+      "step": 8935
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.800718455836921e-05,
+      "loss": 0.2535,
+      "step": 8936
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.798655446046972e-05,
+      "loss": 0.2734,
+      "step": 8937
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.796592534713085e-05,
+      "loss": 0.2381,
+      "step": 8938
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.794529721927522e-05,
+      "loss": 0.2629,
+      "step": 8939
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.792467007782544e-05,
+      "loss": 0.2624,
+      "step": 8940
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.790404392370409e-05,
+      "loss": 0.2645,
+      "step": 8941
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.788341875783365e-05,
+      "loss": 0.2747,
+      "step": 8942
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.786279458113661e-05,
+      "loss": 0.2617,
+      "step": 8943
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.784217139453538e-05,
+      "loss": 0.2564,
+      "step": 8944
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.782154919895236e-05,
+      "loss": 0.2616,
+      "step": 8945
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.780092799530986e-05,
+      "loss": 0.2444,
+      "step": 8946
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.778030778453021e-05,
+      "loss": 0.2472,
+      "step": 8947
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.775968856753566e-05,
+      "loss": 0.253,
+      "step": 8948
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.773907034524837e-05,
+      "loss": 0.2437,
+      "step": 8949
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.771845311859049e-05,
+      "loss": 0.2699,
+      "step": 8950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.769783688848418e-05,
+      "loss": 0.2579,
+      "step": 8951
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.767722165585148e-05,
+      "loss": 0.27,
+      "step": 8952
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.765660742161442e-05,
+      "loss": 0.2497,
+      "step": 8953
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.7635994186695e-05,
+      "loss": 0.2441,
+      "step": 8954
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.761538195201514e-05,
+      "loss": 0.2429,
+      "step": 8955
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.759477071849673e-05,
+      "loss": 0.25,
+      "step": 8956
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.757416048706163e-05,
+      "loss": 0.2376,
+      "step": 8957
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.755355125863162e-05,
+      "loss": 0.2674,
+      "step": 8958
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.753294303412846e-05,
+      "loss": 0.2493,
+      "step": 8959
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.751233581447388e-05,
+      "loss": 0.2488,
+      "step": 8960
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.749172960058952e-05,
+      "loss": 0.2513,
+      "step": 8961
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.747112439339704e-05,
+      "loss": 0.2486,
+      "step": 8962
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.745052019381798e-05,
+      "loss": 0.2497,
+      "step": 8963
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.742991700277387e-05,
+      "loss": 0.2549,
+      "step": 8964
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.740931482118623e-05,
+      "loss": 0.2484,
+      "step": 8965
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.738871364997647e-05,
+      "loss": 0.2542,
+      "step": 8966
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.736811349006601e-05,
+      "loss": 0.2422,
+      "step": 8967
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.734751434237622e-05,
+      "loss": 0.2538,
+      "step": 8968
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.732691620782834e-05,
+      "loss": 0.2667,
+      "step": 8969
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.730631908734366e-05,
+      "loss": 0.2541,
+      "step": 8970
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72857229818434e-05,
+      "loss": 0.2665,
+      "step": 8971
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.726512789224873e-05,
+      "loss": 0.2468,
+      "step": 8972
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.724453381948077e-05,
+      "loss": 0.251,
+      "step": 8973
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72239407644606e-05,
+      "loss": 0.2375,
+      "step": 8974
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.720334872810926e-05,
+      "loss": 0.2643,
+      "step": 8975
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.718275771134773e-05,
+      "loss": 0.2343,
+      "step": 8976
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.716216771509694e-05,
+      "loss": 0.2528,
+      "step": 8977
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.714157874027781e-05,
+      "loss": 0.2425,
+      "step": 8978
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.71209907878112e-05,
+      "loss": 0.2681,
+      "step": 8979
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.710040385861787e-05,
+      "loss": 0.2669,
+      "step": 8980
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.707981795361861e-05,
+      "loss": 0.2686,
+      "step": 8981
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.705923307373414e-05,
+      "loss": 0.2578,
+      "step": 8982
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.70386492198851e-05,
+      "loss": 0.2615,
+      "step": 8983
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.701806639299214e-05,
+      "loss": 0.2479,
+      "step": 8984
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.699748459397582e-05,
+      "loss": 0.2512,
+      "step": 8985
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.697690382375666e-05,
+      "loss": 0.256,
+      "step": 8986
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.695632408325517e-05,
+      "loss": 0.2649,
+      "step": 8987
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.693574537339181e-05,
+      "loss": 0.2376,
+      "step": 8988
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.691516769508689e-05,
+      "loss": 0.251,
+      "step": 8989
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.68945910492608e-05,
+      "loss": 0.2577,
+      "step": 8990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.687401543683384e-05,
+      "loss": 0.2663,
+      "step": 8991
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.685344085872625e-05,
+      "loss": 0.2414,
+      "step": 8992
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.683286731585826e-05,
+      "loss": 0.256,
+      "step": 8993
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.681229480915003e-05,
+      "loss": 0.2561,
+      "step": 8994
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.679172333952165e-05,
+      "loss": 0.2409,
+      "step": 8995
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67711529078932e-05,
+      "loss": 0.2625,
+      "step": 8996
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.675058351518471e-05,
+      "loss": 0.2636,
+      "step": 8997
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.673001516231614e-05,
+      "loss": 0.2578,
+      "step": 8998
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.670944785020743e-05,
+      "loss": 0.2542,
+      "step": 8999
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.668888157977845e-05,
+      "loss": 0.2473,
+      "step": 9000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.666831635194904e-05,
+      "loss": 0.25,
+      "step": 9001
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.6647752167639e-05,
+      "loss": 0.2592,
+      "step": 9002
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.662718902776807e-05,
+      "loss": 0.2465,
+      "step": 9003
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.660662693325592e-05,
+      "loss": 0.2624,
+      "step": 9004
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.658606588502221e-05,
+      "loss": 0.2547,
+      "step": 9005
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.656550588398655e-05,
+      "loss": 0.2395,
+      "step": 9006
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.65449469310685e-05,
+      "loss": 0.2686,
+      "step": 9007
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.652438902718757e-05,
+      "loss": 0.2378,
+      "step": 9008
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.65038321732632e-05,
+      "loss": 0.2371,
+      "step": 9009
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64832763702148e-05,
+      "loss": 0.2478,
+      "step": 9010
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.646272161896176e-05,
+      "loss": 0.2653,
+      "step": 9011
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64421679204234e-05,
+      "loss": 0.2567,
+      "step": 9012
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.642161527551896e-05,
+      "loss": 0.2514,
+      "step": 9013
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.640106368516772e-05,
+      "loss": 0.2597,
+      "step": 9014
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.638051315028882e-05,
+      "loss": 0.2492,
+      "step": 9015
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.635996367180139e-05,
+      "loss": 0.2612,
+      "step": 9016
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.633941525062452e-05,
+      "loss": 0.2488,
+      "step": 9017
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.631886788767726e-05,
+      "loss": 0.2525,
+      "step": 9018
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.629832158387858e-05,
+      "loss": 0.2622,
+      "step": 9019
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.627777634014743e-05,
+      "loss": 0.2625,
+      "step": 9020
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.62572321574027e-05,
+      "loss": 0.2634,
+      "step": 9021
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.623668903656326e-05,
+      "loss": 0.2589,
+      "step": 9022
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.621614697854788e-05,
+      "loss": 0.2569,
+      "step": 9023
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.619560598427533e-05,
+      "loss": 0.246,
+      "step": 9024
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.617506605466431e-05,
+      "loss": 0.2566,
+      "step": 9025
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.615452719063349e-05,
+      "loss": 0.2549,
+      "step": 9026
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.613398939310145e-05,
+      "loss": 0.2504,
+      "step": 9027
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.611345266298681e-05,
+      "loss": 0.2491,
+      "step": 9028
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6092917001208e-05,
+      "loss": 0.2818,
+      "step": 9029
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.607238240868354e-05,
+      "loss": 0.2615,
+      "step": 9030
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.605184888633183e-05,
+      "loss": 0.2277,
+      "step": 9031
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.603131643507122e-05,
+      "loss": 0.2704,
+      "step": 9032
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.60107850558201e-05,
+      "loss": 0.2524,
+      "step": 9033
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.599025474949668e-05,
+      "loss": 0.2716,
+      "step": 9034
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.596972551701921e-05,
+      "loss": 0.2587,
+      "step": 9035
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.594919735930587e-05,
+      "loss": 0.2452,
+      "step": 9036
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.59286702772748e-05,
+      "loss": 0.2516,
+      "step": 9037
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.590814427184407e-05,
+      "loss": 0.2521,
+      "step": 9038
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.588761934393172e-05,
+      "loss": 0.2456,
+      "step": 9039
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.586709549445572e-05,
+      "loss": 0.2579,
+      "step": 9040
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.584657272433401e-05,
+      "loss": 0.2363,
+      "step": 9041
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.582605103448452e-05,
+      "loss": 0.2416,
+      "step": 9042
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.580553042582503e-05,
+      "loss": 0.2449,
+      "step": 9043
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.578501089927338e-05,
+      "loss": 0.2518,
+      "step": 9044
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57644924557473e-05,
+      "loss": 0.2562,
+      "step": 9045
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.574397509616447e-05,
+      "loss": 0.2525,
+      "step": 9046
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.572345882144256e-05,
+      "loss": 0.2508,
+      "step": 9047
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57029436324992e-05,
+      "loss": 0.2497,
+      "step": 9048
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.568242953025186e-05,
+      "loss": 0.261,
+      "step": 9049
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.566191651561807e-05,
+      "loss": 0.2617,
+      "step": 9050
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.564140458951531e-05,
+      "loss": 0.2534,
+      "step": 9051
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.562089375286097e-05,
+      "loss": 0.2447,
+      "step": 9052
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.56003840065724e-05,
+      "loss": 0.2528,
+      "step": 9053
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.557987535156692e-05,
+      "loss": 0.2563,
+      "step": 9054
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.555936778876177e-05,
+      "loss": 0.261,
+      "step": 9055
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.553886131907419e-05,
+      "loss": 0.2386,
+      "step": 9056
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.551835594342131e-05,
+      "loss": 0.2515,
+      "step": 9057
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.549785166272025e-05,
+      "loss": 0.2489,
+      "step": 9058
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.547734847788806e-05,
+      "loss": 0.2522,
+      "step": 9059
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.545684638984179e-05,
+      "loss": 0.2684,
+      "step": 9060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.543634539949838e-05,
+      "loss": 0.2615,
+      "step": 9061
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.541584550777475e-05,
+      "loss": 0.2728,
+      "step": 9062
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.539534671558776e-05,
+      "loss": 0.2578,
+      "step": 9063
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.537484902385423e-05,
+      "loss": 0.2634,
+      "step": 9064
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.535435243349094e-05,
+      "loss": 0.2653,
+      "step": 9065
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.533385694541459e-05,
+      "loss": 0.2708,
+      "step": 9066
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.53133625605419e-05,
+      "loss": 0.2421,
+      "step": 9067
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.529286927978941e-05,
+      "loss": 0.2466,
+      "step": 9068
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.527237710407374e-05,
+      "loss": 0.2569,
+      "step": 9069
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.525188603431138e-05,
+      "loss": 0.231,
+      "step": 9070
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.523139607141882e-05,
+      "loss": 0.2396,
+      "step": 9071
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.52109072163125e-05,
+      "loss": 0.2711,
+      "step": 9072
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.519041946990876e-05,
+      "loss": 0.26,
+      "step": 9073
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.516993283312396e-05,
+      "loss": 0.2568,
+      "step": 9074
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.514944730687433e-05,
+      "loss": 0.2677,
+      "step": 9075
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.512896289207612e-05,
+      "loss": 0.2665,
+      "step": 9076
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.510847958964552e-05,
+      "loss": 0.255,
+      "step": 9077
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.508799740049863e-05,
+      "loss": 0.2533,
+      "step": 9078
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.506751632555153e-05,
+      "loss": 0.2606,
+      "step": 9079
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.504703636572024e-05,
+      "loss": 0.2509,
+      "step": 9080
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.502655752192074e-05,
+      "loss": 0.2423,
+      "step": 9081
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.500607979506894e-05,
+      "loss": 0.2552,
+      "step": 9082
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.498560318608073e-05,
+      "loss": 0.2606,
+      "step": 9083
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.496512769587194e-05,
+      "loss": 0.2577,
+      "step": 9084
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.494465332535834e-05,
+      "loss": 0.2506,
+      "step": 9085
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.492418007545565e-05,
+      "loss": 0.2501,
+      "step": 9086
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.490370794707955e-05,
+      "loss": 0.2552,
+      "step": 9087
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.48832369411457e-05,
+      "loss": 0.2509,
+      "step": 9088
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.486276705856959e-05,
+      "loss": 0.2604,
+      "step": 9089
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.48422983002668e-05,
+      "loss": 0.2689,
+      "step": 9090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.48218306671528e-05,
+      "loss": 0.2546,
+      "step": 9091
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.480136416014303e-05,
+      "loss": 0.26,
+      "step": 9092
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.478089878015281e-05,
+      "loss": 0.2637,
+      "step": 9093
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.47604345280975e-05,
+      "loss": 0.2271,
+      "step": 9094
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.473997140489237e-05,
+      "loss": 0.2657,
+      "step": 9095
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.471950941145262e-05,
+      "loss": 0.2445,
+      "step": 9096
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.469904854869348e-05,
+      "loss": 0.2562,
+      "step": 9097
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.467858881752999e-05,
+      "loss": 0.2521,
+      "step": 9098
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.465813021887728e-05,
+      "loss": 0.2452,
+      "step": 9099
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.463767275365034e-05,
+      "loss": 0.257,
+      "step": 9100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.461721642276416e-05,
+      "loss": 0.2561,
+      "step": 9101
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.459676122713364e-05,
+      "loss": 0.2483,
+      "step": 9102
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.457630716767366e-05,
+      "loss": 0.2634,
+      "step": 9103
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.455585424529902e-05,
+      "loss": 0.2591,
+      "step": 9104
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.45354024609245e-05,
+      "loss": 0.2485,
+      "step": 9105
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.451495181546482e-05,
+      "loss": 0.2388,
+      "step": 9106
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.449450230983462e-05,
+      "loss": 0.2534,
+      "step": 9107
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.447405394494857e-05,
+      "loss": 0.2514,
+      "step": 9108
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.445360672172115e-05,
+      "loss": 0.2634,
+      "step": 9109
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.44331606410669e-05,
+      "loss": 0.2587,
+      "step": 9110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.44127157039003e-05,
+      "loss": 0.2674,
+      "step": 9111
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.439227191113575e-05,
+      "loss": 0.2441,
+      "step": 9112
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.43718292636876e-05,
+      "loss": 0.2515,
+      "step": 9113
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.435138776247016e-05,
+      "loss": 0.2383,
+      "step": 9114
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.433094740839768e-05,
+      "loss": 0.2501,
+      "step": 9115
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.431050820238438e-05,
+      "loss": 0.2601,
+      "step": 9116
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.429007014534439e-05,
+      "loss": 0.2533,
+      "step": 9117
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.426963323819183e-05,
+      "loss": 0.2429,
+      "step": 9118
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.424919748184074e-05,
+      "loss": 0.2469,
+      "step": 9119
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.42287628772051e-05,
+      "loss": 0.2542,
+      "step": 9120
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.42083294251989e-05,
+      "loss": 0.2537,
+      "step": 9121
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.418789712673599e-05,
+      "loss": 0.259,
+      "step": 9122
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.416746598273025e-05,
+      "loss": 0.2589,
+      "step": 9123
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.414703599409545e-05,
+      "loss": 0.244,
+      "step": 9124
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.412660716174532e-05,
+      "loss": 0.2602,
+      "step": 9125
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.410617948659357e-05,
+      "loss": 0.2458,
+      "step": 9126
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.408575296955383e-05,
+      "loss": 0.258,
+      "step": 9127
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.406532761153972e-05,
+      "loss": 0.2547,
+      "step": 9128
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.404490341346471e-05,
+      "loss": 0.2527,
+      "step": 9129
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.402448037624229e-05,
+      "loss": 0.2649,
+      "step": 9130
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.400405850078591e-05,
+      "loss": 0.2621,
+      "step": 9131
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.398363778800894e-05,
+      "loss": 0.2584,
+      "step": 9132
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.39632182388247e-05,
+      "loss": 0.269,
+      "step": 9133
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.394279985414649e-05,
+      "loss": 0.2461,
+      "step": 9134
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.392238263488751e-05,
+      "loss": 0.2599,
+      "step": 9135
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.390196658196091e-05,
+      "loss": 0.2526,
+      "step": 9136
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.388155169627984e-05,
+      "loss": 0.2622,
+      "step": 9137
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.386113797875736e-05,
+      "loss": 0.2401,
+      "step": 9138
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.384072543030644e-05,
+      "loss": 0.2659,
+      "step": 9139
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.382031405184011e-05,
+      "loss": 0.2646,
+      "step": 9140
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.379990384427123e-05,
+      "loss": 0.2249,
+      "step": 9141
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.377949480851264e-05,
+      "loss": 0.246,
+      "step": 9142
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.37590869454772e-05,
+      "loss": 0.2484,
+      "step": 9143
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.373868025607761e-05,
+      "loss": 0.2512,
+      "step": 9144
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.371827474122661e-05,
+      "loss": 0.246,
+      "step": 9145
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.36978704018368e-05,
+      "loss": 0.2358,
+      "step": 9146
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.36774672388208e-05,
+      "loss": 0.2347,
+      "step": 9147
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.365706525309116e-05,
+      "loss": 0.247,
+      "step": 9148
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.363666444556032e-05,
+      "loss": 0.2656,
+      "step": 9149
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.361626481714077e-05,
+      "loss": 0.2485,
+      "step": 9150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.359586636874484e-05,
+      "loss": 0.2574,
+      "step": 9151
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.357546910128492e-05,
+      "loss": 0.2476,
+      "step": 9152
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.355507301567323e-05,
+      "loss": 0.2578,
+      "step": 9153
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.353467811282202e-05,
+      "loss": 0.2564,
+      "step": 9154
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.351428439364346e-05,
+      "loss": 0.2586,
+      "step": 9155
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.349389185904967e-05,
+      "loss": 0.2588,
+      "step": 9156
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.347350050995272e-05,
+      "loss": 0.2364,
+      "step": 9157
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.345311034726459e-05,
+      "loss": 0.2497,
+      "step": 9158
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.343272137189727e-05,
+      "loss": 0.2643,
+      "step": 9159
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.341233358476264e-05,
+      "loss": 0.2525,
+      "step": 9160
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.339194698677258e-05,
+      "loss": 0.2614,
+      "step": 9161
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.337156157883889e-05,
+      "loss": 0.2669,
+      "step": 9162
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.335117736187329e-05,
+      "loss": 0.2503,
+      "step": 9163
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.333079433678749e-05,
+      "loss": 0.2672,
+      "step": 9164
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.331041250449312e-05,
+      "loss": 0.237,
+      "step": 9165
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.329003186590177e-05,
+      "loss": 0.2311,
+      "step": 9166
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.326965242192499e-05,
+      "loss": 0.2572,
+      "step": 9167
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.324927417347424e-05,
+      "loss": 0.2542,
+      "step": 9168
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.322889712146094e-05,
+      "loss": 0.2564,
+      "step": 9169
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.32085212667965e-05,
+      "loss": 0.2379,
+      "step": 9170
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.318814661039217e-05,
+      "loss": 0.2491,
+      "step": 9171
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.316777315315926e-05,
+      "loss": 0.2566,
+      "step": 9172
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.314740089600897e-05,
+      "loss": 0.2511,
+      "step": 9173
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.312702983985248e-05,
+      "loss": 0.2401,
+      "step": 9174
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.310665998560087e-05,
+      "loss": 0.247,
+      "step": 9175
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.30862913341652e-05,
+      "loss": 0.2527,
+      "step": 9176
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.306592388645647e-05,
+      "loss": 0.2399,
+      "step": 9177
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.304555764338561e-05,
+      "loss": 0.2546,
+      "step": 9178
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.302519260586351e-05,
+      "loss": 0.2612,
+      "step": 9179
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.300482877480103e-05,
+      "loss": 0.2456,
+      "step": 9180
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.298446615110892e-05,
+      "loss": 0.2552,
+      "step": 9181
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.296410473569791e-05,
+      "loss": 0.2269,
+      "step": 9182
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.29437445294787e-05,
+      "loss": 0.2526,
+      "step": 9183
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.292338553336187e-05,
+      "loss": 0.2401,
+      "step": 9184
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.290302774825803e-05,
+      "loss": 0.2622,
+      "step": 9185
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.288267117507766e-05,
+      "loss": 0.2518,
+      "step": 9186
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.28623158147312e-05,
+      "loss": 0.2254,
+      "step": 9187
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.284196166812914e-05,
+      "loss": 0.2511,
+      "step": 9188
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.28216087361817e-05,
+      "loss": 0.2637,
+      "step": 9189
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.280125701979925e-05,
+      "loss": 0.2512,
+      "step": 9190
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.278090651989201e-05,
+      "loss": 0.2465,
+      "step": 9191
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.276055723737017e-05,
+      "loss": 0.2495,
+      "step": 9192
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.274020917314386e-05,
+      "loss": 0.252,
+      "step": 9193
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.271986232812315e-05,
+      "loss": 0.2483,
+      "step": 9194
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.269951670321806e-05,
+      "loss": 0.2384,
+      "step": 9195
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.267917229933859e-05,
+      "loss": 0.243,
+      "step": 9196
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.26588291173946e-05,
+      "loss": 0.2542,
+      "step": 9197
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.263848715829598e-05,
+      "loss": 0.2439,
+      "step": 9198
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.261814642295251e-05,
+      "loss": 0.2383,
+      "step": 9199
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.259780691227397e-05,
+      "loss": 0.2368,
+      "step": 9200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.257746862717003e-05,
+      "loss": 0.2476,
+      "step": 9201
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.255713156855034e-05,
+      "loss": 0.2467,
+      "step": 9202
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.253679573732449e-05,
+      "loss": 0.2443,
+      "step": 9203
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.251646113440195e-05,
+      "loss": 0.2556,
+      "step": 9204
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.249612776069228e-05,
+      "loss": 0.2386,
+      "step": 9205
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.247579561710485e-05,
+      "loss": 0.2543,
+      "step": 9206
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.245546470454904e-05,
+      "loss": 0.2487,
+      "step": 9207
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.243513502393418e-05,
+      "loss": 0.2574,
+      "step": 9208
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.241480657616946e-05,
+      "loss": 0.2504,
+      "step": 9209
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.23944793621641e-05,
+      "loss": 0.2345,
+      "step": 9210
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.237415338282727e-05,
+      "loss": 0.2435,
+      "step": 9211
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.235382863906804e-05,
+      "loss": 0.2435,
+      "step": 9212
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.233350513179544e-05,
+      "loss": 0.2445,
+      "step": 9213
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.231318286191846e-05,
+      "loss": 0.242,
+      "step": 9214
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.229286183034603e-05,
+      "loss": 0.2538,
+      "step": 9215
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.227254203798698e-05,
+      "loss": 0.2598,
+      "step": 9216
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.225222348575015e-05,
+      "loss": 0.2573,
+      "step": 9217
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.223190617454429e-05,
+      "loss": 0.2576,
+      "step": 9218
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.221159010527811e-05,
+      "loss": 0.257,
+      "step": 9219
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.219127527886024e-05,
+      "loss": 0.2299,
+      "step": 9220
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.217096169619927e-05,
+      "loss": 0.2545,
+      "step": 9221
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.215064935820374e-05,
+      "loss": 0.2655,
+      "step": 9222
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.213033826578215e-05,
+      "loss": 0.2481,
+      "step": 9223
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.211002841984287e-05,
+      "loss": 0.2582,
+      "step": 9224
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.20897198212943e-05,
+      "loss": 0.2498,
+      "step": 9225
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.206941247104474e-05,
+      "loss": 0.2392,
+      "step": 9226
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.204910637000245e-05,
+      "loss": 0.2453,
+      "step": 9227
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.202880151907565e-05,
+      "loss": 0.2553,
+      "step": 9228
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.200849791917244e-05,
+      "loss": 0.2428,
+      "step": 9229
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.198819557120093e-05,
+      "loss": 0.2478,
+      "step": 9230
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.196789447606915e-05,
+      "loss": 0.2453,
+      "step": 9231
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.194759463468507e-05,
+      "loss": 0.2503,
+      "step": 9232
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.192729604795662e-05,
+      "loss": 0.2454,
+      "step": 9233
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.190699871679165e-05,
+      "loss": 0.2464,
+      "step": 9234
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1886702642098e-05,
+      "loss": 0.2479,
+      "step": 9235
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.186640782478335e-05,
+      "loss": 0.2475,
+      "step": 9236
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.184611426575547e-05,
+      "loss": 0.2298,
+      "step": 9237
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.182582196592195e-05,
+      "loss": 0.2473,
+      "step": 9238
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.180553092619039e-05,
+      "loss": 0.2566,
+      "step": 9239
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.17852411474683e-05,
+      "loss": 0.241,
+      "step": 9240
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.176495263066318e-05,
+      "loss": 0.2395,
+      "step": 9241
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.174466537668242e-05,
+      "loss": 0.2433,
+      "step": 9242
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.172437938643337e-05,
+      "loss": 0.256,
+      "step": 9243
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.170409466082335e-05,
+      "loss": 0.2457,
+      "step": 9244
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.168381120075957e-05,
+      "loss": 0.2353,
+      "step": 9245
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.166352900714926e-05,
+      "loss": 0.2466,
+      "step": 9246
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.164324808089952e-05,
+      "loss": 0.2479,
+      "step": 9247
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.162296842291749e-05,
+      "loss": 0.2526,
+      "step": 9248
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.160269003411006e-05,
+      "loss": 0.2521,
+      "step": 9249
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.158241291538428e-05,
+      "loss": 0.2582,
+      "step": 9250
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1562137067647e-05,
+      "loss": 0.2503,
+      "step": 9251
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.154186249180512e-05,
+      "loss": 0.2478,
+      "step": 9252
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.15215891887654e-05,
+      "loss": 0.2511,
+      "step": 9253
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.150131715943457e-05,
+      "loss": 0.2644,
+      "step": 9254
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.148104640471932e-05,
+      "loss": 0.2625,
+      "step": 9255
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.146077692552624e-05,
+      "loss": 0.2545,
+      "step": 9256
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.144050872276193e-05,
+      "loss": 0.2471,
+      "step": 9257
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.142024179733287e-05,
+      "loss": 0.2529,
+      "step": 9258
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.139997615014551e-05,
+      "loss": 0.2534,
+      "step": 9259
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.137971178210623e-05,
+      "loss": 0.2563,
+      "step": 9260
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.135944869412137e-05,
+      "loss": 0.2542,
+      "step": 9261
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.133918688709722e-05,
+      "loss": 0.2599,
+      "step": 9262
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.131892636194e-05,
+      "loss": 0.266,
+      "step": 9263
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.129866711955583e-05,
+      "loss": 0.2448,
+      "step": 9264
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.127840916085085e-05,
+      "loss": 0.2501,
+      "step": 9265
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.125815248673109e-05,
+      "loss": 0.264,
+      "step": 9266
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.123789709810255e-05,
+      "loss": 0.2449,
+      "step": 9267
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.12176429958712e-05,
+      "loss": 0.2362,
+      "step": 9268
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.11973901809428e-05,
+      "loss": 0.2558,
+      "step": 9269
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.117713865422326e-05,
+      "loss": 0.2612,
+      "step": 9270
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.115688841661829e-05,
+      "loss": 0.266,
+      "step": 9271
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.113663946903364e-05,
+      "loss": 0.2532,
+      "step": 9272
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.11163918123749e-05,
+      "loss": 0.2602,
+      "step": 9273
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.109614544754768e-05,
+      "loss": 0.26,
+      "step": 9274
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.107590037545753e-05,
+      "loss": 0.2677,
+      "step": 9275
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.105565659700987e-05,
+      "loss": 0.2676,
+      "step": 9276
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.103541411311015e-05,
+      "loss": 0.2401,
+      "step": 9277
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.10151729246637e-05,
+      "loss": 0.2498,
+      "step": 9278
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.099493303257583e-05,
+      "loss": 0.2629,
+      "step": 9279
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.097469443775177e-05,
+      "loss": 0.2522,
+      "step": 9280
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.095445714109672e-05,
+      "loss": 0.2616,
+      "step": 9281
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.093422114351578e-05,
+      "loss": 0.2538,
+      "step": 9282
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.091398644591401e-05,
+      "loss": 0.258,
+      "step": 9283
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.089375304919642e-05,
+      "loss": 0.2535,
+      "step": 9284
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.087352095426798e-05,
+      "loss": 0.2608,
+      "step": 9285
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.085329016203354e-05,
+      "loss": 0.2579,
+      "step": 9286
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.083306067339796e-05,
+      "loss": 0.2582,
+      "step": 9287
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.081283248926604e-05,
+      "loss": 0.2454,
+      "step": 9288
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.079260561054242e-05,
+      "loss": 0.2617,
+      "step": 9289
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.07723800381318e-05,
+      "loss": 0.2569,
+      "step": 9290
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.075215577293877e-05,
+      "loss": 0.2471,
+      "step": 9291
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.073193281586785e-05,
+      "loss": 0.2519,
+      "step": 9292
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.071171116782356e-05,
+      "loss": 0.2617,
+      "step": 9293
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.06914908297103e-05,
+      "loss": 0.251,
+      "step": 9294
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.067127180243242e-05,
+      "loss": 0.2531,
+      "step": 9295
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.065105408689425e-05,
+      "loss": 0.2583,
+      "step": 9296
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.063083768400001e-05,
+      "loss": 0.2636,
+      "step": 9297
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.061062259465392e-05,
+      "loss": 0.2436,
+      "step": 9298
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.059040881976008e-05,
+      "loss": 0.248,
+      "step": 9299
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.057019636022258e-05,
+      "loss": 0.2485,
+      "step": 9300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.054998521694543e-05,
+      "loss": 0.2449,
+      "step": 9301
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.052977539083254e-05,
+      "loss": 0.2573,
+      "step": 9302
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.050956688278785e-05,
+      "loss": 0.2608,
+      "step": 9303
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.048935969371518e-05,
+      "loss": 0.2375,
+      "step": 9304
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.046915382451831e-05,
+      "loss": 0.2428,
+      "step": 9305
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.044894927610092e-05,
+      "loss": 0.2497,
+      "step": 9306
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.042874604936673e-05,
+      "loss": 0.2458,
+      "step": 9307
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.040854414521932e-05,
+      "loss": 0.2595,
+      "step": 9308
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.03883435645622e-05,
+      "loss": 0.2515,
+      "step": 9309
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.036814430829883e-05,
+      "loss": 0.2467,
+      "step": 9310
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.034794637733268e-05,
+      "loss": 0.2493,
+      "step": 9311
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.03277497725671e-05,
+      "loss": 0.2473,
+      "step": 9312
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.030755449490541e-05,
+      "loss": 0.2473,
+      "step": 9313
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.02873605452508e-05,
+      "loss": 0.2522,
+      "step": 9314
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.026716792450649e-05,
+      "loss": 0.2564,
+      "step": 9315
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.024697663357558e-05,
+      "loss": 0.256,
+      "step": 9316
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.022678667336115e-05,
+      "loss": 0.2531,
+      "step": 9317
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.020659804476621e-05,
+      "loss": 0.2584,
+      "step": 9318
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.018641074869369e-05,
+      "loss": 0.2482,
+      "step": 9319
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.016622478604651e-05,
+      "loss": 0.2308,
+      "step": 9320
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.014604015772743e-05,
+      "loss": 0.2573,
+      "step": 9321
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.012585686463928e-05,
+      "loss": 0.2399,
+      "step": 9322
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.010567490768473e-05,
+      "loss": 0.2587,
+      "step": 9323
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.008549428776646e-05,
+      "loss": 0.2374,
+      "step": 9324
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.006531500578699e-05,
+      "loss": 0.2486,
+      "step": 9325
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.004513706264892e-05,
+      "loss": 0.2494,
+      "step": 9326
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.002496045925468e-05,
+      "loss": 0.2678,
+      "step": 9327
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.000478519650673e-05,
+      "loss": 0.254,
+      "step": 9328
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.998461127530734e-05,
+      "loss": 0.2488,
+      "step": 9329
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.996443869655883e-05,
+      "loss": 0.2462,
+      "step": 9330
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.99442674611634e-05,
+      "loss": 0.2548,
+      "step": 9331
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.992409757002327e-05,
+      "loss": 0.2399,
+      "step": 9332
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.99039290240405e-05,
+      "loss": 0.2583,
+      "step": 9333
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.988376182411717e-05,
+      "loss": 0.2654,
+      "step": 9334
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.986359597115526e-05,
+      "loss": 0.2592,
+      "step": 9335
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.984343146605668e-05,
+      "loss": 0.2487,
+      "step": 9336
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.98232683097233e-05,
+      "loss": 0.2479,
+      "step": 9337
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.980310650305695e-05,
+      "loss": 0.2576,
+      "step": 9338
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.978294604695934e-05,
+      "loss": 0.2373,
+      "step": 9339
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.976278694233217e-05,
+      "loss": 0.2556,
+      "step": 9340
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.974262919007708e-05,
+      "loss": 0.2556,
+      "step": 9341
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.97224727910956e-05,
+      "loss": 0.2575,
+      "step": 9342
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.970231774628927e-05,
+      "loss": 0.249,
+      "step": 9343
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.96821640565595e-05,
+      "loss": 0.2516,
+      "step": 9344
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.966201172280771e-05,
+      "loss": 0.2492,
+      "step": 9345
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.964186074593519e-05,
+      "loss": 0.2327,
+      "step": 9346
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.96217111268432e-05,
+      "loss": 0.2457,
+      "step": 9347
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9601562866433e-05,
+      "loss": 0.2453,
+      "step": 9348
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.958141596560565e-05,
+      "loss": 0.2628,
+      "step": 9349
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.956127042526225e-05,
+      "loss": 0.2567,
+      "step": 9350
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.954112624630382e-05,
+      "loss": 0.2388,
+      "step": 9351
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.952098342963133e-05,
+      "loss": 0.239,
+      "step": 9352
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.950084197614569e-05,
+      "loss": 0.2524,
+      "step": 9353
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.948070188674769e-05,
+      "loss": 0.2511,
+      "step": 9354
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.946056316233814e-05,
+      "loss": 0.2378,
+      "step": 9355
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.944042580381773e-05,
+      "loss": 0.2505,
+      "step": 9356
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.942028981208713e-05,
+      "loss": 0.2497,
+      "step": 9357
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.940015518804691e-05,
+      "loss": 0.2547,
+      "step": 9358
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.938002193259763e-05,
+      "loss": 0.2352,
+      "step": 9359
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.935989004663974e-05,
+      "loss": 0.2313,
+      "step": 9360
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.933975953107365e-05,
+      "loss": 0.2536,
+      "step": 9361
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.931963038679969e-05,
+      "loss": 0.2379,
+      "step": 9362
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.929950261471818e-05,
+      "loss": 0.257,
+      "step": 9363
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.92793762157293e-05,
+      "loss": 0.2598,
+      "step": 9364
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.925925119073325e-05,
+      "loss": 0.2332,
+      "step": 9365
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.923912754063011e-05,
+      "loss": 0.2567,
+      "step": 9366
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.921900526631995e-05,
+      "loss": 0.239,
+      "step": 9367
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.919888436870272e-05,
+      "loss": 0.2623,
+      "step": 9368
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.91787648486783e-05,
+      "loss": 0.2506,
+      "step": 9369
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.915864670714659e-05,
+      "loss": 0.2466,
+      "step": 9370
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.913852994500737e-05,
+      "loss": 0.2508,
+      "step": 9371
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.911841456316037e-05,
+      "loss": 0.2257,
+      "step": 9372
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.2457,
+      "step": 9373
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.907818794394166e-05,
+      "loss": 0.2644,
+      "step": 9374
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.90580767083691e-05,
+      "loss": 0.2449,
+      "step": 9375
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.903796685668706e-05,
+      "loss": 0.2549,
+      "step": 9376
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.901785838979497e-05,
+      "loss": 0.2511,
+      "step": 9377
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.899775130859219e-05,
+      "loss": 0.2704,
+      "step": 9378
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.897764561397804e-05,
+      "loss": 0.2639,
+      "step": 9379
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.895754130685172e-05,
+      "loss": 0.252,
+      "step": 9380
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.89374383881124e-05,
+      "loss": 0.2616,
+      "step": 9381
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.89173368586592e-05,
+      "loss": 0.2494,
+      "step": 9382
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.88972367193912e-05,
+      "loss": 0.2522,
+      "step": 9383
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.887713797120735e-05,
+      "loss": 0.2562,
+      "step": 9384
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.885704061500659e-05,
+      "loss": 0.2519,
+      "step": 9385
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.883694465168778e-05,
+      "loss": 0.2663,
+      "step": 9386
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.881685008214976e-05,
+      "loss": 0.263,
+      "step": 9387
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.879675690729118e-05,
+      "loss": 0.267,
+      "step": 9388
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.877666512801078e-05,
+      "loss": 0.239,
+      "step": 9389
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.875657474520716e-05,
+      "loss": 0.2604,
+      "step": 9390
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.873648575977888e-05,
+      "loss": 0.2693,
+      "step": 9391
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.871639817262441e-05,
+      "loss": 0.2416,
+      "step": 9392
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.869631198464218e-05,
+      "loss": 0.2548,
+      "step": 9393
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.867622719673055e-05,
+      "loss": 0.2469,
+      "step": 9394
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.865614380978782e-05,
+      "loss": 0.235,
+      "step": 9395
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.863606182471225e-05,
+      "loss": 0.2459,
+      "step": 9396
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.861598124240201e-05,
+      "loss": 0.2527,
+      "step": 9397
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.859590206375518e-05,
+      "loss": 0.2384,
+      "step": 9398
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.857582428966985e-05,
+      "loss": 0.2584,
+      "step": 9399
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.855574792104398e-05,
+      "loss": 0.2498,
+      "step": 9400
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.85356729587755e-05,
+      "loss": 0.2529,
+      "step": 9401
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.851559940376229e-05,
+      "loss": 0.2467,
+      "step": 9402
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.849552725690212e-05,
+      "loss": 0.2558,
+      "step": 9403
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.847545651909274e-05,
+      "loss": 0.239,
+      "step": 9404
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.845538719123182e-05,
+      "loss": 0.2511,
+      "step": 9405
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.843531927421696e-05,
+      "loss": 0.2477,
+      "step": 9406
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.841525276894578e-05,
+      "loss": 0.2429,
+      "step": 9407
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.839518767631564e-05,
+      "loss": 0.2423,
+      "step": 9408
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.837512399722402e-05,
+      "loss": 0.2476,
+      "step": 9409
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.835506173256827e-05,
+      "loss": 0.2443,
+      "step": 9410
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.833500088324571e-05,
+      "loss": 0.2438,
+      "step": 9411
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.831494145015352e-05,
+      "loss": 0.2378,
+      "step": 9412
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.82948834341889e-05,
+      "loss": 0.2612,
+      "step": 9413
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.827482683624897e-05,
+      "loss": 0.2505,
+      "step": 9414
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.825477165723074e-05,
+      "loss": 0.2437,
+      "step": 9415
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.823471789803118e-05,
+      "loss": 0.2605,
+      "step": 9416
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.821466555954721e-05,
+      "loss": 0.2565,
+      "step": 9417
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.81946146426757e-05,
+      "loss": 0.2625,
+      "step": 9418
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.817456514831342e-05,
+      "loss": 0.2397,
+      "step": 9419
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.815451707735709e-05,
+      "loss": 0.2536,
+      "step": 9420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.813447043070337e-05,
+      "loss": 0.2509,
+      "step": 9421
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.811442520924885e-05,
+      "loss": 0.2517,
+      "step": 9422
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.809438141389008e-05,
+      "loss": 0.233,
+      "step": 9423
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.807433904552352e-05,
+      "loss": 0.2564,
+      "step": 9424
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.805429810504555e-05,
+      "loss": 0.2532,
+      "step": 9425
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.803425859335253e-05,
+      "loss": 0.2507,
+      "step": 9426
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.801422051134078e-05,
+      "loss": 0.2526,
+      "step": 9427
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.799418385990644e-05,
+      "loss": 0.2449,
+      "step": 9428
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.797414863994567e-05,
+      "loss": 0.2422,
+      "step": 9429
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.795411485235455e-05,
+      "loss": 0.2386,
+      "step": 9430
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.793408249802912e-05,
+      "loss": 0.2388,
+      "step": 9431
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.791405157786535e-05,
+      "loss": 0.252,
+      "step": 9432
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.78940220927591e-05,
+      "loss": 0.234,
+      "step": 9433
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.787399404360621e-05,
+      "loss": 0.2429,
+      "step": 9434
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.785396743130243e-05,
+      "loss": 0.2434,
+      "step": 9435
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.783394225674348e-05,
+      "loss": 0.2387,
+      "step": 9436
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7813918520825e-05,
+      "loss": 0.2434,
+      "step": 9437
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.779389622444254e-05,
+      "loss": 0.2554,
+      "step": 9438
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.77738753684916e-05,
+      "loss": 0.2552,
+      "step": 9439
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.775385595386764e-05,
+      "loss": 0.2537,
+      "step": 9440
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.773383798146602e-05,
+      "loss": 0.2519,
+      "step": 9441
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.771382145218207e-05,
+      "loss": 0.255,
+      "step": 9442
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.769380636691103e-05,
+      "loss": 0.2489,
+      "step": 9443
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.76737927265481e-05,
+      "loss": 0.2436,
+      "step": 9444
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.765378053198835e-05,
+      "loss": 0.2404,
+      "step": 9445
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.763376978412687e-05,
+      "loss": 0.247,
+      "step": 9446
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.761376048385868e-05,
+      "loss": 0.258,
+      "step": 9447
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.759375263207864e-05,
+      "loss": 0.2316,
+      "step": 9448
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.757374622968163e-05,
+      "loss": 0.2428,
+      "step": 9449
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.755374127756244e-05,
+      "loss": 0.2457,
+      "step": 9450
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.753373777661582e-05,
+      "loss": 0.2434,
+      "step": 9451
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.751373572773644e-05,
+      "loss": 0.2438,
+      "step": 9452
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.749373513181888e-05,
+      "loss": 0.2467,
+      "step": 9453
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.747373598975769e-05,
+      "loss": 0.2528,
+      "step": 9454
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.74537383024473e-05,
+      "loss": 0.2344,
+      "step": 9455
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.743374207078219e-05,
+      "loss": 0.2454,
+      "step": 9456
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.741374729565664e-05,
+      "loss": 0.2474,
+      "step": 9457
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.739375397796495e-05,
+      "loss": 0.2445,
+      "step": 9458
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.737376211860131e-05,
+      "loss": 0.2519,
+      "step": 9459
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.735377171845986e-05,
+      "loss": 0.2377,
+      "step": 9460
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733378277843472e-05,
+      "loss": 0.2467,
+      "step": 9461
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731379529941986e-05,
+      "loss": 0.2405,
+      "step": 9462
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.729380928230927e-05,
+      "loss": 0.2685,
+      "step": 9463
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.727382472799678e-05,
+      "loss": 0.245,
+      "step": 9464
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.725384163737625e-05,
+      "loss": 0.2486,
+      "step": 9465
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.723386001134142e-05,
+      "loss": 0.2407,
+      "step": 9466
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7213879850786e-05,
+      "loss": 0.2541,
+      "step": 9467
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.719390115660357e-05,
+      "loss": 0.2444,
+      "step": 9468
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.717392392968771e-05,
+      "loss": 0.2391,
+      "step": 9469
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.71539481709319e-05,
+      "loss": 0.2385,
+      "step": 9470
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.713397388122955e-05,
+      "loss": 0.2411,
+      "step": 9471
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.711400106147404e-05,
+      "loss": 0.25,
+      "step": 9472
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.709402971255866e-05,
+      "loss": 0.2493,
+      "step": 9473
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.707405983537663e-05,
+      "loss": 0.2643,
+      "step": 9474
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.705409143082114e-05,
+      "loss": 0.2476,
+      "step": 9475
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.703412449978522e-05,
+      "loss": 0.259,
+      "step": 9476
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.701415904316197e-05,
+      "loss": 0.2444,
+      "step": 9477
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.699419506184431e-05,
+      "loss": 0.254,
+      "step": 9478
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.697423255672515e-05,
+      "loss": 0.2532,
+      "step": 9479
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.695427152869735e-05,
+      "loss": 0.2499,
+      "step": 9480
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.693431197865362e-05,
+      "loss": 0.2444,
+      "step": 9481
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.691435390748671e-05,
+      "loss": 0.268,
+      "step": 9482
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.689439731608921e-05,
+      "loss": 0.2572,
+      "step": 9483
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.68744422053537e-05,
+      "loss": 0.2659,
+      "step": 9484
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.685448857617272e-05,
+      "loss": 0.2358,
+      "step": 9485
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.683453642943866e-05,
+      "loss": 0.2515,
+      "step": 9486
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.681458576604393e-05,
+      "loss": 0.2456,
+      "step": 9487
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.679463658688078e-05,
+      "loss": 0.2465,
+      "step": 9488
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.677468889284145e-05,
+      "loss": 0.2458,
+      "step": 9489
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.675474268481814e-05,
+      "loss": 0.2502,
+      "step": 9490
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.673479796370294e-05,
+      "loss": 0.2521,
+      "step": 9491
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.671485473038788e-05,
+      "loss": 0.2489,
+      "step": 9492
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.669491298576493e-05,
+      "loss": 0.2405,
+      "step": 9493
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.667497273072601e-05,
+      "loss": 0.2475,
+      "step": 9494
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.665503396616295e-05,
+      "loss": 0.2435,
+      "step": 9495
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.66350966929675e-05,
+      "loss": 0.2524,
+      "step": 9496
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.661516091203139e-05,
+      "loss": 0.2424,
+      "step": 9497
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.659522662424623e-05,
+      "loss": 0.2355,
+      "step": 9498
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.657529383050361e-05,
+      "loss": 0.2544,
+      "step": 9499
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.655536253169503e-05,
+      "loss": 0.2595,
+      "step": 9500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.653543272871191e-05,
+      "loss": 0.2485,
+      "step": 9501
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.651550442244564e-05,
+      "loss": 0.2539,
+      "step": 9502
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.649557761378751e-05,
+      "loss": 0.2767,
+      "step": 9503
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.647565230362874e-05,
+      "loss": 0.2574,
+      "step": 9504
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.645572849286053e-05,
+      "loss": 0.2524,
+      "step": 9505
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.643580618237397e-05,
+      "loss": 0.2379,
+      "step": 9506
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.641588537306011e-05,
+      "loss": 0.2515,
+      "step": 9507
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.639596606580987e-05,
+      "loss": 0.2388,
+      "step": 9508
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.637604826151417e-05,
+      "loss": 0.2287,
+      "step": 9509
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.635613196106386e-05,
+      "loss": 0.246,
+      "step": 9510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.633621716534968e-05,
+      "loss": 0.2545,
+      "step": 9511
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.631630387526235e-05,
+      "loss": 0.2535,
+      "step": 9512
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.629639209169248e-05,
+      "loss": 0.2373,
+      "step": 9513
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.627648181553064e-05,
+      "loss": 0.2466,
+      "step": 9514
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.625657304766734e-05,
+      "loss": 0.2486,
+      "step": 9515
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6236665788993e-05,
+      "loss": 0.2432,
+      "step": 9516
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.621676004039795e-05,
+      "loss": 0.2357,
+      "step": 9517
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.619685580277253e-05,
+      "loss": 0.24,
+      "step": 9518
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.617695307700695e-05,
+      "loss": 0.2377,
+      "step": 9519
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.615705186399134e-05,
+      "loss": 0.2415,
+      "step": 9520
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.613715216461584e-05,
+      "loss": 0.2326,
+      "step": 9521
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.611725397977045e-05,
+      "loss": 0.2519,
+      "step": 9522
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.60973573103451e-05,
+      "loss": 0.2616,
+      "step": 9523
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.607746215722969e-05,
+      "loss": 0.2551,
+      "step": 9524
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.605756852131406e-05,
+      "loss": 0.2448,
+      "step": 9525
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.603767640348795e-05,
+      "loss": 0.2307,
+      "step": 9526
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.601778580464106e-05,
+      "loss": 0.2598,
+      "step": 9527
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.599789672566296e-05,
+      "loss": 0.2652,
+      "step": 9528
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.597800916744322e-05,
+      "loss": 0.2498,
+      "step": 9529
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.595812313087132e-05,
+      "loss": 0.2457,
+      "step": 9530
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.593823861683669e-05,
+      "loss": 0.2483,
+      "step": 9531
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.591835562622866e-05,
+      "loss": 0.2396,
+      "step": 9532
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.589847415993649e-05,
+      "loss": 0.2357,
+      "step": 9533
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.587859421884939e-05,
+      "loss": 0.2455,
+      "step": 9534
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.585871580385654e-05,
+      "loss": 0.2354,
+      "step": 9535
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.583883891584696e-05,
+      "loss": 0.2602,
+      "step": 9536
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.581896355570967e-05,
+      "loss": 0.2337,
+      "step": 9537
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.579908972433362e-05,
+      "loss": 0.2466,
+      "step": 9538
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.577921742260764e-05,
+      "loss": 0.237,
+      "step": 9539
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.575934665142055e-05,
+      "loss": 0.2475,
+      "step": 9540
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.573947741166108e-05,
+      "loss": 0.2533,
+      "step": 9541
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.571960970421788e-05,
+      "loss": 0.2482,
+      "step": 9542
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.569974352997955e-05,
+      "loss": 0.2442,
+      "step": 9543
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.567987888983463e-05,
+      "loss": 0.2477,
+      "step": 9544
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.566001578467152e-05,
+      "loss": 0.2566,
+      "step": 9545
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.564015421537866e-05,
+      "loss": 0.2381,
+      "step": 9546
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.562029418284438e-05,
+      "loss": 0.2644,
+      "step": 9547
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.560043568795687e-05,
+      "loss": 0.2432,
+      "step": 9548
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.558057873160433e-05,
+      "loss": 0.2343,
+      "step": 9549
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.556072331467488e-05,
+      "loss": 0.2455,
+      "step": 9550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.554086943805656e-05,
+      "loss": 0.244,
+      "step": 9551
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.552101710263731e-05,
+      "loss": 0.2315,
+      "step": 9552
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.55011663093051e-05,
+      "loss": 0.248,
+      "step": 9553
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.548131705894772e-05,
+      "loss": 0.2357,
+      "step": 9554
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.546146935245294e-05,
+      "loss": 0.2189,
+      "step": 9555
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.544162319070848e-05,
+      "loss": 0.244,
+      "step": 9556
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.542177857460195e-05,
+      "loss": 0.2462,
+      "step": 9557
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.540193550502088e-05,
+      "loss": 0.2443,
+      "step": 9558
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.538209398285282e-05,
+      "loss": 0.2511,
+      "step": 9559
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.536225400898515e-05,
+      "loss": 0.2478,
+      "step": 9560
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.534241558430522e-05,
+      "loss": 0.2562,
+      "step": 9561
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.532257870970034e-05,
+      "loss": 0.2484,
+      "step": 9562
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.530274338605771e-05,
+      "loss": 0.256,
+      "step": 9563
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.528290961426445e-05,
+      "loss": 0.2815,
+      "step": 9564
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.526307739520768e-05,
+      "loss": 0.2401,
+      "step": 9565
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.524324672977435e-05,
+      "loss": 0.2484,
+      "step": 9566
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.522341761885147e-05,
+      "loss": 0.2486,
+      "step": 9567
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.520359006332582e-05,
+      "loss": 0.238,
+      "step": 9568
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.518376406408424e-05,
+      "loss": 0.227,
+      "step": 9569
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.516393962201344e-05,
+      "loss": 0.2527,
+      "step": 9570
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.514411673800007e-05,
+      "loss": 0.247,
+      "step": 9571
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.512429541293075e-05,
+      "loss": 0.2468,
+      "step": 9572
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.510447564769197e-05,
+      "loss": 0.2525,
+      "step": 9573
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.50846574431702e-05,
+      "loss": 0.2315,
+      "step": 9574
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.506484080025181e-05,
+      "loss": 0.2448,
+      "step": 9575
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.504502571982309e-05,
+      "loss": 0.2388,
+      "step": 9576
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.50252122027703e-05,
+      "loss": 0.2424,
+      "step": 9577
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.500540024997958e-05,
+      "loss": 0.2435,
+      "step": 9578
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.498558986233705e-05,
+      "loss": 0.2558,
+      "step": 9579
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.496578104072874e-05,
+      "loss": 0.2272,
+      "step": 9580
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.494597378604061e-05,
+      "loss": 0.248,
+      "step": 9581
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.492616809915853e-05,
+      "loss": 0.2468,
+      "step": 9582
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.490636398096832e-05,
+      "loss": 0.2374,
+      "step": 9583
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.488656143235574e-05,
+      "loss": 0.2311,
+      "step": 9584
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.486676045420647e-05,
+      "loss": 0.2439,
+      "step": 9585
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.48469610474061e-05,
+      "loss": 0.2476,
+      "step": 9586
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.482716321284021e-05,
+      "loss": 0.2597,
+      "step": 9587
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.48073669513942e-05,
+      "loss": 0.2498,
+      "step": 9588
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.478757226395351e-05,
+      "loss": 0.2431,
+      "step": 9589
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.476777915140344e-05,
+      "loss": 0.2519,
+      "step": 9590
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.474798761462926e-05,
+      "loss": 0.2427,
+      "step": 9591
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.472819765451615e-05,
+      "loss": 0.2371,
+      "step": 9592
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.470840927194923e-05,
+      "loss": 0.2316,
+      "step": 9593
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.468862246781356e-05,
+      "loss": 0.2547,
+      "step": 9594
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.466883724299408e-05,
+      "loss": 0.2427,
+      "step": 9595
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.46490535983757e-05,
+      "loss": 0.2377,
+      "step": 9596
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.462927153484327e-05,
+      "loss": 0.2577,
+      "step": 9597
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.460949105328154e-05,
+      "loss": 0.2484,
+      "step": 9598
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.45897121545752e-05,
+      "loss": 0.2502,
+      "step": 9599
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.456993483960889e-05,
+      "loss": 0.2326,
+      "step": 9600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.455015910926711e-05,
+      "loss": 0.2333,
+      "step": 9601
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.453038496443439e-05,
+      "loss": 0.2502,
+      "step": 9602
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.45106124059951e-05,
+      "loss": 0.2402,
+      "step": 9603
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.44908414348336e-05,
+      "loss": 0.2536,
+      "step": 9604
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.447107205183413e-05,
+      "loss": 0.2477,
+      "step": 9605
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.445130425788089e-05,
+      "loss": 0.2471,
+      "step": 9606
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.443153805385805e-05,
+      "loss": 0.2415,
+      "step": 9607
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.44117734406496e-05,
+      "loss": 0.2547,
+      "step": 9608
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.439201041913955e-05,
+      "loss": 0.2595,
+      "step": 9609
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.437224899021179e-05,
+      "loss": 0.2564,
+      "step": 9610
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.435248915475019e-05,
+      "loss": 0.2544,
+      "step": 9611
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.43327309136385e-05,
+      "loss": 0.2381,
+      "step": 9612
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.431297426776041e-05,
+      "loss": 0.2467,
+      "step": 9613
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.429321921799955e-05,
+      "loss": 0.2404,
+      "step": 9614
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.427346576523946e-05,
+      "loss": 0.259,
+      "step": 9615
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.425371391036363e-05,
+      "loss": 0.2371,
+      "step": 9616
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.423396365425549e-05,
+      "loss": 0.2495,
+      "step": 9617
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.421421499779837e-05,
+      "loss": 0.2446,
+      "step": 9618
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.419446794187549e-05,
+      "loss": 0.2509,
+      "step": 9619
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.417472248737011e-05,
+      "loss": 0.2368,
+      "step": 9620
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.415497863516533e-05,
+      "loss": 0.2552,
+      "step": 9621
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.413523638614419e-05,
+      "loss": 0.2489,
+      "step": 9622
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.41154957411897e-05,
+      "loss": 0.2557,
+      "step": 9623
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.409575670118472e-05,
+      "loss": 0.2417,
+      "step": 9624
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.407601926701213e-05,
+      "loss": 0.2453,
+      "step": 9625
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.405628343955468e-05,
+      "loss": 0.2644,
+      "step": 9626
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.403654921969511e-05,
+      "loss": 0.2407,
+      "step": 9627
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.401681660831594e-05,
+      "loss": 0.2411,
+      "step": 9628
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.399708560629978e-05,
+      "loss": 0.2497,
+      "step": 9629
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.397735621452912e-05,
+      "loss": 0.257,
+      "step": 9630
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.395762843388633e-05,
+      "loss": 0.2551,
+      "step": 9631
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.393790226525376e-05,
+      "loss": 0.2582,
+      "step": 9632
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.391817770951367e-05,
+      "loss": 0.2432,
+      "step": 9633
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.389845476754825e-05,
+      "loss": 0.2426,
+      "step": 9634
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.387873344023964e-05,
+      "loss": 0.2513,
+      "step": 9635
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.385901372846984e-05,
+      "loss": 0.2305,
+      "step": 9636
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.383929563312086e-05,
+      "loss": 0.2392,
+      "step": 9637
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.381957915507457e-05,
+      "loss": 0.2374,
+      "step": 9638
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.379986429521282e-05,
+      "loss": 0.2437,
+      "step": 9639
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.378015105441736e-05,
+      "loss": 0.2486,
+      "step": 9640
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.376043943356987e-05,
+      "loss": 0.2567,
+      "step": 9641
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.374072943355195e-05,
+      "loss": 0.2626,
+      "step": 9642
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.372102105524516e-05,
+      "loss": 0.2762,
+      "step": 9643
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.370131429953097e-05,
+      "loss": 0.2597,
+      "step": 9644
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.368160916729074e-05,
+      "loss": 0.2617,
+      "step": 9645
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.366190565940581e-05,
+      "loss": 0.2637,
+      "step": 9646
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.364220377675747e-05,
+      "loss": 0.2549,
+      "step": 9647
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.362250352022681e-05,
+      "loss": 0.2611,
+      "step": 9648
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.360280489069497e-05,
+      "loss": 0.2588,
+      "step": 9649
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.358310788904299e-05,
+      "loss": 0.2483,
+      "step": 9650
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.356341251615181e-05,
+      "loss": 0.2563,
+      "step": 9651
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.354371877290233e-05,
+      "loss": 0.2436,
+      "step": 9652
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.352402666017534e-05,
+      "loss": 0.2412,
+      "step": 9653
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.350433617885161e-05,
+      "loss": 0.2514,
+      "step": 9654
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.348464732981179e-05,
+      "loss": 0.256,
+      "step": 9655
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.346496011393646e-05,
+      "loss": 0.2417,
+      "step": 9656
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.344527453210614e-05,
+      "loss": 0.2662,
+      "step": 9657
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.34255905852013e-05,
+      "loss": 0.2558,
+      "step": 9658
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.340590827410229e-05,
+      "loss": 0.2533,
+      "step": 9659
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.338622759968941e-05,
+      "loss": 0.2346,
+      "step": 9660
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.336654856284291e-05,
+      "loss": 0.2434,
+      "step": 9661
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.334687116444291e-05,
+      "loss": 0.2394,
+      "step": 9662
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.332719540536952e-05,
+      "loss": 0.2473,
+      "step": 9663
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.330752128650273e-05,
+      "loss": 0.2427,
+      "step": 9664
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.328784880872247e-05,
+      "loss": 0.2278,
+      "step": 9665
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.326817797290863e-05,
+      "loss": 0.234,
+      "step": 9666
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.324850877994098e-05,
+      "loss": 0.2309,
+      "step": 9667
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.32288412306992e-05,
+      "loss": 0.2493,
+      "step": 9668
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.320917532606294e-05,
+      "loss": 0.2456,
+      "step": 9669
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31895110669118e-05,
+      "loss": 0.2357,
+      "step": 9670
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.316984845412524e-05,
+      "loss": 0.2326,
+      "step": 9671
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31501874885827e-05,
+      "loss": 0.2624,
+      "step": 9672
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31305281711635e-05,
+      "loss": 0.2443,
+      "step": 9673
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.311087050274695e-05,
+      "loss": 0.2499,
+      "step": 9674
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.30912144842122e-05,
+      "loss": 0.2524,
+      "step": 9675
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.307156011643843e-05,
+      "loss": 0.2621,
+      "step": 9676
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.305190740030464e-05,
+      "loss": 0.2318,
+      "step": 9677
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.303225633668986e-05,
+      "loss": 0.2486,
+      "step": 9678
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.30126069264729e-05,
+      "loss": 0.2502,
+      "step": 9679
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.299295917053268e-05,
+      "loss": 0.247,
+      "step": 9680
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.297331306974792e-05,
+      "loss": 0.2382,
+      "step": 9681
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.295366862499728e-05,
+      "loss": 0.2473,
+      "step": 9682
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.293402583715939e-05,
+      "loss": 0.2481,
+      "step": 9683
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.291438470711278e-05,
+      "loss": 0.241,
+      "step": 9684
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.289474523573591e-05,
+      "loss": 0.2348,
+      "step": 9685
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.287510742390714e-05,
+      "loss": 0.2547,
+      "step": 9686
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.285547127250485e-05,
+      "loss": 0.2472,
+      "step": 9687
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.283583678240718e-05,
+      "loss": 0.2294,
+      "step": 9688
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.281620395449234e-05,
+      "loss": 0.249,
+      "step": 9689
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.279657278963839e-05,
+      "loss": 0.2504,
+      "step": 9690
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.277694328872341e-05,
+      "loss": 0.2434,
+      "step": 9691
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.275731545262527e-05,
+      "loss": 0.247,
+      "step": 9692
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.273768928222184e-05,
+      "loss": 0.2515,
+      "step": 9693
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.271806477839092e-05,
+      "loss": 0.2401,
+      "step": 9694
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.269844194201024e-05,
+      "loss": 0.243,
+      "step": 9695
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.267882077395741e-05,
+      "loss": 0.2648,
+      "step": 9696
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.265920127511002e-05,
+      "loss": 0.2358,
+      "step": 9697
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.263958344634557e-05,
+      "loss": 0.2414,
+      "step": 9698
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.261996728854142e-05,
+      "loss": 0.239,
+      "step": 9699
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.260035280257499e-05,
+      "loss": 0.2405,
+      "step": 9700
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.258073998932347e-05,
+      "loss": 0.2495,
+      "step": 9701
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.256112884966409e-05,
+      "loss": 0.2516,
+      "step": 9702
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.254151938447397e-05,
+      "loss": 0.2376,
+      "step": 9703
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.252191159463015e-05,
+      "loss": 0.237,
+      "step": 9704
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.250230548100957e-05,
+      "loss": 0.2444,
+      "step": 9705
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.248270104448915e-05,
+      "loss": 0.2386,
+      "step": 9706
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.246309828594575e-05,
+      "loss": 0.2377,
+      "step": 9707
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2443497206256e-05,
+      "loss": 0.2337,
+      "step": 9708
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.242389780629664e-05,
+      "loss": 0.2338,
+      "step": 9709
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.240430008694425e-05,
+      "loss": 0.2375,
+      "step": 9710
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.238470404907532e-05,
+      "loss": 0.2366,
+      "step": 9711
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.236510969356635e-05,
+      "loss": 0.2305,
+      "step": 9712
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.234551702129364e-05,
+      "loss": 0.2336,
+      "step": 9713
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.232592603313353e-05,
+      "loss": 0.2458,
+      "step": 9714
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.230633672996221e-05,
+      "loss": 0.2329,
+      "step": 9715
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.228674911265582e-05,
+      "loss": 0.2349,
+      "step": 9716
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.226716318209044e-05,
+      "loss": 0.2487,
+      "step": 9717
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.224757893914206e-05,
+      "loss": 0.2341,
+      "step": 9718
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.222799638468658e-05,
+      "loss": 0.2293,
+      "step": 9719
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.220841551959982e-05,
+      "loss": 0.2518,
+      "step": 9720
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.218883634475759e-05,
+      "loss": 0.2343,
+      "step": 9721
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.216925886103553e-05,
+      "loss": 0.2359,
+      "step": 9722
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.214968306930928e-05,
+      "loss": 0.2538,
+      "step": 9723
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.213010897045435e-05,
+      "loss": 0.2249,
+      "step": 9724
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.211053656534624e-05,
+      "loss": 0.2452,
+      "step": 9725
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.209096585486031e-05,
+      "loss": 0.2459,
+      "step": 9726
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.20713968398719e-05,
+      "loss": 0.2355,
+      "step": 9727
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.205182952125618e-05,
+      "loss": 0.24,
+      "step": 9728
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.203226389988832e-05,
+      "loss": 0.234,
+      "step": 9729
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.201269997664344e-05,
+      "loss": 0.2372,
+      "step": 9730
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.199313775239653e-05,
+      "loss": 0.2495,
+      "step": 9731
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.197357722802249e-05,
+      "loss": 0.252,
+      "step": 9732
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.19540184043962e-05,
+      "loss": 0.244,
+      "step": 9733
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.193446128239243e-05,
+      "loss": 0.2352,
+      "step": 9734
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.191490586288589e-05,
+      "loss": 0.2331,
+      "step": 9735
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.18953521467512e-05,
+      "loss": 0.2277,
+      "step": 9736
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.187580013486289e-05,
+      "loss": 0.2532,
+      "step": 9737
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.185624982809545e-05,
+      "loss": 0.232,
+      "step": 9738
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.183670122732327e-05,
+      "loss": 0.2559,
+      "step": 9739
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.181715433342069e-05,
+      "loss": 0.2401,
+      "step": 9740
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.179760914726192e-05,
+      "loss": 0.2568,
+      "step": 9741
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.177806566972114e-05,
+      "loss": 0.2506,
+      "step": 9742
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.175852390167245e-05,
+      "loss": 0.2571,
+      "step": 9743
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.173898384398988e-05,
+      "loss": 0.2403,
+      "step": 9744
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.171944549754731e-05,
+      "loss": 0.2502,
+      "step": 9745
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.169990886321863e-05,
+      "loss": 0.2498,
+      "step": 9746
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.168037394187766e-05,
+      "loss": 0.2431,
+      "step": 9747
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.166084073439805e-05,
+      "loss": 0.2374,
+      "step": 9748
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.164130924165346e-05,
+      "loss": 0.2573,
+      "step": 9749
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.162177946451744e-05,
+      "loss": 0.2418,
+      "step": 9750
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.160225140386345e-05,
+      "loss": 0.2507,
+      "step": 9751
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.158272506056493e-05,
+      "loss": 0.2374,
+      "step": 9752
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.156320043549517e-05,
+      "loss": 0.2512,
+      "step": 9753
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.154367752952743e-05,
+      "loss": 0.2426,
+      "step": 9754
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.15241563435349e-05,
+      "loss": 0.2358,
+      "step": 9755
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.150463687839065e-05,
+      "loss": 0.2428,
+      "step": 9756
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.148511913496768e-05,
+      "loss": 0.2296,
+      "step": 9757
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.146560311413896e-05,
+      "loss": 0.2363,
+      "step": 9758
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.144608881677733e-05,
+      "loss": 0.2239,
+      "step": 9759
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.142657624375561e-05,
+      "loss": 0.2445,
+      "step": 9760
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.140706539594646e-05,
+      "loss": 0.2433,
+      "step": 9761
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.138755627422254e-05,
+      "loss": 0.2427,
+      "step": 9762
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.13680488794564e-05,
+      "loss": 0.2637,
+      "step": 9763
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.134854321252052e-05,
+      "loss": 0.2259,
+      "step": 9764
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.132903927428728e-05,
+      "loss": 0.2414,
+      "step": 9765
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.130953706562904e-05,
+      "loss": 0.2413,
+      "step": 9766
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.129003658741801e-05,
+      "loss": 0.2368,
+      "step": 9767
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.127053784052639e-05,
+      "loss": 0.2371,
+      "step": 9768
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.125104082582623e-05,
+      "loss": 0.2389,
+      "step": 9769
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.123154554418953e-05,
+      "loss": 0.2498,
+      "step": 9770
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.121205199648827e-05,
+      "loss": 0.2316,
+      "step": 9771
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.119256018359429e-05,
+      "loss": 0.2331,
+      "step": 9772
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.117307010637938e-05,
+      "loss": 0.2329,
+      "step": 9773
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.115358176571523e-05,
+      "loss": 0.2497,
+      "step": 9774
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.113409516247346e-05,
+      "loss": 0.2471,
+      "step": 9775
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.111461029752564e-05,
+      "loss": 0.2511,
+      "step": 9776
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.10951271717432e-05,
+      "loss": 0.2331,
+      "step": 9777
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.107564578599756e-05,
+      "loss": 0.2374,
+      "step": 9778
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.105616614116003e-05,
+      "loss": 0.2455,
+      "step": 9779
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.103668823810185e-05,
+      "loss": 0.2591,
+      "step": 9780
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.101721207769415e-05,
+      "loss": 0.2513,
+      "step": 9781
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.099773766080804e-05,
+      "loss": 0.246,
+      "step": 9782
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0978264988314506e-05,
+      "loss": 0.2536,
+      "step": 9783
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0958794061084466e-05,
+      "loss": 0.2424,
+      "step": 9784
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.093932487998879e-05,
+      "loss": 0.2337,
+      "step": 9785
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.091985744589823e-05,
+      "loss": 0.2471,
+      "step": 9786
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0900391759683514e-05,
+      "loss": 0.2469,
+      "step": 9787
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0880927822215176e-05,
+      "loss": 0.2326,
+      "step": 9788
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.086146563436377e-05,
+      "loss": 0.2465,
+      "step": 9789
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0842005196999784e-05,
+      "loss": 0.2361,
+      "step": 9790
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.082254651099357e-05,
+      "loss": 0.2478,
+      "step": 9791
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.080308957721543e-05,
+      "loss": 0.2493,
+      "step": 9792
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0783634396535606e-05,
+      "loss": 0.2346,
+      "step": 9793
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0764180969824205e-05,
+      "loss": 0.2198,
+      "step": 9794
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.074472929795131e-05,
+      "loss": 0.2245,
+      "step": 9795
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.072527938178689e-05,
+      "loss": 0.2597,
+      "step": 9796
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.070583122220088e-05,
+      "loss": 0.2473,
+      "step": 9797
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.068638482006308e-05,
+      "loss": 0.2483,
+      "step": 9798
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.066694017624324e-05,
+      "loss": 0.2513,
+      "step": 9799
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.064749729161103e-05,
+      "loss": 0.2329,
+      "step": 9800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0628056167036066e-05,
+      "loss": 0.2522,
+      "step": 9801
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.060861680338783e-05,
+      "loss": 0.2551,
+      "step": 9802
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.058917920153577e-05,
+      "loss": 0.2562,
+      "step": 9803
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.056974336234924e-05,
+      "loss": 0.2543,
+      "step": 9804
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.055030928669752e-05,
+      "loss": 0.2544,
+      "step": 9805
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.053087697544979e-05,
+      "loss": 0.2431,
+      "step": 9806
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.051144642947524e-05,
+      "loss": 0.2343,
+      "step": 9807
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.04920176496428e-05,
+      "loss": 0.2448,
+      "step": 9808
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.047259063682147e-05,
+      "loss": 0.2306,
+      "step": 9809
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.045316539188015e-05,
+      "loss": 0.2424,
+      "step": 9810
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0433741915687627e-05,
+      "loss": 0.2368,
+      "step": 9811
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.041432020911264e-05,
+      "loss": 0.2387,
+      "step": 9812
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.039490027302381e-05,
+      "loss": 0.24,
+      "step": 9813
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0375482108289736e-05,
+      "loss": 0.2567,
+      "step": 9814
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.035606571577888e-05,
+      "loss": 0.2434,
+      "step": 9815
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.033665109635964e-05,
+      "loss": 0.2524,
+      "step": 9816
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.031723825090036e-05,
+      "loss": 0.2573,
+      "step": 9817
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.029782718026927e-05,
+      "loss": 0.2701,
+      "step": 9818
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.027841788533457e-05,
+      "loss": 0.2546,
+      "step": 9819
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0259010366964306e-05,
+      "loss": 0.2715,
+      "step": 9820
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0239604626026536e-05,
+      "loss": 0.2715,
+      "step": 9821
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.022020066338916e-05,
+      "loss": 0.2433,
+      "step": 9822
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.020079847992004e-05,
+      "loss": 0.2606,
+      "step": 9823
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.018139807648692e-05,
+      "loss": 0.2472,
+      "step": 9824
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0161999453957504e-05,
+      "loss": 0.2597,
+      "step": 9825
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.014260261319944e-05,
+      "loss": 0.2402,
+      "step": 9826
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.012320755508023e-05,
+      "loss": 0.2511,
+      "step": 9827
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.010381428046731e-05,
+      "loss": 0.2819,
+      "step": 9828
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0084422790228055e-05,
+      "loss": 0.2451,
+      "step": 9829
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0065033085229795e-05,
+      "loss": 0.2548,
+      "step": 9830
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.00456451663397e-05,
+      "loss": 0.2392,
+      "step": 9831
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0026259034424936e-05,
+      "loss": 0.2468,
+      "step": 9832
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.000687469035253e-05,
+      "loss": 0.2527,
+      "step": 9833
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.998749213498951e-05,
+      "loss": 0.2483,
+      "step": 9834
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9968111369202684e-05,
+      "loss": 0.2454,
+      "step": 9835
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.994873239385891e-05,
+      "loss": 0.2558,
+      "step": 9836
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.992935520982492e-05,
+      "loss": 0.2394,
+      "step": 9837
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990997981796737e-05,
+      "loss": 0.2383,
+      "step": 9838
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.989060621915281e-05,
+      "loss": 0.2557,
+      "step": 9839
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9871234414247755e-05,
+      "loss": 0.2582,
+      "step": 9840
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.985186440411863e-05,
+      "loss": 0.2692,
+      "step": 9841
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9832496189631726e-05,
+      "loss": 0.2491,
+      "step": 9842
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.981312977165332e-05,
+      "loss": 0.2473,
+      "step": 9843
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.97937651510496e-05,
+      "loss": 0.2437,
+      "step": 9844
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.977440232868661e-05,
+      "loss": 0.2503,
+      "step": 9845
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9755041305430396e-05,
+      "loss": 0.2672,
+      "step": 9846
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9735682082146924e-05,
+      "loss": 0.2314,
+      "step": 9847
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.971632465970195e-05,
+      "loss": 0.2394,
+      "step": 9848
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.969696903896129e-05,
+      "loss": 0.2463,
+      "step": 9849
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.967761522079065e-05,
+      "loss": 0.252,
+      "step": 9850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.965826320605561e-05,
+      "loss": 0.2428,
+      "step": 9851
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.96389129956217e-05,
+      "loss": 0.2512,
+      "step": 9852
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.961956459035438e-05,
+      "loss": 0.2181,
+      "step": 9853
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.960021799111901e-05,
+      "loss": 0.2534,
+      "step": 9854
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9580873198780876e-05,
+      "loss": 0.2452,
+      "step": 9855
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9561530214205184e-05,
+      "loss": 0.2378,
+      "step": 9856
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.954218903825706e-05,
+      "loss": 0.2397,
+      "step": 9857
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.952284967180153e-05,
+      "loss": 0.2385,
+      "step": 9858
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.950351211570355e-05,
+      "loss": 0.2475,
+      "step": 9859
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.948417637082804e-05,
+      "loss": 0.2277,
+      "step": 9860
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9464842438039757e-05,
+      "loss": 0.2448,
+      "step": 9861
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9445510318203444e-05,
+      "loss": 0.2479,
+      "step": 9862
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.942618001218374e-05,
+      "loss": 0.2631,
+      "step": 9863
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.940685152084519e-05,
+      "loss": 0.2462,
+      "step": 9864
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9387524845052265e-05,
+      "loss": 0.2589,
+      "step": 9865
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.936819998566936e-05,
+      "loss": 0.2499,
+      "step": 9866
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.934887694356084e-05,
+      "loss": 0.2368,
+      "step": 9867
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9329555719590856e-05,
+      "loss": 0.2356,
+      "step": 9868
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.931023631462358e-05,
+      "loss": 0.2496,
+      "step": 9869
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.929091872952309e-05,
+      "loss": 0.2527,
+      "step": 9870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.927160296515338e-05,
+      "loss": 0.2322,
+      "step": 9871
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.925228902237835e-05,
+      "loss": 0.247,
+      "step": 9872
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.923297690206182e-05,
+      "loss": 0.2341,
+      "step": 9873
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.921366660506754e-05,
+      "loss": 0.2506,
+      "step": 9874
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.919435813225916e-05,
+      "loss": 0.2373,
+      "step": 9875
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.917505148450028e-05,
+      "loss": 0.254,
+      "step": 9876
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.915574666265438e-05,
+      "loss": 0.2542,
+      "step": 9877
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.913644366758489e-05,
+      "loss": 0.234,
+      "step": 9878
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.911714250015512e-05,
+      "loss": 0.234,
+      "step": 9879
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.909784316122834e-05,
+      "loss": 0.2519,
+      "step": 9880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.907854565166773e-05,
+      "loss": 0.2428,
+      "step": 9881
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.905924997233636e-05,
+      "loss": 0.2492,
+      "step": 9882
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.903995612409725e-05,
+      "loss": 0.2279,
+      "step": 9883
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9020664107813325e-05,
+      "loss": 0.2429,
+      "step": 9884
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9001373924347425e-05,
+      "loss": 0.2409,
+      "step": 9885
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.898208557456231e-05,
+      "loss": 0.2494,
+      "step": 9886
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8962799059320714e-05,
+      "loss": 0.2438,
+      "step": 9887
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8943514379485133e-05,
+      "loss": 0.2422,
+      "step": 9888
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.892423153591813e-05,
+      "loss": 0.2455,
+      "step": 9889
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.890495052948213e-05,
+      "loss": 0.2439,
+      "step": 9890
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8885671361039505e-05,
+      "loss": 0.2477,
+      "step": 9891
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.886639403145252e-05,
+      "loss": 0.2304,
+      "step": 9892
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.884711854158334e-05,
+      "loss": 0.2644,
+      "step": 9893
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.882784489229409e-05,
+      "loss": 0.2367,
+      "step": 9894
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.88085730844468e-05,
+      "loss": 0.2521,
+      "step": 9895
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.878930311890338e-05,
+      "loss": 0.2375,
+      "step": 9896
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.877003499652571e-05,
+      "loss": 0.2589,
+      "step": 9897
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.875076871817555e-05,
+      "loss": 0.2463,
+      "step": 9898
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.873150428471461e-05,
+      "loss": 0.2326,
+      "step": 9899
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.871224169700451e-05,
+      "loss": 0.2425,
+      "step": 9900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.869298095590673e-05,
+      "loss": 0.2435,
+      "step": 9901
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8673722062282746e-05,
+      "loss": 0.2414,
+      "step": 9902
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8654465016993924e-05,
+      "loss": 0.2328,
+      "step": 9903
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8635209820901515e-05,
+      "loss": 0.2422,
+      "step": 9904
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.861595647486675e-05,
+      "loss": 0.2271,
+      "step": 9905
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.859670497975074e-05,
+      "loss": 0.2389,
+      "step": 9906
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8577455336414513e-05,
+      "loss": 0.2462,
+      "step": 9907
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.855820754571899e-05,
+      "loss": 0.2587,
+      "step": 9908
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8538961608525054e-05,
+      "loss": 0.2483,
+      "step": 9909
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.85197175256935e-05,
+      "loss": 0.2521,
+      "step": 9910
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.850047529808502e-05,
+      "loss": 0.2332,
+      "step": 9911
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.848123492656024e-05,
+      "loss": 0.2583,
+      "step": 9912
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.846199641197967e-05,
+      "loss": 0.2409,
+      "step": 9913
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.844275975520377e-05,
+      "loss": 0.2341,
+      "step": 9914
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.842352495709291e-05,
+      "loss": 0.26,
+      "step": 9915
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.840429201850737e-05,
+      "loss": 0.2399,
+      "step": 9916
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8385060940307357e-05,
+      "loss": 0.2435,
+      "step": 9917
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.836583172335298e-05,
+      "loss": 0.2397,
+      "step": 9918
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.834660436850432e-05,
+      "loss": 0.2413,
+      "step": 9919
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8327378876621244e-05,
+      "loss": 0.2275,
+      "step": 9920
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.83081552485637e-05,
+      "loss": 0.2245,
+      "step": 9921
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.828893348519142e-05,
+      "loss": 0.2533,
+      "step": 9922
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.826971358736415e-05,
+      "loss": 0.2526,
+      "step": 9923
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8250495555941456e-05,
+      "loss": 0.2471,
+      "step": 9924
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.823127939178292e-05,
+      "loss": 0.2319,
+      "step": 9925
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.821206509574799e-05,
+      "loss": 0.2462,
+      "step": 9926
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.819285266869602e-05,
+      "loss": 0.2357,
+      "step": 9927
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8173642111486236e-05,
+      "loss": 0.2383,
+      "step": 9928
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.815443342497795e-05,
+      "loss": 0.2253,
+      "step": 9929
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.813522661003017e-05,
+      "loss": 0.2492,
+      "step": 9930
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8116021667502005e-05,
+      "loss": 0.2305,
+      "step": 9931
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8096818598252356e-05,
+      "loss": 0.2355,
+      "step": 9932
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.807761740314015e-05,
+      "loss": 0.2374,
+      "step": 9933
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8058418083024105e-05,
+      "loss": 0.2304,
+      "step": 9934
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.803922063876296e-05,
+      "loss": 0.2253,
+      "step": 9935
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.802002507121529e-05,
+      "loss": 0.2458,
+      "step": 9936
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8000831381239664e-05,
+      "loss": 0.2416,
+      "step": 9937
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.798163956969449e-05,
+      "loss": 0.2462,
+      "step": 9938
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.796244963743819e-05,
+      "loss": 0.2553,
+      "step": 9939
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.794326158532899e-05,
+      "loss": 0.238,
+      "step": 9940
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7924075414225066e-05,
+      "loss": 0.2468,
+      "step": 9941
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.790489112498461e-05,
+      "loss": 0.2394,
+      "step": 9942
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.788570871846554e-05,
+      "loss": 0.2322,
+      "step": 9943
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7866528195525896e-05,
+      "loss": 0.2427,
+      "step": 9944
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.784734955702345e-05,
+      "loss": 0.2287,
+      "step": 9945
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.782817280381605e-05,
+      "loss": 0.2414,
+      "step": 9946
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.780899793676136e-05,
+      "loss": 0.2344,
+      "step": 9947
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7789824956716944e-05,
+      "loss": 0.2301,
+      "step": 9948
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.777065386454037e-05,
+      "loss": 0.2444,
+      "step": 9949
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.775148466108903e-05,
+      "loss": 0.2299,
+      "step": 9950
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7732317347220336e-05,
+      "loss": 0.2377,
+      "step": 9951
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.771315192379152e-05,
+      "loss": 0.2468,
+      "step": 9952
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7693988391659714e-05,
+      "loss": 0.2399,
+      "step": 9953
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.767482675168212e-05,
+      "loss": 0.2365,
+      "step": 9954
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.765566700471563e-05,
+      "loss": 0.2391,
+      "step": 9955
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.763650915161729e-05,
+      "loss": 0.2426,
+      "step": 9956
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.761735319324385e-05,
+      "loss": 0.2422,
+      "step": 9957
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7598199130452145e-05,
+      "loss": 0.2293,
+      "step": 9958
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.757904696409876e-05,
+      "loss": 0.2407,
+      "step": 9959
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7559896695040386e-05,
+      "loss": 0.2445,
+      "step": 9960
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.754074832413343e-05,
+      "loss": 0.2369,
+      "step": 9961
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7521601852234385e-05,
+      "loss": 0.2345,
+      "step": 9962
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.750245728019952e-05,
+      "loss": 0.2304,
+      "step": 9963
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7483314608885166e-05,
+      "loss": 0.2476,
+      "step": 9964
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.746417383914739e-05,
+      "loss": 0.2405,
+      "step": 9965
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.744503497184238e-05,
+      "loss": 0.2508,
+      "step": 9966
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.742589800782606e-05,
+      "loss": 0.2358,
+      "step": 9967
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.740676294795435e-05,
+      "loss": 0.2259,
+      "step": 9968
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.738762979308303e-05,
+      "loss": 0.2376,
+      "step": 9969
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.736849854406793e-05,
+      "loss": 0.2278,
+      "step": 9970
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7349369201764616e-05,
+      "loss": 0.237,
+      "step": 9971
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.733024176702873e-05,
+      "loss": 0.2306,
+      "step": 9972
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7311116240715676e-05,
+      "loss": 0.2364,
+      "step": 9973
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7291992623680935e-05,
+      "loss": 0.2407,
+      "step": 9974
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7272870916779744e-05,
+      "loss": 0.268,
+      "step": 9975
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.72537511208674e-05,
+      "loss": 0.2422,
+      "step": 9976
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.723463323679896e-05,
+      "loss": 0.2515,
+      "step": 9977
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.721551726542957e-05,
+      "loss": 0.2455,
+      "step": 9978
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.719640320761413e-05,
+      "loss": 0.2419,
+      "step": 9979
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.717729106420759e-05,
+      "loss": 0.2493,
+      "step": 9980
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7158180836064666e-05,
+      "loss": 0.242,
+      "step": 9981
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.713907252404016e-05,
+      "loss": 0.2505,
+      "step": 9982
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.71199661289886e-05,
+      "loss": 0.2393,
+      "step": 9983
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.710086165176463e-05,
+      "loss": 0.2374,
+      "step": 9984
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.708175909322262e-05,
+      "loss": 0.2439,
+      "step": 9985
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.706265845421701e-05,
+      "loss": 0.232,
+      "step": 9986
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.704355973560207e-05,
+      "loss": 0.2345,
+      "step": 9987
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.702446293823196e-05,
+      "loss": 0.2292,
+      "step": 9988
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.700536806296077e-05,
+      "loss": 0.2419,
+      "step": 9989
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.698627511064263e-05,
+      "loss": 0.2449,
+      "step": 9990
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6967184082131366e-05,
+      "loss": 0.2344,
+      "step": 9991
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.694809497828093e-05,
+      "loss": 0.2443,
+      "step": 9992
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.692900779994499e-05,
+      "loss": 0.22,
+      "step": 9993
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6909922547977336e-05,
+      "loss": 0.2328,
+      "step": 9994
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6890839223231475e-05,
+      "loss": 0.243,
+      "step": 9995
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6871757826560976e-05,
+      "loss": 0.2347,
+      "step": 9996
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.685267835881922e-05,
+      "loss": 0.2395,
+      "step": 9997
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.68336008208596e-05,
+      "loss": 0.2336,
+      "step": 9998
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6814525213535276e-05,
+      "loss": 0.2462,
+      "step": 9999
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.679545153769952e-05,
+      "loss": 0.2253,
+      "step": 10000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6776379794205315e-05,
+      "loss": 0.2457,
+      "step": 10001
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.675730998390574e-05,
+      "loss": 0.2505,
+      "step": 10002
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6738242107653605e-05,
+      "loss": 0.2462,
+      "step": 10003
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6719176166301826e-05,
+      "loss": 0.2368,
+      "step": 10004
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.670011216070308e-05,
+      "loss": 0.2347,
+      "step": 10005
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6681050091709983e-05,
+      "loss": 0.242,
+      "step": 10006
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.666198996017519e-05,
+      "loss": 0.247,
+      "step": 10007
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.66429317669511e-05,
+      "loss": 0.2473,
+      "step": 10008
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.662387551289008e-05,
+      "loss": 0.2326,
+      "step": 10009
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6604821198844496e-05,
+      "loss": 0.2403,
+      "step": 10010
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.658576882566648e-05,
+      "loss": 0.2296,
+      "step": 10011
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6566718394208264e-05,
+      "loss": 0.2345,
+      "step": 10012
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6547669905321786e-05,
+      "loss": 0.2379,
+      "step": 10013
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.652862335985907e-05,
+      "loss": 0.2471,
+      "step": 10014
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.650957875867191e-05,
+      "loss": 0.2401,
+      "step": 10015
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.649053610261217e-05,
+      "loss": 0.2336,
+      "step": 10016
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.647149539253149e-05,
+      "loss": 0.2416,
+      "step": 10017
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.645245662928144e-05,
+      "loss": 0.2464,
+      "step": 10018
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.643341981371362e-05,
+      "loss": 0.2376,
+      "step": 10019
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6414384946679356e-05,
+      "loss": 0.2382,
+      "step": 10020
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6395352029030104e-05,
+      "loss": 0.2567,
+      "step": 10021
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.637632106161702e-05,
+      "loss": 0.2355,
+      "step": 10022
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6357292045291355e-05,
+      "loss": 0.2379,
+      "step": 10023
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.633826498090411e-05,
+      "loss": 0.2429,
+      "step": 10024
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6319239869306364e-05,
+      "loss": 0.2476,
+      "step": 10025
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.630021671134893e-05,
+      "loss": 0.2459,
+      "step": 10026
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6281195507882736e-05,
+      "loss": 0.2408,
+      "step": 10027
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.626217625975843e-05,
+      "loss": 0.2376,
+      "step": 10028
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.624315896782668e-05,
+      "loss": 0.2439,
+      "step": 10029
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.622414363293802e-05,
+      "loss": 0.2361,
+      "step": 10030
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.620513025594296e-05,
+      "loss": 0.2578,
+      "step": 10031
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6186118837691824e-05,
+      "loss": 0.2382,
+      "step": 10032
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6167109379035e-05,
+      "loss": 0.2454,
+      "step": 10033
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.614810188082258e-05,
+      "loss": 0.2433,
+      "step": 10034
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6129096343904775e-05,
+      "loss": 0.231,
+      "step": 10035
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.611009276913155e-05,
+      "loss": 0.2522,
+      "step": 10036
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6091091157352915e-05,
+      "loss": 0.2384,
+      "step": 10037
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.607209150941864e-05,
+      "loss": 0.2369,
+      "step": 10038
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.605309382617859e-05,
+      "loss": 0.2432,
+      "step": 10039
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.603409810848235e-05,
+      "loss": 0.2529,
+      "step": 10040
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.601510435717961e-05,
+      "loss": 0.2441,
+      "step": 10041
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5996112573119755e-05,
+      "loss": 0.2515,
+      "step": 10042
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.597712275715231e-05,
+      "loss": 0.2368,
+      "step": 10043
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.595813491012652e-05,
+      "loss": 0.2324,
+      "step": 10044
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.59391490328917e-05,
+      "loss": 0.2636,
+      "step": 10045
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5920165126296934e-05,
+      "loss": 0.2503,
+      "step": 10046
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.590118319119139e-05,
+      "loss": 0.2466,
+      "step": 10047
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.588220322842388e-05,
+      "loss": 0.2475,
+      "step": 10048
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.586322523884342e-05,
+      "loss": 0.2586,
+      "step": 10049
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5844249223298726e-05,
+      "loss": 0.2448,
+      "step": 10050
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.582527518263859e-05,
+      "loss": 0.254,
+      "step": 10051
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5806303117711555e-05,
+      "loss": 0.2402,
+      "step": 10052
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.578733302936623e-05,
+      "loss": 0.2722,
+      "step": 10053
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5768364918450985e-05,
+      "loss": 0.2477,
+      "step": 10054
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5749398785814244e-05,
+      "loss": 0.2579,
+      "step": 10055
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.573043463230422e-05,
+      "loss": 0.2434,
+      "step": 10056
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.571147245876915e-05,
+      "loss": 0.2551,
+      "step": 10057
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.569251226605705e-05,
+      "loss": 0.2365,
+      "step": 10058
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.567355405501602e-05,
+      "loss": 0.2294,
+      "step": 10059
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.565459782649387e-05,
+      "loss": 0.2397,
+      "step": 10060
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.563564358133852e-05,
+      "loss": 0.2435,
+      "step": 10061
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.561669132039762e-05,
+      "loss": 0.2456,
+      "step": 10062
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.559774104451889e-05,
+      "loss": 0.2449,
+      "step": 10063
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5578792754549826e-05,
+      "loss": 0.2416,
+      "step": 10064
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5559846451337984e-05,
+      "loss": 0.2441,
+      "step": 10065
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5540902135730634e-05,
+      "loss": 0.2393,
+      "step": 10066
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.552195980857522e-05,
+      "loss": 0.2462,
+      "step": 10067
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5503019470718766e-05,
+      "loss": 0.2355,
+      "step": 10068
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5484081123008504e-05,
+      "loss": 0.2441,
+      "step": 10069
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.546514476629139e-05,
+      "loss": 0.2211,
+      "step": 10070
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.544621040141444e-05,
+      "loss": 0.2202,
+      "step": 10071
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5427278029224436e-05,
+      "loss": 0.2368,
+      "step": 10072
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.540834765056819e-05,
+      "loss": 0.2465,
+      "step": 10073
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5389419266292306e-05,
+      "loss": 0.2578,
+      "step": 10074
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.537049287724343e-05,
+      "loss": 0.2445,
+      "step": 10075
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5351568484267994e-05,
+      "loss": 0.2415,
+      "step": 10076
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.533264608821247e-05,
+      "loss": 0.2297,
+      "step": 10077
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.531372568992308e-05,
+      "loss": 0.2463,
+      "step": 10078
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.529480729024613e-05,
+      "loss": 0.2407,
+      "step": 10079
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.527589089002768e-05,
+      "loss": 0.2386,
+      "step": 10080
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5256976490113854e-05,
+      "loss": 0.2219,
+      "step": 10081
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.523806409135057e-05,
+      "loss": 0.2433,
+      "step": 10082
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5219153694583645e-05,
+      "loss": 0.2504,
+      "step": 10083
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5200245300658936e-05,
+      "loss": 0.2249,
+      "step": 10084
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.518133891042205e-05,
+      "loss": 0.2418,
+      "step": 10085
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5162434524718674e-05,
+      "loss": 0.2473,
+      "step": 10086
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.514353214439425e-05,
+      "loss": 0.2371,
+      "step": 10087
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.512463177029416e-05,
+      "loss": 0.2286,
+      "step": 10088
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5105733403263835e-05,
+      "loss": 0.239,
+      "step": 10089
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.508683704414842e-05,
+      "loss": 0.244,
+      "step": 10090
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.506794269379314e-05,
+      "loss": 0.2556,
+      "step": 10091
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5049050353042964e-05,
+      "loss": 0.2361,
+      "step": 10092
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.503016002274295e-05,
+      "loss": 0.2349,
+      "step": 10093
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.50112717037379e-05,
+      "loss": 0.2604,
+      "step": 10094
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.499238539687267e-05,
+      "loss": 0.2344,
+      "step": 10095
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.497350110299193e-05,
+      "loss": 0.2319,
+      "step": 10096
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.495461882294024e-05,
+      "loss": 0.2349,
+      "step": 10097
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.49357385575622e-05,
+      "loss": 0.2369,
+      "step": 10098
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.491686030770215e-05,
+      "loss": 0.2321,
+      "step": 10099
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.489798407420452e-05,
+      "loss": 0.2562,
+      "step": 10100
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.487910985791349e-05,
+      "loss": 0.253,
+      "step": 10101
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.486023765967327e-05,
+      "loss": 0.2405,
+      "step": 10102
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.484136748032786e-05,
+      "loss": 0.227,
+      "step": 10103
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.48224993207213e-05,
+      "loss": 0.2426,
+      "step": 10104
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.480363318169743e-05,
+      "loss": 0.259,
+      "step": 10105
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.478476906410009e-05,
+      "loss": 0.2336,
+      "step": 10106
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4765906968772974e-05,
+      "loss": 0.2399,
+      "step": 10107
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.474704689655967e-05,
+      "loss": 0.2585,
+      "step": 10108
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.472818884830371e-05,
+      "loss": 0.2311,
+      "step": 10109
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.470933282484855e-05,
+      "loss": 0.2396,
+      "step": 10110
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.469047882703748e-05,
+      "loss": 0.2526,
+      "step": 10111
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.467162685571384e-05,
+      "loss": 0.2409,
+      "step": 10112
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4652776911720705e-05,
+      "loss": 0.2306,
+      "step": 10113
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4633928995901226e-05,
+      "loss": 0.2437,
+      "step": 10114
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.461508310909831e-05,
+      "loss": 0.2433,
+      "step": 10115
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.459623925215491e-05,
+      "loss": 0.2537,
+      "step": 10116
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.457739742591377e-05,
+      "loss": 0.2381,
+      "step": 10117
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.455855763121765e-05,
+      "loss": 0.237,
+      "step": 10118
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.453971986890911e-05,
+      "loss": 0.2397,
+      "step": 10119
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4520884139830765e-05,
+      "loss": 0.2459,
+      "step": 10120
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.450205044482495e-05,
+      "loss": 0.2363,
+      "step": 10121
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4483218784734084e-05,
+      "loss": 0.2441,
+      "step": 10122
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.446438916040036e-05,
+      "loss": 0.2328,
+      "step": 10123
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4445561572666025e-05,
+      "loss": 0.2377,
+      "step": 10124
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.442673602237305e-05,
+      "loss": 0.2462,
+      "step": 10125
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4407912510363505e-05,
+      "loss": 0.2385,
+      "step": 10126
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.438909103747924e-05,
+      "loss": 0.242,
+      "step": 10127
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.437027160456205e-05,
+      "loss": 0.2554,
+      "step": 10128
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4351454212453614e-05,
+      "loss": 0.2381,
+      "step": 10129
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.433263886199561e-05,
+      "loss": 0.2498,
+      "step": 10130
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.431382555402951e-05,
+      "loss": 0.239,
+      "step": 10131
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.429501428939678e-05,
+      "loss": 0.2248,
+      "step": 10132
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.427620506893873e-05,
+      "loss": 0.2429,
+      "step": 10133
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.425739789349667e-05,
+      "loss": 0.2319,
+      "step": 10134
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.423859276391168e-05,
+      "loss": 0.2407,
+      "step": 10135
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.42197896810249e-05,
+      "loss": 0.2433,
+      "step": 10136
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.420098864567724e-05,
+      "loss": 0.2342,
+      "step": 10137
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4182189658709647e-05,
+      "loss": 0.2324,
+      "step": 10138
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.416339272096285e-05,
+      "loss": 0.2412,
+      "step": 10139
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4144597833277635e-05,
+      "loss": 0.2228,
+      "step": 10140
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.41258049964945e-05,
+      "loss": 0.2452,
+      "step": 10141
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.410701421145408e-05,
+      "loss": 0.2331,
+      "step": 10142
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.408822547899668e-05,
+      "loss": 0.2364,
+      "step": 10143
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4069438799962755e-05,
+      "loss": 0.2396,
+      "step": 10144
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4050654175192436e-05,
+      "loss": 0.231,
+      "step": 10145
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4031871605525986e-05,
+      "loss": 0.2313,
+      "step": 10146
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.401309109180339e-05,
+      "loss": 0.2334,
+      "step": 10147
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.399431263486463e-05,
+      "loss": 0.2282,
+      "step": 10148
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.397553623554954e-05,
+      "loss": 0.2473,
+      "step": 10149
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.395676189469798e-05,
+      "loss": 0.2394,
+      "step": 10150
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3937989613149553e-05,
+      "loss": 0.2204,
+      "step": 10151
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.391921939174396e-05,
+      "loss": 0.237,
+      "step": 10152
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3900451231320615e-05,
+      "loss": 0.2378,
+      "step": 10153
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3881685132718985e-05,
+      "loss": 0.2372,
+      "step": 10154
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.386292109677835e-05,
+      "loss": 0.2418,
+      "step": 10155
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.384415912433801e-05,
+      "loss": 0.2319,
+      "step": 10156
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3825399216237014e-05,
+      "loss": 0.2327,
+      "step": 10157
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.380664137331449e-05,
+      "loss": 0.2298,
+      "step": 10158
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.37878855964093e-05,
+      "loss": 0.2535,
+      "step": 10159
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3769131886360414e-05,
+      "loss": 0.2327,
+      "step": 10160
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3750380244006534e-05,
+      "loss": 0.2372,
+      "step": 10161
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.373163067018629e-05,
+      "loss": 0.2321,
+      "step": 10162
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.371288316573837e-05,
+      "loss": 0.2272,
+      "step": 10163
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.369413773150116e-05,
+      "loss": 0.2311,
+      "step": 10164
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.367539436831315e-05,
+      "loss": 0.2559,
+      "step": 10165
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3656653077012556e-05,
+      "loss": 0.2213,
+      "step": 10166
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.363791385843768e-05,
+      "loss": 0.2507,
+      "step": 10167
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.36191767134266e-05,
+      "loss": 0.2317,
+      "step": 10168
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.36004416428173e-05,
+      "loss": 0.2274,
+      "step": 10169
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3581708647447805e-05,
+      "loss": 0.2379,
+      "step": 10170
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.356297772815585e-05,
+      "loss": 0.2383,
+      "step": 10171
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.354424888577928e-05,
+      "loss": 0.2316,
+      "step": 10172
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.352552212115571e-05,
+      "loss": 0.2447,
+      "step": 10173
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.350679743512266e-05,
+      "loss": 0.2447,
+      "step": 10174
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3488074828517684e-05,
+      "loss": 0.2488,
+      "step": 10175
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3469354302178076e-05,
+      "loss": 0.2541,
+      "step": 10176
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.34506358569412e-05,
+      "loss": 0.2263,
+      "step": 10177
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3431919493644165e-05,
+      "loss": 0.234,
+      "step": 10178
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.341320521312414e-05,
+      "loss": 0.2237,
+      "step": 10179
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.339449301621806e-05,
+      "loss": 0.2393,
+      "step": 10180
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3375782903762913e-05,
+      "loss": 0.2424,
+      "step": 10181
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.335707487659545e-05,
+      "loss": 0.2372,
+      "step": 10182
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.333836893555244e-05,
+      "loss": 0.2487,
+      "step": 10183
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.331966508147047e-05,
+      "loss": 0.244,
+      "step": 10184
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3300963315186146e-05,
+      "loss": 0.2375,
+      "step": 10185
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.328226363753582e-05,
+      "loss": 0.2406,
+      "step": 10186
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.326356604935598e-05,
+      "loss": 0.2404,
+      "step": 10187
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3244870551482715e-05,
+      "loss": 0.2325,
+      "step": 10188
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.322617714475231e-05,
+      "loss": 0.2377,
+      "step": 10189
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3207485830000755e-05,
+      "loss": 0.2506,
+      "step": 10190
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3188796608064106e-05,
+      "loss": 0.2334,
+      "step": 10191
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.317010947977818e-05,
+      "loss": 0.2382,
+      "step": 10192
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3151424445978825e-05,
+      "loss": 0.2312,
+      "step": 10193
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.313274150750167e-05,
+      "loss": 0.2222,
+      "step": 10194
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3114060665182384e-05,
+      "loss": 0.2368,
+      "step": 10195
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.309538191985642e-05,
+      "loss": 0.2301,
+      "step": 10196
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.307670527235926e-05,
+      "loss": 0.2473,
+      "step": 10197
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.305803072352613e-05,
+      "loss": 0.2378,
+      "step": 10198
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.303935827419236e-05,
+      "loss": 0.2338,
+      "step": 10199
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.302068792519299e-05,
+      "loss": 0.2251,
+      "step": 10200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.300201967736316e-05,
+      "loss": 0.2391,
+      "step": 10201
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2983353531537694e-05,
+      "loss": 0.241,
+      "step": 10202
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2964689488551556e-05,
+      "loss": 0.2382,
+      "step": 10203
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.294602754923941e-05,
+      "loss": 0.2321,
+      "step": 10204
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2927367714436005e-05,
+      "loss": 0.2342,
+      "step": 10205
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2908709984975836e-05,
+      "loss": 0.2416,
+      "step": 10206
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.289005436169349e-05,
+      "loss": 0.233,
+      "step": 10207
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.287140084542318e-05,
+      "loss": 0.2286,
+      "step": 10208
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.285274943699932e-05,
+      "loss": 0.2297,
+      "step": 10209
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.283410013725603e-05,
+      "loss": 0.2449,
+      "step": 10210
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2815452947027475e-05,
+      "loss": 0.2188,
+      "step": 10211
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.279680786714758e-05,
+      "loss": 0.2425,
+      "step": 10212
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.277816489845036e-05,
+      "loss": 0.2567,
+      "step": 10213
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.275952404176949e-05,
+      "loss": 0.2401,
+      "step": 10214
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.274088529793883e-05,
+      "loss": 0.2435,
+      "step": 10215
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2722248667791905e-05,
+      "loss": 0.2348,
+      "step": 10216
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.270361415216232e-05,
+      "loss": 0.2334,
+      "step": 10217
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2684981751883435e-05,
+      "loss": 0.2354,
+      "step": 10218
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2666351467788664e-05,
+      "loss": 0.231,
+      "step": 10219
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.264772330071121e-05,
+      "loss": 0.2222,
+      "step": 10220
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.262909725148425e-05,
+      "loss": 0.2253,
+      "step": 10221
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2610473320940804e-05,
+      "loss": 0.2256,
+      "step": 10222
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.25918515099139e-05,
+      "loss": 0.231,
+      "step": 10223
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2573231819236326e-05,
+      "loss": 0.2228,
+      "step": 10224
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.255461424974093e-05,
+      "loss": 0.2421,
+      "step": 10225
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2535998802260366e-05,
+      "loss": 0.2356,
+      "step": 10226
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2517385477627214e-05,
+      "loss": 0.2445,
+      "step": 10227
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2498774276673915e-05,
+      "loss": 0.2399,
+      "step": 10228
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2480165200232934e-05,
+      "loss": 0.2258,
+      "step": 10229
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2461558249136525e-05,
+      "loss": 0.2338,
+      "step": 10230
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.244295342421692e-05,
+      "loss": 0.2355,
+      "step": 10231
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2424350726306204e-05,
+      "loss": 0.2266,
+      "step": 10232
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2405750156236435e-05,
+      "loss": 0.2353,
+      "step": 10233
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2387151714839454e-05,
+      "loss": 0.2415,
+      "step": 10234
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.236855540294716e-05,
+      "loss": 0.2162,
+      "step": 10235
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.234996122139122e-05,
+      "loss": 0.2343,
+      "step": 10236
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.233136917100334e-05,
+      "loss": 0.2493,
+      "step": 10237
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.231277925261502e-05,
+      "loss": 0.2422,
+      "step": 10238
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.229419146705764e-05,
+      "loss": 0.2298,
+      "step": 10239
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.227560581516265e-05,
+      "loss": 0.2435,
+      "step": 10240
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.225702229776122e-05,
+      "loss": 0.2474,
+      "step": 10241
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2238440915684575e-05,
+      "loss": 0.2304,
+      "step": 10242
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2219861669763694e-05,
+      "loss": 0.2298,
+      "step": 10243
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.220128456082962e-05,
+      "loss": 0.2277,
+      "step": 10244
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.218270958971316e-05,
+      "loss": 0.2547,
+      "step": 10245
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.216413675724516e-05,
+      "loss": 0.2329,
+      "step": 10246
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.214556606425625e-05,
+      "loss": 0.2539,
+      "step": 10247
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.212699751157697e-05,
+      "loss": 0.2292,
+      "step": 10248
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.210843110003789e-05,
+      "loss": 0.2405,
+      "step": 10249
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.208986683046937e-05,
+      "loss": 0.2389,
+      "step": 10250
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.207130470370166e-05,
+      "loss": 0.2421,
+      "step": 10251
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.205274472056503e-05,
+      "loss": 0.2514,
+      "step": 10252
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2034186881889524e-05,
+      "loss": 0.2231,
+      "step": 10253
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.20156311885052e-05,
+      "loss": 0.2448,
+      "step": 10254
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.199707764124191e-05,
+      "loss": 0.2505,
+      "step": 10255
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.197852624092954e-05,
+      "loss": 0.2356,
+      "step": 10256
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1959976988397743e-05,
+      "loss": 0.2282,
+      "step": 10257
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.194142988447619e-05,
+      "loss": 0.2338,
+      "step": 10258
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.192288492999436e-05,
+      "loss": 0.2335,
+      "step": 10259
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1904342125781755e-05,
+      "loss": 0.2369,
+      "step": 10260
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.188580147266764e-05,
+      "loss": 0.2291,
+      "step": 10261
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.18672629714813e-05,
+      "loss": 0.2334,
+      "step": 10262
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.184872662305182e-05,
+      "loss": 0.2576,
+      "step": 10263
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.183019242820835e-05,
+      "loss": 0.2386,
+      "step": 10264
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1811660387779714e-05,
+      "loss": 0.2501,
+      "step": 10265
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1793130502594866e-05,
+      "loss": 0.2356,
+      "step": 10266
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1774602773482525e-05,
+      "loss": 0.2264,
+      "step": 10267
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.175607720127136e-05,
+      "loss": 0.2426,
+      "step": 10268
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.173755378678986e-05,
+      "loss": 0.2292,
+      "step": 10269
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1719032530866615e-05,
+      "loss": 0.2492,
+      "step": 10270
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.170051343432989e-05,
+      "loss": 0.2321,
+      "step": 10271
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1681996498008054e-05,
+      "loss": 0.2381,
+      "step": 10272
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.166348172272919e-05,
+      "loss": 0.229,
+      "step": 10273
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.164496910932149e-05,
+      "loss": 0.2254,
+      "step": 10274
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1626458658612795e-05,
+      "loss": 0.2286,
+      "step": 10275
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.160795037143114e-05,
+      "loss": 0.2197,
+      "step": 10276
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1589444248604216e-05,
+      "loss": 0.2493,
+      "step": 10277
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.157094029095977e-05,
+      "loss": 0.2323,
+      "step": 10278
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.155243849932534e-05,
+      "loss": 0.2379,
+      "step": 10279
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.153393887452852e-05,
+      "loss": 0.2278,
+      "step": 10280
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.151544141739662e-05,
+      "loss": 0.2451,
+      "step": 10281
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.149694612875701e-05,
+      "loss": 0.2334,
+      "step": 10282
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.147845300943683e-05,
+      "loss": 0.2331,
+      "step": 10283
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.14599620602633e-05,
+      "loss": 0.23,
+      "step": 10284
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1441473282063325e-05,
+      "loss": 0.2333,
+      "step": 10285
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.142298667566391e-05,
+      "loss": 0.2268,
+      "step": 10286
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.140450224189184e-05,
+      "loss": 0.2383,
+      "step": 10287
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.138601998157384e-05,
+      "loss": 0.2347,
+      "step": 10288
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1367539895536486e-05,
+      "loss": 0.2395,
+      "step": 10289
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.13490619846064e-05,
+      "loss": 0.2396,
+      "step": 10290
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1330586249609925e-05,
+      "loss": 0.2456,
+      "step": 10291
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1312112691373485e-05,
+      "loss": 0.232,
+      "step": 10292
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.129364131072323e-05,
+      "loss": 0.2308,
+      "step": 10293
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1275172108485395e-05,
+      "loss": 0.2452,
+      "step": 10294
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1256705085485926e-05,
+      "loss": 0.2185,
+      "step": 10295
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.123824024255084e-05,
+      "loss": 0.2208,
+      "step": 10296
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.121977758050594e-05,
+      "loss": 0.2169,
+      "step": 10297
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1201317100177015e-05,
+      "loss": 0.2269,
+      "step": 10298
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.118285880238967e-05,
+      "loss": 0.2443,
+      "step": 10299
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.116440268796954e-05,
+      "loss": 0.2418,
+      "step": 10300
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1145948757741965e-05,
+      "loss": 0.2309,
+      "step": 10301
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.112749701253242e-05,
+      "loss": 0.2333,
+      "step": 10302
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.110904745316608e-05,
+      "loss": 0.2195,
+      "step": 10303
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1090600080468186e-05,
+      "loss": 0.2323,
+      "step": 10304
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.107215489526377e-05,
+      "loss": 0.2208,
+      "step": 10305
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.105371189837775e-05,
+      "loss": 0.2314,
+      "step": 10306
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.103527109063508e-05,
+      "loss": 0.2354,
+      "step": 10307
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.101683247286051e-05,
+      "loss": 0.2348,
+      "step": 10308
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.099839604587864e-05,
+      "loss": 0.2402,
+      "step": 10309
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.097996181051417e-05,
+      "loss": 0.2372,
+      "step": 10310
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.096152976759145e-05,
+      "loss": 0.2189,
+      "step": 10311
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.094309991793498e-05,
+      "loss": 0.2256,
+      "step": 10312
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.092467226236896e-05,
+      "loss": 0.228,
+      "step": 10313
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.090624680171763e-05,
+      "loss": 0.2316,
+      "step": 10314
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.088782353680501e-05,
+      "loss": 0.2352,
+      "step": 10315
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.086940246845517e-05,
+      "loss": 0.2446,
+      "step": 10316
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.085098359749198e-05,
+      "loss": 0.2272,
+      "step": 10317
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0832566924739146e-05,
+      "loss": 0.2416,
+      "step": 10318
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.081415245102048e-05,
+      "loss": 0.2238,
+      "step": 10319
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.079574017715948e-05,
+      "loss": 0.2435,
+      "step": 10320
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0777330103979726e-05,
+      "loss": 0.2224,
+      "step": 10321
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.075892223230454e-05,
+      "loss": 0.2433,
+      "step": 10322
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0740516562957284e-05,
+      "loss": 0.24,
+      "step": 10323
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0722113096761116e-05,
+      "loss": 0.229,
+      "step": 10324
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.070371183453918e-05,
+      "loss": 0.2513,
+      "step": 10325
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.068531277711442e-05,
+      "loss": 0.2393,
+      "step": 10326
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.066691592530983e-05,
+      "loss": 0.236,
+      "step": 10327
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.064852127994815e-05,
+      "loss": 0.2394,
+      "step": 10328
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.063012884185211e-05,
+      "loss": 0.2382,
+      "step": 10329
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0611738611844274e-05,
+      "loss": 0.2514,
+      "step": 10330
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0593350590747224e-05,
+      "loss": 0.2291,
+      "step": 10331
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.05749647793833e-05,
+      "loss": 0.2442,
+      "step": 10332
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.05565811785749e-05,
+      "loss": 0.2407,
+      "step": 10333
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.053819978914416e-05,
+      "loss": 0.2447,
+      "step": 10334
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.051982061191325e-05,
+      "loss": 0.2325,
+      "step": 10335
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.050144364770415e-05,
+      "loss": 0.2547,
+      "step": 10336
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.04830688973388e-05,
+      "loss": 0.2325,
+      "step": 10337
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0464696361638996e-05,
+      "loss": 0.2389,
+      "step": 10338
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.044632604142651e-05,
+      "loss": 0.2326,
+      "step": 10339
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.042795793752289e-05,
+      "loss": 0.2457,
+      "step": 10340
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.040959205074971e-05,
+      "loss": 0.2334,
+      "step": 10341
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.039122838192834e-05,
+      "loss": 0.2323,
+      "step": 10342
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0372866931880184e-05,
+      "loss": 0.2419,
+      "step": 10343
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.035450770142638e-05,
+      "loss": 0.2412,
+      "step": 10344
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.033615069138812e-05,
+      "loss": 0.2371,
+      "step": 10345
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.031779590258636e-05,
+      "loss": 0.2366,
+      "step": 10346
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.029944333584215e-05,
+      "loss": 0.2454,
+      "step": 10347
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0281092991976155e-05,
+      "loss": 0.2248,
+      "step": 10348
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0262744871809206e-05,
+      "loss": 0.2397,
+      "step": 10349
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0244398976161864e-05,
+      "loss": 0.2461,
+      "step": 10350
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0226055305854755e-05,
+      "loss": 0.2301,
+      "step": 10351
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.020771386170819e-05,
+      "loss": 0.2393,
+      "step": 10352
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.01893746445426e-05,
+      "loss": 0.2282,
+      "step": 10353
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.017103765517812e-05,
+      "loss": 0.2263,
+      "step": 10354
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.015270289443499e-05,
+      "loss": 0.2329,
+      "step": 10355
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.013437036313312e-05,
+      "loss": 0.2475,
+      "step": 10356
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.011604006209254e-05,
+      "loss": 0.2309,
+      "step": 10357
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.009771199213302e-05,
+      "loss": 0.2356,
+      "step": 10358
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0079386154074326e-05,
+      "loss": 0.2285,
+      "step": 10359
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.006106254873605e-05,
+      "loss": 0.2267,
+      "step": 10360
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.004274117693778e-05,
+      "loss": 0.239,
+      "step": 10361
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.002442203949886e-05,
+      "loss": 0.2359,
+      "step": 10362
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0006105137238734e-05,
+      "loss": 0.2548,
+      "step": 10363
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9987790470976525e-05,
+      "loss": 0.2286,
+      "step": 10364
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.996947804153146e-05,
+      "loss": 0.2342,
+      "step": 10365
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.995116784972248e-05,
+      "loss": 0.2371,
+      "step": 10366
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.993285989636863e-05,
+      "loss": 0.241,
+      "step": 10367
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.99145541822886e-05,
+      "loss": 0.2437,
+      "step": 10368
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.989625070830122e-05,
+      "loss": 0.2438,
+      "step": 10369
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.987794947522506e-05,
+      "loss": 0.23,
+      "step": 10370
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9859650483878714e-05,
+      "loss": 0.2198,
+      "step": 10371
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.984135373508054e-05,
+      "loss": 0.2299,
+      "step": 10372
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.982305922964895e-05,
+      "loss": 0.2185,
+      "step": 10373
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.98047669684021e-05,
+      "loss": 0.2276,
+      "step": 10374
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.978647695215817e-05,
+      "loss": 0.2334,
+      "step": 10375
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.976818918173514e-05,
+      "loss": 0.236,
+      "step": 10376
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9749903657951e-05,
+      "loss": 0.2265,
+      "step": 10377
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.973162038162351e-05,
+      "loss": 0.2269,
+      "step": 10378
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9713339353570475e-05,
+      "loss": 0.251,
+      "step": 10379
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.969506057460943e-05,
+      "loss": 0.2269,
+      "step": 10380
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9676784045558e-05,
+      "loss": 0.237,
+      "step": 10381
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.965850976723355e-05,
+      "loss": 0.2286,
+      "step": 10382
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9640237740453374e-05,
+      "loss": 0.2281,
+      "step": 10383
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.962196796603479e-05,
+      "loss": 0.2386,
+      "step": 10384
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.960370044479481e-05,
+      "loss": 0.242,
+      "step": 10385
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9585435177550564e-05,
+      "loss": 0.2291,
+      "step": 10386
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.956717216511893e-05,
+      "loss": 0.2458,
+      "step": 10387
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.954891140831667e-05,
+      "loss": 0.2346,
+      "step": 10388
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9530652907960595e-05,
+      "loss": 0.2331,
+      "step": 10389
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9512396664867256e-05,
+      "loss": 0.2253,
+      "step": 10390
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.949414267985324e-05,
+      "loss": 0.2194,
+      "step": 10391
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.947589095373489e-05,
+      "loss": 0.2297,
+      "step": 10392
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9457641487328586e-05,
+      "loss": 0.2254,
+      "step": 10393
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9439394281450525e-05,
+      "loss": 0.2309,
+      "step": 10394
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9421149336916775e-05,
+      "loss": 0.231,
+      "step": 10395
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.940290665454341e-05,
+      "loss": 0.2368,
+      "step": 10396
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.938466623514628e-05,
+      "loss": 0.2191,
+      "step": 10397
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.936642807954127e-05,
+      "loss": 0.2284,
+      "step": 10398
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.934819218854402e-05,
+      "loss": 0.2315,
+      "step": 10399
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9329958562970205e-05,
+      "loss": 0.2389,
+      "step": 10400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9311727203635263e-05,
+      "loss": 0.2212,
+      "step": 10401
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.929349811135465e-05,
+      "loss": 0.2293,
+      "step": 10402
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9275271286943625e-05,
+      "loss": 0.2342,
+      "step": 10403
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.925704673121746e-05,
+      "loss": 0.2418,
+      "step": 10404
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9238824444991164e-05,
+      "loss": 0.2317,
+      "step": 10405
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.922060442907983e-05,
+      "loss": 0.2394,
+      "step": 10406
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9202386684298304e-05,
+      "loss": 0.2253,
+      "step": 10407
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.918417121146139e-05,
+      "loss": 0.2358,
+      "step": 10408
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.916595801138375e-05,
+      "loss": 0.2368,
+      "step": 10409
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9147747084880046e-05,
+      "loss": 0.2303,
+      "step": 10410
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.91295384327647e-05,
+      "loss": 0.2307,
+      "step": 10411
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.911133205585215e-05,
+      "loss": 0.2269,
+      "step": 10412
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9093127954956654e-05,
+      "loss": 0.2352,
+      "step": 10413
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.907492613089244e-05,
+      "loss": 0.2411,
+      "step": 10414
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.905672658447352e-05,
+      "loss": 0.2288,
+      "step": 10415
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9038529316513974e-05,
+      "loss": 0.2299,
+      "step": 10416
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.902033432782758e-05,
+      "loss": 0.2323,
+      "step": 10417
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.90021416192282e-05,
+      "loss": 0.2383,
+      "step": 10418
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.898395119152944e-05,
+      "loss": 0.2353,
+      "step": 10419
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.896576304554495e-05,
+      "loss": 0.233,
+      "step": 10420
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8947577182088125e-05,
+      "loss": 0.23,
+      "step": 10421
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.892939360197242e-05,
+      "loss": 0.2289,
+      "step": 10422
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8911212306011e-05,
+      "loss": 0.2317,
+      "step": 10423
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8893033295017135e-05,
+      "loss": 0.2405,
+      "step": 10424
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.887485656980381e-05,
+      "loss": 0.2201,
+      "step": 10425
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.885668213118405e-05,
+      "loss": 0.2372,
+      "step": 10426
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8838509979970684e-05,
+      "loss": 0.2262,
+      "step": 10427
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8820340116976456e-05,
+      "loss": 0.2206,
+      "step": 10428
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8802172543014013e-05,
+      "loss": 0.2399,
+      "step": 10429
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.878400725889595e-05,
+      "loss": 0.2243,
+      "step": 10430
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.876584426543467e-05,
+      "loss": 0.2278,
+      "step": 10431
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.874768356344257e-05,
+      "loss": 0.237,
+      "step": 10432
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8729525153731835e-05,
+      "loss": 0.2233,
+      "step": 10433
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8711369037114684e-05,
+      "loss": 0.2441,
+      "step": 10434
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8693215214403064e-05,
+      "loss": 0.2307,
+      "step": 10435
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8675063686409015e-05,
+      "loss": 0.224,
+      "step": 10436
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.865691445394428e-05,
+      "loss": 0.2254,
+      "step": 10437
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.863876751782066e-05,
+      "loss": 0.2257,
+      "step": 10438
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.862062287884973e-05,
+      "loss": 0.2246,
+      "step": 10439
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.860248053784308e-05,
+      "loss": 0.2365,
+      "step": 10440
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8584340495612047e-05,
+      "loss": 0.2361,
+      "step": 10441
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8566202752968036e-05,
+      "loss": 0.2373,
+      "step": 10442
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.85480673107222e-05,
+      "loss": 0.2167,
+      "step": 10443
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.852993416968572e-05,
+      "loss": 0.2421,
+      "step": 10444
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.851180333066953e-05,
+      "loss": 0.232,
+      "step": 10445
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.849367479448463e-05,
+      "loss": 0.2324,
+      "step": 10446
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8475548561941774e-05,
+      "loss": 0.2247,
+      "step": 10447
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8457424633851675e-05,
+      "loss": 0.2317,
+      "step": 10448
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.843930301102488e-05,
+      "loss": 0.2463,
+      "step": 10449
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.842118369427199e-05,
+      "loss": 0.2479,
+      "step": 10450
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.840306668440328e-05,
+      "loss": 0.23,
+      "step": 10451
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8384951982229154e-05,
+      "loss": 0.2264,
+      "step": 10452
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.836683958855972e-05,
+      "loss": 0.2365,
+      "step": 10453
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8348729504205116e-05,
+      "loss": 0.2392,
+      "step": 10454
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.833062172997527e-05,
+      "loss": 0.2353,
+      "step": 10455
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.831251626668013e-05,
+      "loss": 0.2282,
+      "step": 10456
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.829441311512938e-05,
+      "loss": 0.2136,
+      "step": 10457
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8276312276132785e-05,
+      "loss": 0.2277,
+      "step": 10458
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8258213750499835e-05,
+      "loss": 0.2228,
+      "step": 10459
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8240117539040064e-05,
+      "loss": 0.2147,
+      "step": 10460
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.822202364256279e-05,
+      "loss": 0.2153,
+      "step": 10461
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.820393206187724e-05,
+      "loss": 0.2344,
+      "step": 10462
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.818584279779265e-05,
+      "loss": 0.2225,
+      "step": 10463
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.816775585111797e-05,
+      "loss": 0.2154,
+      "step": 10464
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.814967122266225e-05,
+      "loss": 0.2227,
+      "step": 10465
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8131588913234236e-05,
+      "loss": 0.2358,
+      "step": 10466
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8113508923642755e-05,
+      "loss": 0.2445,
+      "step": 10467
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.809543125469641e-05,
+      "loss": 0.2387,
+      "step": 10468
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.807735590720368e-05,
+      "loss": 0.2272,
+      "step": 10469
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.805928288197307e-05,
+      "loss": 0.2237,
+      "step": 10470
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.804121217981283e-05,
+      "loss": 0.235,
+      "step": 10471
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.802314380153127e-05,
+      "loss": 0.2441,
+      "step": 10472
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8005077747936444e-05,
+      "loss": 0.2538,
+      "step": 10473
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.798701401983633e-05,
+      "loss": 0.2253,
+      "step": 10474
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7968952618038935e-05,
+      "loss": 0.2341,
+      "step": 10475
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.795089354335196e-05,
+      "loss": 0.24,
+      "step": 10476
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.793283679658319e-05,
+      "loss": 0.2299,
+      "step": 10477
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7914782378540144e-05,
+      "loss": 0.2253,
+      "step": 10478
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7896730290030404e-05,
+      "loss": 0.2233,
+      "step": 10479
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7878680531861256e-05,
+      "loss": 0.2285,
+      "step": 10480
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.786063310484007e-05,
+      "loss": 0.2298,
+      "step": 10481
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7842588009773956e-05,
+      "loss": 0.2248,
+      "step": 10482
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.782454524747005e-05,
+      "loss": 0.224,
+      "step": 10483
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7806504818735265e-05,
+      "loss": 0.2368,
+      "step": 10484
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.778846672437654e-05,
+      "loss": 0.2279,
+      "step": 10485
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.777043096520053e-05,
+      "loss": 0.2194,
+      "step": 10486
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.775239754201405e-05,
+      "loss": 0.2354,
+      "step": 10487
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7734366455623456e-05,
+      "loss": 0.226,
+      "step": 10488
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.771633770683536e-05,
+      "loss": 0.2339,
+      "step": 10489
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7698311296455975e-05,
+      "loss": 0.2418,
+      "step": 10490
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7680287225291664e-05,
+      "loss": 0.2277,
+      "step": 10491
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.766226549414846e-05,
+      "loss": 0.2326,
+      "step": 10492
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.764424610383247e-05,
+      "loss": 0.2255,
+      "step": 10493
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7626229055149544e-05,
+      "loss": 0.2304,
+      "step": 10494
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.760821434890558e-05,
+      "loss": 0.2378,
+      "step": 10495
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.759020198590622e-05,
+      "loss": 0.2189,
+      "step": 10496
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7572191966957146e-05,
+      "loss": 0.2293,
+      "step": 10497
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.755418429286379e-05,
+      "loss": 0.2392,
+      "step": 10498
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.753617896443162e-05,
+      "loss": 0.2261,
+      "step": 10499
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.751817598246587e-05,
+      "loss": 0.2305,
+      "step": 10500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.75001753477718e-05,
+      "loss": 0.2277,
+      "step": 10501
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.748217706115443e-05,
+      "loss": 0.2297,
+      "step": 10502
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7464181123418806e-05,
+      "loss": 0.2173,
+      "step": 10503
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7446187535369727e-05,
+      "loss": 0.2225,
+      "step": 10504
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.742819629781203e-05,
+      "loss": 0.2183,
+      "step": 10505
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7410207411550325e-05,
+      "loss": 0.2376,
+      "step": 10506
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7392220877389285e-05,
+      "loss": 0.2312,
+      "step": 10507
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7374236696133204e-05,
+      "loss": 0.2288,
+      "step": 10508
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7356254868586534e-05,
+      "loss": 0.2233,
+      "step": 10509
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7338275395553475e-05,
+      "loss": 0.2206,
+      "step": 10510
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.732029827783823e-05,
+      "loss": 0.2273,
+      "step": 10511
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.730232351624474e-05,
+      "loss": 0.2411,
+      "step": 10512
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.728435111157703e-05,
+      "loss": 0.247,
+      "step": 10513
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7266381064638846e-05,
+      "loss": 0.2377,
+      "step": 10514
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.724841337623396e-05,
+      "loss": 0.237,
+      "step": 10515
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.723044804716593e-05,
+      "loss": 0.2311,
+      "step": 10516
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.721248507823835e-05,
+      "loss": 0.2226,
+      "step": 10517
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.719452447025451e-05,
+      "loss": 0.2286,
+      "step": 10518
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7176566224017804e-05,
+      "loss": 0.2202,
+      "step": 10519
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7158610340331353e-05,
+      "loss": 0.2229,
+      "step": 10520
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.71406568199983e-05,
+      "loss": 0.2153,
+      "step": 10521
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.712270566382155e-05,
+      "loss": 0.2276,
+      "step": 10522
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7104756872604075e-05,
+      "loss": 0.2337,
+      "step": 10523
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7086810447148536e-05,
+      "loss": 0.2238,
+      "step": 10524
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.706886638825767e-05,
+      "loss": 0.2335,
+      "step": 10525
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7050924696734025e-05,
+      "loss": 0.2371,
+      "step": 10526
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.703298537338003e-05,
+      "loss": 0.2242,
+      "step": 10527
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.701504841899799e-05,
+      "loss": 0.2364,
+      "step": 10528
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.699711383439023e-05,
+      "loss": 0.2304,
+      "step": 10529
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.697918162035879e-05,
+      "loss": 0.2402,
+      "step": 10530
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.696125177770578e-05,
+      "loss": 0.221,
+      "step": 10531
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.694332430723305e-05,
+      "loss": 0.2347,
+      "step": 10532
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.692539920974248e-05,
+      "loss": 0.2303,
+      "step": 10533
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.69074764860357e-05,
+      "loss": 0.2382,
+      "step": 10534
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6889556136914404e-05,
+      "loss": 0.244,
+      "step": 10535
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.687163816318e-05,
+      "loss": 0.2433,
+      "step": 10536
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.685372256563394e-05,
+      "loss": 0.2355,
+      "step": 10537
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.683580934507748e-05,
+      "loss": 0.232,
+      "step": 10538
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6817898502311773e-05,
+      "loss": 0.2361,
+      "step": 10539
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6799990038137955e-05,
+      "loss": 0.2326,
+      "step": 10540
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6782083953356894e-05,
+      "loss": 0.2304,
+      "step": 10541
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.676418024876955e-05,
+      "loss": 0.2251,
+      "step": 10542
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.674627892517658e-05,
+      "loss": 0.2222,
+      "step": 10543
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.672837998337871e-05,
+      "loss": 0.2171,
+      "step": 10544
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.671048342417641e-05,
+      "loss": 0.239,
+      "step": 10545
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6692589248370176e-05,
+      "loss": 0.2183,
+      "step": 10546
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.667469745676031e-05,
+      "loss": 0.2244,
+      "step": 10547
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.665680805014697e-05,
+      "loss": 0.227,
+      "step": 10548
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6638921029330355e-05,
+      "loss": 0.2365,
+      "step": 10549
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.662103639511043e-05,
+      "loss": 0.229,
+      "step": 10550
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.660315414828705e-05,
+      "loss": 0.2294,
+      "step": 10551
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6585274289660086e-05,
+      "loss": 0.2337,
+      "step": 10552
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6567396820029164e-05,
+      "loss": 0.2209,
+      "step": 10553
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.65495217401939e-05,
+      "loss": 0.2271,
+      "step": 10554
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6531649050953715e-05,
+      "loss": 0.2181,
+      "step": 10555
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.651377875310805e-05,
+      "loss": 0.232,
+      "step": 10556
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.649591084745608e-05,
+      "loss": 0.222,
+      "step": 10557
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.647804533479702e-05,
+      "loss": 0.2375,
+      "step": 10558
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.646018221592985e-05,
+      "loss": 0.2276,
+      "step": 10559
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.644232149165357e-05,
+      "loss": 0.2101,
+      "step": 10560
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.642446316276694e-05,
+      "loss": 0.223,
+      "step": 10561
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.640660723006877e-05,
+      "loss": 0.2368,
+      "step": 10562
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6388753694357566e-05,
+      "loss": 0.2314,
+      "step": 10563
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.637090255643194e-05,
+      "loss": 0.2241,
+      "step": 10564
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.635305381709021e-05,
+      "loss": 0.2217,
+      "step": 10565
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.633520747713073e-05,
+      "loss": 0.2393,
+      "step": 10566
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6317363537351654e-05,
+      "loss": 0.2151,
+      "step": 10567
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.629952199855108e-05,
+      "loss": 0.2164,
+      "step": 10568
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.62816828615269e-05,
+      "loss": 0.2254,
+      "step": 10569
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6263846127077095e-05,
+      "loss": 0.2277,
+      "step": 10570
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.624601179599931e-05,
+      "loss": 0.2232,
+      "step": 10571
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.622817986909129e-05,
+      "loss": 0.2371,
+      "step": 10572
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.621035034715049e-05,
+      "loss": 0.2182,
+      "step": 10573
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6192523230974424e-05,
+      "loss": 0.234,
+      "step": 10574
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.617469852136035e-05,
+      "loss": 0.2328,
+      "step": 10575
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6156876219105536e-05,
+      "loss": 0.2251,
+      "step": 10576
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6139056325007036e-05,
+      "loss": 0.2256,
+      "step": 10577
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6121238839861924e-05,
+      "loss": 0.2209,
+      "step": 10578
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6103423764467014e-05,
+      "loss": 0.2273,
+      "step": 10579
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6085611099619174e-05,
+      "loss": 0.2214,
+      "step": 10580
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6067800846115014e-05,
+      "loss": 0.2182,
+      "step": 10581
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6049993004751166e-05,
+      "loss": 0.2227,
+      "step": 10582
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.603218757632402e-05,
+      "loss": 0.2281,
+      "step": 10583
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.601438456163002e-05,
+      "loss": 0.2234,
+      "step": 10584
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.599658396146532e-05,
+      "loss": 0.2351,
+      "step": 10585
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5978785776626146e-05,
+      "loss": 0.2139,
+      "step": 10586
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.596099000790849e-05,
+      "loss": 0.2277,
+      "step": 10587
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.594319665610828e-05,
+      "loss": 0.23,
+      "step": 10588
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.592540572202128e-05,
+      "loss": 0.2296,
+      "step": 10589
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5907617206443275e-05,
+      "loss": 0.2321,
+      "step": 10590
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.58898311101698e-05,
+      "loss": 0.2276,
+      "step": 10591
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.587204743399641e-05,
+      "loss": 0.2206,
+      "step": 10592
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.585426617871843e-05,
+      "loss": 0.229,
+      "step": 10593
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.583648734513118e-05,
+      "loss": 0.2515,
+      "step": 10594
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.581871093402977e-05,
+      "loss": 0.2291,
+      "step": 10595
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.580093694620934e-05,
+      "loss": 0.2483,
+      "step": 10596
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.578316538246474e-05,
+      "loss": 0.2185,
+      "step": 10597
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.576539624359091e-05,
+      "loss": 0.23,
+      "step": 10598
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.574762953038249e-05,
+      "loss": 0.2204,
+      "step": 10599
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.572986524363418e-05,
+      "loss": 0.231,
+      "step": 10600
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5712103384140426e-05,
+      "loss": 0.2124,
+      "step": 10601
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.569434395269572e-05,
+      "loss": 0.2132,
+      "step": 10602
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.56765869500943e-05,
+      "loss": 0.225,
+      "step": 10603
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.565883237713033e-05,
+      "loss": 0.2239,
+      "step": 10604
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.564108023459796e-05,
+      "loss": 0.2159,
+      "step": 10605
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.562333052329111e-05,
+      "loss": 0.2279,
+      "step": 10606
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.560558324400369e-05,
+      "loss": 0.2185,
+      "step": 10607
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.558783839752942e-05,
+      "loss": 0.2342,
+      "step": 10608
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.557009598466192e-05,
+      "loss": 0.2169,
+      "step": 10609
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.555235600619481e-05,
+      "loss": 0.2363,
+      "step": 10610
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.553461846292141e-05,
+      "loss": 0.2306,
+      "step": 10611
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.551688335563514e-05,
+      "loss": 0.2155,
+      "step": 10612
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5499150685129135e-05,
+      "loss": 0.229,
+      "step": 10613
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5481420452196565e-05,
+      "loss": 0.2233,
+      "step": 10614
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5463692657630384e-05,
+      "loss": 0.2247,
+      "step": 10615
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.544596730222344e-05,
+      "loss": 0.2292,
+      "step": 10616
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.542824438676857e-05,
+      "loss": 0.2321,
+      "step": 10617
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.541052391205839e-05,
+      "loss": 0.2326,
+      "step": 10618
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.53928058788855e-05,
+      "loss": 0.2242,
+      "step": 10619
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.53750902880423e-05,
+      "loss": 0.2214,
+      "step": 10620
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5357377140321176e-05,
+      "loss": 0.2091,
+      "step": 10621
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5339666436514305e-05,
+      "loss": 0.2282,
+      "step": 10622
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.532195817741387e-05,
+      "loss": 0.2372,
+      "step": 10623
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5304252363811806e-05,
+      "loss": 0.2186,
+      "step": 10624
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5286548996500075e-05,
+      "loss": 0.2218,
+      "step": 10625
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.526884807627041e-05,
+      "loss": 0.2211,
+      "step": 10626
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.525114960391457e-05,
+      "loss": 0.2268,
+      "step": 10627
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.523345358022407e-05,
+      "loss": 0.2098,
+      "step": 10628
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.521576000599038e-05,
+      "loss": 0.2221,
+      "step": 10629
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5198068882004826e-05,
+      "loss": 0.2239,
+      "step": 10630
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5180380209058726e-05,
+      "loss": 0.2115,
+      "step": 10631
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5162693987943114e-05,
+      "loss": 0.2252,
+      "step": 10632
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5145010219449116e-05,
+      "loss": 0.2207,
+      "step": 10633
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.512732890436755e-05,
+      "loss": 0.2357,
+      "step": 10634
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.510965004348932e-05,
+      "loss": 0.2292,
+      "step": 10635
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5091973637605014e-05,
+      "loss": 0.2188,
+      "step": 10636
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5074299687505303e-05,
+      "loss": 0.2316,
+      "step": 10637
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.505662819398061e-05,
+      "loss": 0.2367,
+      "step": 10638
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.503895915782135e-05,
+      "loss": 0.2371,
+      "step": 10639
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.502129257981771e-05,
+      "loss": 0.2253,
+      "step": 10640
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5003628460759916e-05,
+      "loss": 0.2433,
+      "step": 10641
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4985966801437915e-05,
+      "loss": 0.2152,
+      "step": 10642
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4968307602641716e-05,
+      "loss": 0.2121,
+      "step": 10643
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.495065086516106e-05,
+      "loss": 0.2235,
+      "step": 10644
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4932996589785715e-05,
+      "loss": 0.222,
+      "step": 10645
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.491534477730522e-05,
+      "loss": 0.2278,
+      "step": 10646
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.489769542850916e-05,
+      "loss": 0.2189,
+      "step": 10647
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.488004854418676e-05,
+      "loss": 0.2241,
+      "step": 10648
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.486240412512741e-05,
+      "loss": 0.2447,
+      "step": 10649
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.484476217212017e-05,
+      "loss": 0.2322,
+      "step": 10650
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.482712268595416e-05,
+      "loss": 0.2177,
+      "step": 10651
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.480948566741826e-05,
+      "loss": 0.2288,
+      "step": 10652
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4791851117301344e-05,
+      "loss": 0.224,
+      "step": 10653
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.477421903639206e-05,
+      "loss": 0.2213,
+      "step": 10654
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.475658942547908e-05,
+      "loss": 0.2237,
+      "step": 10655
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.473896228535083e-05,
+      "loss": 0.2284,
+      "step": 10656
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.472133761679577e-05,
+      "loss": 0.2315,
+      "step": 10657
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4703715420602075e-05,
+      "loss": 0.2323,
+      "step": 10658
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4686095697558e-05,
+      "loss": 0.2304,
+      "step": 10659
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.46684784484515e-05,
+      "loss": 0.2269,
+      "step": 10660
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.465086367407061e-05,
+      "loss": 0.2488,
+      "step": 10661
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.463325137520307e-05,
+      "loss": 0.2331,
+      "step": 10662
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.461564155263668e-05,
+      "loss": 0.2103,
+      "step": 10663
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.459803420715896e-05,
+      "loss": 0.232,
+      "step": 10664
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4580429339557494e-05,
+      "loss": 0.2046,
+      "step": 10665
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4562826950619585e-05,
+      "loss": 0.2392,
+      "step": 10666
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4545227041132634e-05,
+      "loss": 0.2274,
+      "step": 10667
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.452762961188363e-05,
+      "loss": 0.2281,
+      "step": 10668
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.451003466365975e-05,
+      "loss": 0.2309,
+      "step": 10669
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.449244219724786e-05,
+      "loss": 0.2354,
+      "step": 10670
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.447485221343487e-05,
+      "loss": 0.246,
+      "step": 10671
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.445726471300742e-05,
+      "loss": 0.2874,
+      "step": 10672
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.443967969675219e-05,
+      "loss": 0.2408,
+      "step": 10673
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.442209716545561e-05,
+      "loss": 0.2825,
+      "step": 10674
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.440451711990413e-05,
+      "loss": 0.2495,
+      "step": 10675
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.438693956088397e-05,
+      "loss": 0.2628,
+      "step": 10676
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4369364489181345e-05,
+      "loss": 0.2479,
+      "step": 10677
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.435179190558224e-05,
+      "loss": 0.2533,
+      "step": 10678
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.433422181087268e-05,
+      "loss": 0.2498,
+      "step": 10679
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4316654205838415e-05,
+      "loss": 0.2463,
+      "step": 10680
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.429908909126523e-05,
+      "loss": 0.2451,
+      "step": 10681
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4281526467938714e-05,
+      "loss": 0.2357,
+      "step": 10682
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4263966336644316e-05,
+      "loss": 0.2444,
+      "step": 10683
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.424640869816749e-05,
+      "loss": 0.2479,
+      "step": 10684
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4228853553293425e-05,
+      "loss": 0.224,
+      "step": 10685
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421130090280737e-05,
+      "loss": 0.2423,
+      "step": 10686
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4193750747494345e-05,
+      "loss": 0.2347,
+      "step": 10687
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4176203088139234e-05,
+      "loss": 0.2349,
+      "step": 10688
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.415865792552696e-05,
+      "loss": 0.241,
+      "step": 10689
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.414111526044214e-05,
+      "loss": 0.2435,
+      "step": 10690
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4123575093669454e-05,
+      "loss": 0.2303,
+      "step": 10691
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.410603742599333e-05,
+      "loss": 0.2345,
+      "step": 10692
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4088502258198216e-05,
+      "loss": 0.2276,
+      "step": 10693
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.407096959106835e-05,
+      "loss": 0.2293,
+      "step": 10694
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.405343942538784e-05,
+      "loss": 0.229,
+      "step": 10695
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.403591176194082e-05,
+      "loss": 0.2216,
+      "step": 10696
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.401838660151112e-05,
+      "loss": 0.2252,
+      "step": 10697
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.400086394488267e-05,
+      "loss": 0.2263,
+      "step": 10698
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.398334379283906e-05,
+      "loss": 0.2233,
+      "step": 10699
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3965826146164e-05,
+      "loss": 0.2179,
+      "step": 10700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3948311005640884e-05,
+      "loss": 0.2272,
+      "step": 10701
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.393079837205316e-05,
+      "loss": 0.2412,
+      "step": 10702
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.391328824618401e-05,
+      "loss": 0.2309,
+      "step": 10703
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.389578062881667e-05,
+      "loss": 0.2289,
+      "step": 10704
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.387827552073409e-05,
+      "loss": 0.2121,
+      "step": 10705
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.386077292271926e-05,
+      "loss": 0.2057,
+      "step": 10706
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.384327283555496e-05,
+      "loss": 0.2309,
+      "step": 10707
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3825775260023903e-05,
+      "loss": 0.2094,
+      "step": 10708
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3808280196908615e-05,
+      "loss": 0.2234,
+      "step": 10709
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.379078764699166e-05,
+      "loss": 0.2296,
+      "step": 10710
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3773297611055316e-05,
+      "loss": 0.2328,
+      "step": 10711
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.375581008988192e-05,
+      "loss": 0.2298,
+      "step": 10712
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.373832508425353e-05,
+      "loss": 0.2381,
+      "step": 10713
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.372084259495223e-05,
+      "loss": 0.2299,
+      "step": 10714
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.370336262275987e-05,
+      "loss": 0.2209,
+      "step": 10715
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.368588516845832e-05,
+      "loss": 0.2242,
+      "step": 10716
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3668410232829195e-05,
+      "loss": 0.2228,
+      "step": 10717
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3650937816654145e-05,
+      "loss": 0.2269,
+      "step": 10718
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.363346792071455e-05,
+      "loss": 0.2263,
+      "step": 10719
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3616000545791836e-05,
+      "loss": 0.2241,
+      "step": 10720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3598535692667164e-05,
+      "loss": 0.2242,
+      "step": 10721
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.358107336212174e-05,
+      "loss": 0.2285,
+      "step": 10722
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3563613554936474e-05,
+      "loss": 0.22,
+      "step": 10723
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.354615627189237e-05,
+      "loss": 0.2314,
+      "step": 10724
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.352870151377012e-05,
+      "loss": 0.2121,
+      "step": 10725
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.351124928135046e-05,
+      "loss": 0.2315,
+      "step": 10726
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.349379957541394e-05,
+      "loss": 0.2214,
+      "step": 10727
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.347635239674098e-05,
+      "loss": 0.2276,
+      "step": 10728
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3458907746111896e-05,
+      "loss": 0.23,
+      "step": 10729
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.344146562430697e-05,
+      "loss": 0.2066,
+      "step": 10730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.342402603210622e-05,
+      "loss": 0.2286,
+      "step": 10731
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.340658897028974e-05,
+      "loss": 0.2181,
+      "step": 10732
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3389154439637315e-05,
+      "loss": 0.2211,
+      "step": 10733
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.33717224409288e-05,
+      "loss": 0.2249,
+      "step": 10734
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.335429297494377e-05,
+      "loss": 0.2153,
+      "step": 10735
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.333686604246183e-05,
+      "loss": 0.2176,
+      "step": 10736
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.331944164426235e-05,
+      "loss": 0.2285,
+      "step": 10737
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.33020197811247e-05,
+      "loss": 0.2165,
+      "step": 10738
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.328460045382802e-05,
+      "loss": 0.2263,
+      "step": 10739
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3267183663151464e-05,
+      "loss": 0.222,
+      "step": 10740
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.324976940987393e-05,
+      "loss": 0.2359,
+      "step": 10741
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.323235769477435e-05,
+      "loss": 0.222,
+      "step": 10742
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.321494851863141e-05,
+      "loss": 0.222,
+      "step": 10743
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3197541882223805e-05,
+      "loss": 0.2134,
+      "step": 10744
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.318013778632998e-05,
+      "loss": 0.2097,
+      "step": 10745
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.316273623172842e-05,
+      "loss": 0.2207,
+      "step": 10746
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.314533721919737e-05,
+      "loss": 0.2275,
+      "step": 10747
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.312794074951502e-05,
+      "loss": 0.2256,
+      "step": 10748
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3110546823459396e-05,
+      "loss": 0.2195,
+      "step": 10749
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.309315544180851e-05,
+      "loss": 0.2236,
+      "step": 10750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.307576660534014e-05,
+      "loss": 0.2204,
+      "step": 10751
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.305838031483207e-05,
+      "loss": 0.219,
+      "step": 10752
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.304099657106185e-05,
+      "loss": 0.2252,
+      "step": 10753
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.302361537480704e-05,
+      "loss": 0.2089,
+      "step": 10754
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.300623672684495e-05,
+      "loss": 0.2134,
+      "step": 10755
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.298886062795291e-05,
+      "loss": 0.2162,
+      "step": 10756
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.297148707890802e-05,
+      "loss": 0.2311,
+      "step": 10757
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2954116080487374e-05,
+      "loss": 0.2236,
+      "step": 10758
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2936747633467874e-05,
+      "loss": 0.2256,
+      "step": 10759
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2919381738626285e-05,
+      "loss": 0.2275,
+      "step": 10760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.290201839673938e-05,
+      "loss": 0.2215,
+      "step": 10761
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.288465760858368e-05,
+      "loss": 0.2192,
+      "step": 10762
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.286729937493571e-05,
+      "loss": 0.2244,
+      "step": 10763
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2849943696571756e-05,
+      "loss": 0.2174,
+      "step": 10764
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.283259057426814e-05,
+      "loss": 0.2303,
+      "step": 10765
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.281524000880091e-05,
+      "loss": 0.2192,
+      "step": 10766
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.279789200094614e-05,
+      "loss": 0.2183,
+      "step": 10767
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.278054655147969e-05,
+      "loss": 0.2327,
+      "step": 10768
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.276320366117733e-05,
+      "loss": 0.2267,
+      "step": 10769
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2745863330814774e-05,
+      "loss": 0.2151,
+      "step": 10770
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.272852556116755e-05,
+      "loss": 0.224,
+      "step": 10771
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2711190353011075e-05,
+      "loss": 0.226,
+      "step": 10772
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.269385770712072e-05,
+      "loss": 0.218,
+      "step": 10773
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2676527624271635e-05,
+      "loss": 0.22,
+      "step": 10774
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.265920010523899e-05,
+      "loss": 0.2289,
+      "step": 10775
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2641875150797696e-05,
+      "loss": 0.2189,
+      "step": 10776
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.262455276172268e-05,
+      "loss": 0.2117,
+      "step": 10777
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.260723293878862e-05,
+      "loss": 0.2173,
+      "step": 10778
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2589915682770245e-05,
+      "loss": 0.2196,
+      "step": 10779
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.257260099444198e-05,
+      "loss": 0.2271,
+      "step": 10780
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2555288874578326e-05,
+      "loss": 0.2161,
+      "step": 10781
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.253797932395348e-05,
+      "loss": 0.2209,
+      "step": 10782
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2520672343341706e-05,
+      "loss": 0.2286,
+      "step": 10783
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.250336793351699e-05,
+      "loss": 0.2193,
+      "step": 10784
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.248606609525334e-05,
+      "loss": 0.2211,
+      "step": 10785
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.246876682932455e-05,
+      "loss": 0.2303,
+      "step": 10786
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2451470136504404e-05,
+      "loss": 0.231,
+      "step": 10787
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.243417601756639e-05,
+      "loss": 0.2233,
+      "step": 10788
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.241688447328408e-05,
+      "loss": 0.2155,
+      "step": 10789
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.239959550443078e-05,
+      "loss": 0.2283,
+      "step": 10790
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.238230911177983e-05,
+      "loss": 0.2147,
+      "step": 10791
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.23650252961043e-05,
+      "loss": 0.2373,
+      "step": 10792
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.234774405817726e-05,
+      "loss": 0.2247,
+      "step": 10793
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2330465398771594e-05,
+      "loss": 0.2241,
+      "step": 10794
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.231318931866012e-05,
+      "loss": 0.2205,
+      "step": 10795
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2295915818615484e-05,
+      "loss": 0.2204,
+      "step": 10796
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.22786448994103e-05,
+      "loss": 0.2205,
+      "step": 10797
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.226137656181696e-05,
+      "loss": 0.2203,
+      "step": 10798
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.224411080660785e-05,
+      "loss": 0.2325,
+      "step": 10799
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.222684763455514e-05,
+      "loss": 0.2257,
+      "step": 10800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2209587046431e-05,
+      "loss": 0.2241,
+      "step": 10801
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.219232904300733e-05,
+      "loss": 0.2185,
+      "step": 10802
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2175073625056086e-05,
+      "loss": 0.2286,
+      "step": 10803
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2157820793348934e-05,
+      "loss": 0.2259,
+      "step": 10804
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.214057054865761e-05,
+      "loss": 0.223,
+      "step": 10805
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.212332289175356e-05,
+      "loss": 0.22,
+      "step": 10806
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2106077823408294e-05,
+      "loss": 0.2301,
+      "step": 10807
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2088835344392976e-05,
+      "loss": 0.2314,
+      "step": 10808
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.207159545547886e-05,
+      "loss": 0.2138,
+      "step": 10809
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.205435815743697e-05,
+      "loss": 0.2201,
+      "step": 10810
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2037123451038296e-05,
+      "loss": 0.236,
+      "step": 10811
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.201989133705361e-05,
+      "loss": 0.2269,
+      "step": 10812
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.20026618162537e-05,
+      "loss": 0.2242,
+      "step": 10813
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.198543488940908e-05,
+      "loss": 0.2223,
+      "step": 10814
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.196821055729032e-05,
+      "loss": 0.2154,
+      "step": 10815
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.195098882066769e-05,
+      "loss": 0.2284,
+      "step": 10816
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1933769680311544e-05,
+      "loss": 0.2302,
+      "step": 10817
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.191655313699191e-05,
+      "loss": 0.2362,
+      "step": 10818
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.189933919147889e-05,
+      "loss": 0.2302,
+      "step": 10819
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.188212784454233e-05,
+      "loss": 0.2251,
+      "step": 10820
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.186491909695206e-05,
+      "loss": 0.2085,
+      "step": 10821
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.184771294947768e-05,
+      "loss": 0.2219,
+      "step": 10822
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.183050940288884e-05,
+      "loss": 0.2074,
+      "step": 10823
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.181330845795487e-05,
+      "loss": 0.2181,
+      "step": 10824
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.179611011544519e-05,
+      "loss": 0.2286,
+      "step": 10825
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.177891437612894e-05,
+      "loss": 0.224,
+      "step": 10826
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.176172124077523e-05,
+      "loss": 0.2136,
+      "step": 10827
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.174453071015299e-05,
+      "loss": 0.2281,
+      "step": 10828
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.172734278503113e-05,
+      "loss": 0.2124,
+      "step": 10829
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.171015746617833e-05,
+      "loss": 0.2311,
+      "step": 10830
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.169297475436328e-05,
+      "loss": 0.2217,
+      "step": 10831
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1675794650354406e-05,
+      "loss": 0.2227,
+      "step": 10832
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.165861715492018e-05,
+      "loss": 0.2293,
+      "step": 10833
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.164144226882879e-05,
+      "loss": 0.231,
+      "step": 10834
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1624269992848465e-05,
+      "loss": 0.2203,
+      "step": 10835
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1607100327747165e-05,
+      "loss": 0.2169,
+      "step": 10836
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1589933274292894e-05,
+      "loss": 0.2199,
+      "step": 10837
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1572768833253404e-05,
+      "loss": 0.2222,
+      "step": 10838
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.155560700539637e-05,
+      "loss": 0.2179,
+      "step": 10839
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1538447791489396e-05,
+      "loss": 0.2309,
+      "step": 10840
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1521291192299905e-05,
+      "loss": 0.2275,
+      "step": 10841
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.150413720859528e-05,
+      "loss": 0.2223,
+      "step": 10842
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.148698584114268e-05,
+      "loss": 0.2169,
+      "step": 10843
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1469837090709275e-05,
+      "loss": 0.2359,
+      "step": 10844
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.145269095806197e-05,
+      "loss": 0.2181,
+      "step": 10845
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1435547443967707e-05,
+      "loss": 0.2274,
+      "step": 10846
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.141840654919321e-05,
+      "loss": 0.2158,
+      "step": 10847
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.140126827450507e-05,
+      "loss": 0.2153,
+      "step": 10848
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.138413262066988e-05,
+      "loss": 0.2348,
+      "step": 10849
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1366999588454005e-05,
+      "loss": 0.2167,
+      "step": 10850
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.134986917862368e-05,
+      "loss": 0.2212,
+      "step": 10851
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1332741391945154e-05,
+      "loss": 0.2307,
+      "step": 10852
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.131561622918439e-05,
+      "loss": 0.2285,
+      "step": 10853
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.129849369110741e-05,
+      "loss": 0.2391,
+      "step": 10854
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.128137377847993e-05,
+      "loss": 0.2197,
+      "step": 10855
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1264256492067735e-05,
+      "loss": 0.2131,
+      "step": 10856
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.124714183263633e-05,
+      "loss": 0.2134,
+      "step": 10857
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1230029800951234e-05,
+      "loss": 0.218,
+      "step": 10858
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.121292039777772e-05,
+      "loss": 0.221,
+      "step": 10859
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1195813623881106e-05,
+      "loss": 0.2079,
+      "step": 10860
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.117870948002641e-05,
+      "loss": 0.2141,
+      "step": 10861
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.116160796697869e-05,
+      "loss": 0.2299,
+      "step": 10862
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1144509085502755e-05,
+      "loss": 0.2213,
+      "step": 10863
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1127412836363444e-05,
+      "loss": 0.2212,
+      "step": 10864
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.11103192203253e-05,
+      "loss": 0.226,
+      "step": 10865
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.109322823815291e-05,
+      "loss": 0.2317,
+      "step": 10866
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.107613989061066e-05,
+      "loss": 0.2197,
+      "step": 10867
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.105905417846283e-05,
+      "loss": 0.2117,
+      "step": 10868
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.104197110247354e-05,
+      "loss": 0.23,
+      "step": 10869
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1024890663406925e-05,
+      "loss": 0.2198,
+      "step": 10870
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.100781286202683e-05,
+      "loss": 0.2279,
+      "step": 10871
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.099073769909713e-05,
+      "loss": 0.2276,
+      "step": 10872
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.097366517538146e-05,
+      "loss": 0.2166,
+      "step": 10873
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.095659529164347e-05,
+      "loss": 0.2152,
+      "step": 10874
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0939528048646545e-05,
+      "loss": 0.221,
+      "step": 10875
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.092246344715409e-05,
+      "loss": 0.2251,
+      "step": 10876
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.090540148792926e-05,
+      "loss": 0.2184,
+      "step": 10877
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0888342171735226e-05,
+      "loss": 0.2105,
+      "step": 10878
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0871285499334896e-05,
+      "loss": 0.2225,
+      "step": 10879
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.085423147149123e-05,
+      "loss": 0.2317,
+      "step": 10880
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.083718008896689e-05,
+      "loss": 0.2311,
+      "step": 10881
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.082013135252457e-05,
+      "loss": 0.2223,
+      "step": 10882
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.080308526292672e-05,
+      "loss": 0.2143,
+      "step": 10883
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0786041820935806e-05,
+      "loss": 0.2204,
+      "step": 10884
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0769001027314026e-05,
+      "loss": 0.216,
+      "step": 10885
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0751962882823614e-05,
+      "loss": 0.2065,
+      "step": 10886
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0734927388226576e-05,
+      "loss": 0.227,
+      "step": 10887
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.071789454428482e-05,
+      "loss": 0.2105,
+      "step": 10888
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0700864351760115e-05,
+      "loss": 0.2252,
+      "step": 10889
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.068383681141422e-05,
+      "loss": 0.2155,
+      "step": 10890
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.066681192400863e-05,
+      "loss": 0.2332,
+      "step": 10891
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.064978969030485e-05,
+      "loss": 0.2224,
+      "step": 10892
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0632770111064145e-05,
+      "loss": 0.2119,
+      "step": 10893
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.061575318704779e-05,
+      "loss": 0.2189,
+      "step": 10894
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.05987389190168e-05,
+      "loss": 0.2246,
+      "step": 10895
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.058172730773222e-05,
+      "loss": 0.2248,
+      "step": 10896
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.056471835395483e-05,
+      "loss": 0.2368,
+      "step": 10897
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.054771205844544e-05,
+      "loss": 0.2218,
+      "step": 10898
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.053070842196459e-05,
+      "loss": 0.2256,
+      "step": 10899
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.051370744527283e-05,
+      "loss": 0.2062,
+      "step": 10900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.04967091291305e-05,
+      "loss": 0.218,
+      "step": 10901
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0479713474297874e-05,
+      "loss": 0.2088,
+      "step": 10902
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.046272048153511e-05,
+      "loss": 0.2092,
+      "step": 10903
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.044573015160216e-05,
+      "loss": 0.2328,
+      "step": 10904
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0428742485258985e-05,
+      "loss": 0.2256,
+      "step": 10905
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.041175748326533e-05,
+      "loss": 0.2184,
+      "step": 10906
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.03947751463809e-05,
+      "loss": 0.2331,
+      "step": 10907
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.037779547536521e-05,
+      "loss": 0.2167,
+      "step": 10908
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.036081847097766e-05,
+      "loss": 0.2246,
+      "step": 10909
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.034384413397759e-05,
+      "loss": 0.2236,
+      "step": 10910
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0326872465124153e-05,
+      "loss": 0.2137,
+      "step": 10911
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0309903465176454e-05,
+      "loss": 0.2288,
+      "step": 10912
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0292937134893395e-05,
+      "loss": 0.2284,
+      "step": 10913
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0275973475033856e-05,
+      "loss": 0.2025,
+      "step": 10914
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.025901248635651e-05,
+      "loss": 0.2216,
+      "step": 10915
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0242054169619906e-05,
+      "loss": 0.2224,
+      "step": 10916
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.022509852558259e-05,
+      "loss": 0.2123,
+      "step": 10917
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.020814555500284e-05,
+      "loss": 0.2295,
+      "step": 10918
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0191195258638944e-05,
+      "loss": 0.2201,
+      "step": 10919
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.017424763724895e-05,
+      "loss": 0.2226,
+      "step": 10920
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0157302691590924e-05,
+      "loss": 0.2211,
+      "step": 10921
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.014036042242265e-05,
+      "loss": 0.2242,
+      "step": 10922
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0123420830501965e-05,
+      "loss": 0.2103,
+      "step": 10923
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.010648391658642e-05,
+      "loss": 0.2168,
+      "step": 10924
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.008954968143359e-05,
+      "loss": 0.2242,
+      "step": 10925
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.007261812580081e-05,
+      "loss": 0.2169,
+      "step": 10926
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.005568925044544e-05,
+      "loss": 0.2118,
+      "step": 10927
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0038763056124504e-05,
+      "loss": 0.2318,
+      "step": 10928
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.002183954359513e-05,
+      "loss": 0.228,
+      "step": 10929
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.000491871361417e-05,
+      "loss": 0.2257,
+      "step": 10930
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.998800056693847e-05,
+      "loss": 0.2241,
+      "step": 10931
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9971085104324637e-05,
+      "loss": 0.2319,
+      "step": 10932
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9954172326529305e-05,
+      "loss": 0.2169,
+      "step": 10933
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.993726223430881e-05,
+      "loss": 0.2117,
+      "step": 10934
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.992035482841956e-05,
+      "loss": 0.2261,
+      "step": 10935
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9903450109617656e-05,
+      "loss": 0.2257,
+      "step": 10936
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.988654807865925e-05,
+      "loss": 0.2299,
+      "step": 10937
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9869648736300215e-05,
+      "loss": 0.2235,
+      "step": 10938
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.985275208329645e-05,
+      "loss": 0.221,
+      "step": 10939
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98358581204036e-05,
+      "loss": 0.2255,
+      "step": 10940
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.981896684837734e-05,
+      "loss": 0.2218,
+      "step": 10941
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.980207826797303e-05,
+      "loss": 0.229,
+      "step": 10942
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9785192379946134e-05,
+      "loss": 0.221,
+      "step": 10943
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.976830918505178e-05,
+      "loss": 0.2096,
+      "step": 10944
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.975142868404514e-05,
+      "loss": 0.2232,
+      "step": 10945
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.973455087768117e-05,
+      "loss": 0.2139,
+      "step": 10946
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.97176757667148e-05,
+      "loss": 0.2228,
+      "step": 10947
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9700803351900664e-05,
+      "loss": 0.208,
+      "step": 10948
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.968393363399347e-05,
+      "loss": 0.22,
+      "step": 10949
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9667066613747685e-05,
+      "loss": 0.222,
+      "step": 10950
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.965020229191773e-05,
+      "loss": 0.2136,
+      "step": 10951
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.963334066925782e-05,
+      "loss": 0.2093,
+      "step": 10952
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.961648174652216e-05,
+      "loss": 0.2264,
+      "step": 10953
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.959962552446471e-05,
+      "loss": 0.2291,
+      "step": 10954
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.958277200383943e-05,
+      "loss": 0.2211,
+      "step": 10955
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.956592118540003e-05,
+      "loss": 0.2128,
+      "step": 10956
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9549073069900266e-05,
+      "loss": 0.2275,
+      "step": 10957
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9532227658093576e-05,
+      "loss": 0.2102,
+      "step": 10958
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.951538495073346e-05,
+      "loss": 0.2178,
+      "step": 10959
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9498544948573145e-05,
+      "loss": 0.2177,
+      "step": 10960
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.948170765236588e-05,
+      "loss": 0.2279,
+      "step": 10961
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9464873062864646e-05,
+      "loss": 0.227,
+      "step": 10962
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.944804118082245e-05,
+      "loss": 0.2189,
+      "step": 10963
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9431212006992036e-05,
+      "loss": 0.2348,
+      "step": 10964
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.941438554212615e-05,
+      "loss": 0.2175,
+      "step": 10965
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.939756178697732e-05,
+      "loss": 0.2217,
+      "step": 10966
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9380740742298086e-05,
+      "loss": 0.2207,
+      "step": 10967
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9363922408840636e-05,
+      "loss": 0.2239,
+      "step": 10968
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.934710678735728e-05,
+      "loss": 0.2223,
+      "step": 10969
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9330293878600035e-05,
+      "loss": 0.2106,
+      "step": 10970
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.931348368332094e-05,
+      "loss": 0.2193,
+      "step": 10971
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9296676202271766e-05,
+      "loss": 0.2285,
+      "step": 10972
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.92798714362043e-05,
+      "loss": 0.2277,
+      "step": 10973
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9263069385870076e-05,
+      "loss": 0.231,
+      "step": 10974
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.924627005202064e-05,
+      "loss": 0.2218,
+      "step": 10975
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.922947343540729e-05,
+      "loss": 0.2276,
+      "step": 10976
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.921267953678133e-05,
+      "loss": 0.2184,
+      "step": 10977
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9195888356893784e-05,
+      "loss": 0.229,
+      "step": 10978
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.917909989649573e-05,
+      "loss": 0.2261,
+      "step": 10979
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.916231415633801e-05,
+      "loss": 0.2304,
+      "step": 10980
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.914553113717132e-05,
+      "loss": 0.2208,
+      "step": 10981
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.912875083974638e-05,
+      "loss": 0.2275,
+      "step": 10982
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.91119732648136e-05,
+      "loss": 0.2225,
+      "step": 10983
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.909519841312347e-05,
+      "loss": 0.2259,
+      "step": 10984
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.907842628542615e-05,
+      "loss": 0.2301,
+      "step": 10985
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9061656882471856e-05,
+      "loss": 0.214,
+      "step": 10986
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.904489020501058e-05,
+      "loss": 0.221,
+      "step": 10987
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9028126253792184e-05,
+      "loss": 0.2304,
+      "step": 10988
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9011365029566506e-05,
+      "loss": 0.2133,
+      "step": 10989
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.899460653308313e-05,
+      "loss": 0.2349,
+      "step": 10990
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.897785076509166e-05,
+      "loss": 0.2339,
+      "step": 10991
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.896109772634143e-05,
+      "loss": 0.2048,
+      "step": 10992
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.89443474175818e-05,
+      "loss": 0.2318,
+      "step": 10993
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8927599839561914e-05,
+      "loss": 0.2253,
+      "step": 10994
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.891085499303075e-05,
+      "loss": 0.2264,
+      "step": 10995
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.889411287873732e-05,
+      "loss": 0.229,
+      "step": 10996
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8877373497430345e-05,
+      "loss": 0.2261,
+      "step": 10997
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.886063684985858e-05,
+      "loss": 0.231,
+      "step": 10998
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8843902936770485e-05,
+      "loss": 0.2201,
+      "step": 10999
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8827171758914586e-05,
+      "loss": 0.2206,
+      "step": 11000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.88104433170391e-05,
+      "loss": 0.2262,
+      "step": 11001
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.879371761189231e-05,
+      "loss": 0.2224,
+      "step": 11002
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.87769946442222e-05,
+      "loss": 0.2335,
+      "step": 11003
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8760274414776765e-05,
+      "loss": 0.2252,
+      "step": 11004
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.874355692430377e-05,
+      "loss": 0.223,
+      "step": 11005
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8726842173550984e-05,
+      "loss": 0.2153,
+      "step": 11006
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8710130163265935e-05,
+      "loss": 0.2225,
+      "step": 11007
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8693420894196087e-05,
+      "loss": 0.2224,
+      "step": 11008
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.867671436708873e-05,
+      "loss": 0.224,
+      "step": 11009
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.866001058269113e-05,
+      "loss": 0.2245,
+      "step": 11010
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8643309541750314e-05,
+      "loss": 0.2261,
+      "step": 11011
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.862661124501331e-05,
+      "loss": 0.2126,
+      "step": 11012
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.860991569322688e-05,
+      "loss": 0.2255,
+      "step": 11013
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.859322288713781e-05,
+      "loss": 0.2171,
+      "step": 11014
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.857653282749264e-05,
+      "loss": 0.2188,
+      "step": 11015
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8559845515037895e-05,
+      "loss": 0.2281,
+      "step": 11016
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.854316095051985e-05,
+      "loss": 0.2167,
+      "step": 11017
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.85264791346848e-05,
+      "loss": 0.2308,
+      "step": 11018
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.850980006827878e-05,
+      "loss": 0.2249,
+      "step": 11019
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.849312375204785e-05,
+      "loss": 0.2211,
+      "step": 11020
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8476450186737764e-05,
+      "loss": 0.2251,
+      "step": 11021
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8459779373094354e-05,
+      "loss": 0.2253,
+      "step": 11022
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8443111311863147e-05,
+      "loss": 0.2144,
+      "step": 11023
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.842644600378971e-05,
+      "loss": 0.226,
+      "step": 11024
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8409783449619305e-05,
+      "loss": 0.2356,
+      "step": 11025
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8393123650097285e-05,
+      "loss": 0.2126,
+      "step": 11026
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.83764666059687e-05,
+      "loss": 0.2219,
+      "step": 11027
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.835981231797855e-05,
+      "loss": 0.2275,
+      "step": 11028
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8343160786871666e-05,
+      "loss": 0.2189,
+      "step": 11029
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.832651201339289e-05,
+      "loss": 0.2164,
+      "step": 11030
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.830986599828674e-05,
+      "loss": 0.2114,
+      "step": 11031
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.829322274229782e-05,
+      "loss": 0.2132,
+      "step": 11032
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.827658224617041e-05,
+      "loss": 0.2127,
+      "step": 11033
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.825994451064884e-05,
+      "loss": 0.2233,
+      "step": 11034
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.824330953647718e-05,
+      "loss": 0.2245,
+      "step": 11035
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.82266773243995e-05,
+      "loss": 0.2234,
+      "step": 11036
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8210047875159595e-05,
+      "loss": 0.2177,
+      "step": 11037
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8193421189501325e-05,
+      "loss": 0.2326,
+      "step": 11038
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.817679726816824e-05,
+      "loss": 0.2254,
+      "step": 11039
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8160176111903923e-05,
+      "loss": 0.2222,
+      "step": 11040
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.814355772145168e-05,
+      "loss": 0.2224,
+      "step": 11041
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8126942097554876e-05,
+      "loss": 0.2258,
+      "step": 11042
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.811032924095655e-05,
+      "loss": 0.2312,
+      "step": 11043
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.809371915239981e-05,
+      "loss": 0.2264,
+      "step": 11044
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.807711183262746e-05,
+      "loss": 0.2104,
+      "step": 11045
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.806050728238237e-05,
+      "loss": 0.2188,
+      "step": 11046
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.804390550240712e-05,
+      "loss": 0.2193,
+      "step": 11047
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8027306493444245e-05,
+      "loss": 0.222,
+      "step": 11048
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8010710256236105e-05,
+      "loss": 0.2147,
+      "step": 11049
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.799411679152505e-05,
+      "loss": 0.2248,
+      "step": 11050
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.797752610005314e-05,
+      "loss": 0.2295,
+      "step": 11051
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.79609381825625e-05,
+      "loss": 0.2164,
+      "step": 11052
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.794435303979494e-05,
+      "loss": 0.1995,
+      "step": 11053
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.792777067249231e-05,
+      "loss": 0.224,
+      "step": 11054
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.79111910813962e-05,
+      "loss": 0.2268,
+      "step": 11055
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7894614267248205e-05,
+      "loss": 0.2104,
+      "step": 11056
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7878040230789666e-05,
+      "loss": 0.2163,
+      "step": 11057
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7861468972761926e-05,
+      "loss": 0.2301,
+      "step": 11058
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.784490049390611e-05,
+      "loss": 0.2244,
+      "step": 11059
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7828334794963214e-05,
+      "loss": 0.2303,
+      "step": 11060
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.781177187667422e-05,
+      "loss": 0.2097,
+      "step": 11061
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.779521173977984e-05,
+      "loss": 0.226,
+      "step": 11062
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.77786543850208e-05,
+      "loss": 0.2272,
+      "step": 11063
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7762099813137555e-05,
+      "loss": 0.2302,
+      "step": 11064
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7745548024870606e-05,
+      "loss": 0.2232,
+      "step": 11065
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.772899902096016e-05,
+      "loss": 0.2298,
+      "step": 11066
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.771245280214644e-05,
+      "loss": 0.2224,
+      "step": 11067
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7695909369169455e-05,
+      "loss": 0.232,
+      "step": 11068
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.767936872276908e-05,
+      "loss": 0.2228,
+      "step": 11069
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7662830863685185e-05,
+      "loss": 0.2254,
+      "step": 11070
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7646295792657375e-05,
+      "loss": 0.2206,
+      "step": 11071
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7629763510425167e-05,
+      "loss": 0.2103,
+      "step": 11072
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.761323401772805e-05,
+      "loss": 0.2164,
+      "step": 11073
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.759670731530523e-05,
+      "loss": 0.2202,
+      "step": 11074
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.758018340389593e-05,
+      "loss": 0.2302,
+      "step": 11075
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7563662284239155e-05,
+      "loss": 0.2208,
+      "step": 11076
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7547143957073846e-05,
+      "loss": 0.2121,
+      "step": 11077
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.753062842313877e-05,
+      "loss": 0.2084,
+      "step": 11078
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7514115683172616e-05,
+      "loss": 0.2101,
+      "step": 11079
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.749760573791388e-05,
+      "loss": 0.2091,
+      "step": 11080
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.748109858810104e-05,
+      "loss": 0.2102,
+      "step": 11081
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7464594234472303e-05,
+      "loss": 0.2172,
+      "step": 11082
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.744809267776593e-05,
+      "loss": 0.2151,
+      "step": 11083
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.743159391871986e-05,
+      "loss": 0.2204,
+      "step": 11084
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.74150979580721e-05,
+      "loss": 0.2712,
+      "step": 11085
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7398604796560364e-05,
+      "loss": 0.2265,
+      "step": 11086
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.738211443492241e-05,
+      "loss": 0.223,
+      "step": 11087
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.736562687389564e-05,
+      "loss": 0.2177,
+      "step": 11088
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7349142114217574e-05,
+      "loss": 0.2247,
+      "step": 11089
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.733266015662543e-05,
+      "loss": 0.2127,
+      "step": 11090
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7316181001856454e-05,
+      "loss": 0.2111,
+      "step": 11091
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7299704650647595e-05,
+      "loss": 0.2224,
+      "step": 11092
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.728323110373583e-05,
+      "loss": 0.2372,
+      "step": 11093
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.726676036185789e-05,
+      "loss": 0.2154,
+      "step": 11094
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.725029242575051e-05,
+      "loss": 0.2092,
+      "step": 11095
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.723382729615014e-05,
+      "loss": 0.2161,
+      "step": 11096
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.721736497379326e-05,
+      "loss": 0.2176,
+      "step": 11097
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.72009054594161e-05,
+      "loss": 0.2109,
+      "step": 11098
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7184448753754874e-05,
+      "loss": 0.2182,
+      "step": 11099
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.716799485754555e-05,
+      "loss": 0.2197,
+      "step": 11100
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.715154377152411e-05,
+      "loss": 0.231,
+      "step": 11101
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.713509549642627e-05,
+      "loss": 0.2153,
+      "step": 11102
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.711865003298772e-05,
+      "loss": 0.2128,
+      "step": 11103
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.710220738194398e-05,
+      "loss": 0.2138,
+      "step": 11104
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.708576754403047e-05,
+      "loss": 0.2244,
+      "step": 11105
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.706933051998242e-05,
+      "loss": 0.2189,
+      "step": 11106
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7052896310535114e-05,
+      "loss": 0.2187,
+      "step": 11107
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7036464916423396e-05,
+      "loss": 0.2232,
+      "step": 11108
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.702003633838229e-05,
+      "loss": 0.2123,
+      "step": 11109
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.70036105771465e-05,
+      "loss": 0.2256,
+      "step": 11110
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6987187633450745e-05,
+      "loss": 0.2251,
+      "step": 11111
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.697076750802948e-05,
+      "loss": 0.2214,
+      "step": 11112
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.695435020161718e-05,
+      "loss": 0.2157,
+      "step": 11113
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6937935714948025e-05,
+      "loss": 0.2172,
+      "step": 11114
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.692152404875625e-05,
+      "loss": 0.204,
+      "step": 11115
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6905115203775786e-05,
+      "loss": 0.2214,
+      "step": 11116
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6888709180740625e-05,
+      "loss": 0.211,
+      "step": 11117
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6872305980384424e-05,
+      "loss": 0.2277,
+      "step": 11118
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.685590560344091e-05,
+      "loss": 0.2096,
+      "step": 11119
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6839508050643545e-05,
+      "loss": 0.2095,
+      "step": 11120
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.682311332272575e-05,
+      "loss": 0.21,
+      "step": 11121
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6806721420420757e-05,
+      "loss": 0.2069,
+      "step": 11122
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6790332344461743e-05,
+      "loss": 0.2157,
+      "step": 11123
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.677394609558169e-05,
+      "loss": 0.2217,
+      "step": 11124
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6757562674513435e-05,
+      "loss": 0.2111,
+      "step": 11125
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6741182081989825e-05,
+      "loss": 0.2134,
+      "step": 11126
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.672480431874342e-05,
+      "loss": 0.2169,
+      "step": 11127
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6708429385506726e-05,
+      "loss": 0.2072,
+      "step": 11128
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.669205728301217e-05,
+      "loss": 0.2325,
+      "step": 11129
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6675688011991926e-05,
+      "loss": 0.213,
+      "step": 11130
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.66593215731782e-05,
+      "loss": 0.2238,
+      "step": 11131
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.66429579673029e-05,
+      "loss": 0.2025,
+      "step": 11132
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.662659719509798e-05,
+      "loss": 0.2226,
+      "step": 11133
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.661023925729511e-05,
+      "loss": 0.2163,
+      "step": 11134
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.659388415462597e-05,
+      "loss": 0.2112,
+      "step": 11135
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6577531887822013e-05,
+      "loss": 0.2171,
+      "step": 11136
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.656118245761459e-05,
+      "loss": 0.2236,
+      "step": 11137
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.654483586473496e-05,
+      "loss": 0.2179,
+      "step": 11138
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6528492109914206e-05,
+      "loss": 0.2028,
+      "step": 11139
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.651215119388336e-05,
+      "loss": 0.2288,
+      "step": 11140
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.64958131173732e-05,
+      "loss": 0.2181,
+      "step": 11141
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.647947788111453e-05,
+      "loss": 0.2067,
+      "step": 11142
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.646314548583789e-05,
+      "loss": 0.2103,
+      "step": 11143
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6446815932273816e-05,
+      "loss": 0.2195,
+      "step": 11144
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.643048922115258e-05,
+      "loss": 0.2122,
+      "step": 11145
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6414165353204464e-05,
+      "loss": 0.2174,
+      "step": 11146
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.639784432915954e-05,
+      "loss": 0.2134,
+      "step": 11147
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6381526149747736e-05,
+      "loss": 0.219,
+      "step": 11148
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6365210815698945e-05,
+      "loss": 0.2147,
+      "step": 11149
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6348898327742866e-05,
+      "loss": 0.2183,
+      "step": 11150
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6332588686609026e-05,
+      "loss": 0.2266,
+      "step": 11151
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.631628189302696e-05,
+      "loss": 0.2193,
+      "step": 11152
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.629997794772592e-05,
+      "loss": 0.2007,
+      "step": 11153
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6283676851435175e-05,
+      "loss": 0.2178,
+      "step": 11154
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.626737860488375e-05,
+      "loss": 0.2187,
+      "step": 11155
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.625108320880064e-05,
+      "loss": 0.2198,
+      "step": 11156
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6234790663914595e-05,
+      "loss": 0.2276,
+      "step": 11157
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.621850097095438e-05,
+      "loss": 0.2214,
+      "step": 11158
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6202214130648505e-05,
+      "loss": 0.2064,
+      "step": 11159
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.618593014372544e-05,
+      "loss": 0.2222,
+      "step": 11160
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6169649010913444e-05,
+      "loss": 0.2223,
+      "step": 11161
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.615337073294077e-05,
+      "loss": 0.2235,
+      "step": 11162
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6137095310535394e-05,
+      "loss": 0.2353,
+      "step": 11163
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.612082274442532e-05,
+      "loss": 0.2143,
+      "step": 11164
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.610455303533826e-05,
+      "loss": 0.2161,
+      "step": 11165
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.608828618400196e-05,
+      "loss": 0.2,
+      "step": 11166
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.607202219114394e-05,
+      "loss": 0.2241,
+      "step": 11167
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.60557610574916e-05,
+      "loss": 0.232,
+      "step": 11168
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6039502783772195e-05,
+      "loss": 0.2093,
+      "step": 11169
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.602324737071294e-05,
+      "loss": 0.2068,
+      "step": 11170
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.600699481904083e-05,
+      "loss": 0.2243,
+      "step": 11171
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.599074512948279e-05,
+      "loss": 0.2198,
+      "step": 11172
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.597449830276557e-05,
+      "loss": 0.2211,
+      "step": 11173
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5958254339615846e-05,
+      "loss": 0.2106,
+      "step": 11174
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5942013240760094e-05,
+      "loss": 0.2299,
+      "step": 11175
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.592577500692478e-05,
+      "loss": 0.2174,
+      "step": 11176
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.590953963883606e-05,
+      "loss": 0.2195,
+      "step": 11177
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.589330713722017e-05,
+      "loss": 0.2115,
+      "step": 11178
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5877077502803026e-05,
+      "loss": 0.2072,
+      "step": 11179
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5860850736310593e-05,
+      "loss": 0.2068,
+      "step": 11180
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.584462683846853e-05,
+      "loss": 0.2152,
+      "step": 11181
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.582840581000254e-05,
+      "loss": 0.2201,
+      "step": 11182
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.581218765163804e-05,
+      "loss": 0.2382,
+      "step": 11183
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5795972364100484e-05,
+      "loss": 0.2084,
+      "step": 11184
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5779759948115e-05,
+      "loss": 0.2153,
+      "step": 11185
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.57635504044068e-05,
+      "loss": 0.2105,
+      "step": 11186
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.574734373370081e-05,
+      "loss": 0.2036,
+      "step": 11187
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.573113993672188e-05,
+      "loss": 0.2201,
+      "step": 11188
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5714939014194706e-05,
+      "loss": 0.2204,
+      "step": 11189
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5698740966843935e-05,
+      "loss": 0.2082,
+      "step": 11190
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.568254579539398e-05,
+      "loss": 0.2177,
+      "step": 11191
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.566635350056924e-05,
+      "loss": 0.2162,
+      "step": 11192
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.565016408309384e-05,
+      "loss": 0.2225,
+      "step": 11193
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.563397754369195e-05,
+      "loss": 0.2164,
+      "step": 11194
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.561779388308744e-05,
+      "loss": 0.2232,
+      "step": 11195
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.56016131020042e-05,
+      "loss": 0.2241,
+      "step": 11196
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.558543520116584e-05,
+      "loss": 0.2061,
+      "step": 11197
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5569260181296014e-05,
+      "loss": 0.2026,
+      "step": 11198
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.555308804311809e-05,
+      "loss": 0.2236,
+      "step": 11199
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.553691878735542e-05,
+      "loss": 0.2204,
+      "step": 11200
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.552075241473112e-05,
+      "loss": 0.2126,
+      "step": 11201
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.550458892596832e-05,
+      "loss": 0.2165,
+      "step": 11202
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.548842832178988e-05,
+      "loss": 0.2113,
+      "step": 11203
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.547227060291859e-05,
+      "loss": 0.2163,
+      "step": 11204
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5456115770077146e-05,
+      "loss": 0.2091,
+      "step": 11205
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.543996382398802e-05,
+      "loss": 0.2119,
+      "step": 11206
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5423814765373695e-05,
+      "loss": 0.21,
+      "step": 11207
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.54076685949564e-05,
+      "loss": 0.2288,
+      "step": 11208
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5391525313458254e-05,
+      "loss": 0.2194,
+      "step": 11209
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5375384921601326e-05,
+      "loss": 0.2228,
+      "step": 11210
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.535924742010746e-05,
+      "loss": 0.2051,
+      "step": 11211
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.534311280969844e-05,
+      "loss": 0.2178,
+      "step": 11212
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5326981091095856e-05,
+      "loss": 0.2148,
+      "step": 11213
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5310852265021264e-05,
+      "loss": 0.2211,
+      "step": 11214
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5294726332196006e-05,
+      "loss": 0.2216,
+      "step": 11215
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.527860329334127e-05,
+      "loss": 0.2162,
+      "step": 11216
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.526248314917826e-05,
+      "loss": 0.2131,
+      "step": 11217
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.524636590042786e-05,
+      "loss": 0.2247,
+      "step": 11218
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.523025154781101e-05,
+      "loss": 0.2079,
+      "step": 11219
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.521414009204835e-05,
+      "loss": 0.2142,
+      "step": 11220
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.519803153386054e-05,
+      "loss": 0.213,
+      "step": 11221
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.518192587396799e-05,
+      "loss": 0.2229,
+      "step": 11222
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5165823113091086e-05,
+      "loss": 0.2061,
+      "step": 11223
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.514972325194996e-05,
+      "loss": 0.2171,
+      "step": 11224
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5133626291264764e-05,
+      "loss": 0.2237,
+      "step": 11225
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.511753223175537e-05,
+      "loss": 0.2237,
+      "step": 11226
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5101441074141696e-05,
+      "loss": 0.2033,
+      "step": 11227
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5085352819143266e-05,
+      "loss": 0.2225,
+      "step": 11228
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.506926746747977e-05,
+      "loss": 0.2253,
+      "step": 11229
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5053185019870564e-05,
+      "loss": 0.2178,
+      "step": 11230
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5037105477034974e-05,
+      "loss": 0.2054,
+      "step": 11231
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5021028839692125e-05,
+      "loss": 0.211,
+      "step": 11232
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.500495510856111e-05,
+      "loss": 0.2222,
+      "step": 11233
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.498888428436077e-05,
+      "loss": 0.2171,
+      "step": 11234
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4972816367809944e-05,
+      "loss": 0.2271,
+      "step": 11235
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4956751359627214e-05,
+      "loss": 0.2147,
+      "step": 11236
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.494068926053116e-05,
+      "loss": 0.2196,
+      "step": 11237
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.492463007124008e-05,
+      "loss": 0.2095,
+      "step": 11238
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.490857379247233e-05,
+      "loss": 0.2282,
+      "step": 11239
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.489252042494594e-05,
+      "loss": 0.2192,
+      "step": 11240
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4876469969378975e-05,
+      "loss": 0.2113,
+      "step": 11241
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.486042242648925e-05,
+      "loss": 0.2213,
+      "step": 11242
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.484437779699453e-05,
+      "loss": 0.2156,
+      "step": 11243
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.482833608161239e-05,
+      "loss": 0.2132,
+      "step": 11244
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.481229728106035e-05,
+      "loss": 0.2142,
+      "step": 11245
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4796261396055677e-05,
+      "loss": 0.2081,
+      "step": 11246
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.478022842731569e-05,
+      "loss": 0.2202,
+      "step": 11247
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.476419837555734e-05,
+      "loss": 0.1975,
+      "step": 11248
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.474817124149769e-05,
+      "loss": 0.2266,
+      "step": 11249
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.473214702585346e-05,
+      "loss": 0.2329,
+      "step": 11250
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.471612572934143e-05,
+      "loss": 0.2144,
+      "step": 11251
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.47001073526781e-05,
+      "loss": 0.2163,
+      "step": 11252
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4684091896579954e-05,
+      "loss": 0.2206,
+      "step": 11253
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4668079361763215e-05,
+      "loss": 0.2074,
+      "step": 11254
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.465206974894413e-05,
+      "loss": 0.2236,
+      "step": 11255
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.463606305883867e-05,
+      "loss": 0.2052,
+      "step": 11256
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.462005929216279e-05,
+      "loss": 0.225,
+      "step": 11257
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4604058449632215e-05,
+      "loss": 0.2127,
+      "step": 11258
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.458806053196265e-05,
+      "loss": 0.2218,
+      "step": 11259
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.457206553986956e-05,
+      "loss": 0.22,
+      "step": 11260
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.455607347406836e-05,
+      "loss": 0.2218,
+      "step": 11261
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4540084335274256e-05,
+      "loss": 0.2276,
+      "step": 11262
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4524098124202445e-05,
+      "loss": 0.2208,
+      "step": 11263
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.450811484156783e-05,
+      "loss": 0.2107,
+      "step": 11264
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.449213448808535e-05,
+      "loss": 0.2342,
+      "step": 11265
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.447615706446965e-05,
+      "loss": 0.2458,
+      "step": 11266
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.446018257143545e-05,
+      "loss": 0.2331,
+      "step": 11267
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.444421100969706e-05,
+      "loss": 0.2439,
+      "step": 11268
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4428242379968936e-05,
+      "loss": 0.2312,
+      "step": 11269
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.44122766829652e-05,
+      "loss": 0.2156,
+      "step": 11270
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4396313919400004e-05,
+      "loss": 0.2339,
+      "step": 11271
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4380354089987196e-05,
+      "loss": 0.2166,
+      "step": 11272
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.436439719544069e-05,
+      "loss": 0.2469,
+      "step": 11273
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.434844323647407e-05,
+      "loss": 0.2241,
+      "step": 11274
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.433249221380097e-05,
+      "loss": 0.2478,
+      "step": 11275
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.431654412813472e-05,
+      "loss": 0.2314,
+      "step": 11276
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.430059898018868e-05,
+      "loss": 0.2352,
+      "step": 11277
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.428465677067595e-05,
+      "loss": 0.2377,
+      "step": 11278
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.42687175003096e-05,
+      "loss": 0.2239,
+      "step": 11279
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4252781169802486e-05,
+      "loss": 0.2303,
+      "step": 11280
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.423684777986734e-05,
+      "loss": 0.2396,
+      "step": 11281
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4220917331216875e-05,
+      "loss": 0.22,
+      "step": 11282
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.420498982456348e-05,
+      "loss": 0.2259,
+      "step": 11283
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.418906526061962e-05,
+      "loss": 0.2206,
+      "step": 11284
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.417314364009746e-05,
+      "loss": 0.2247,
+      "step": 11285
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4157224963709145e-05,
+      "loss": 0.2129,
+      "step": 11286
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.414130923216663e-05,
+      "loss": 0.2087,
+      "step": 11287
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.412539644618171e-05,
+      "loss": 0.2258,
+      "step": 11288
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.410948660646617e-05,
+      "loss": 0.2324,
+      "step": 11289
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4093579713731494e-05,
+      "loss": 0.2235,
+      "step": 11290
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.407767576868922e-05,
+      "loss": 0.2257,
+      "step": 11291
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4061774772050604e-05,
+      "loss": 0.2137,
+      "step": 11292
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4045876724526795e-05,
+      "loss": 0.2245,
+      "step": 11293
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4029981626828924e-05,
+      "loss": 0.22,
+      "step": 11294
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.401408947966781e-05,
+      "loss": 0.2061,
+      "step": 11295
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3998200283754325e-05,
+      "loss": 0.2065,
+      "step": 11296
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.398231403979905e-05,
+      "loss": 0.223,
+      "step": 11297
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3966430748512556e-05,
+      "loss": 0.2246,
+      "step": 11298
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.395055041060519e-05,
+      "loss": 0.21,
+      "step": 11299
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3934673026787246e-05,
+      "loss": 0.2038,
+      "step": 11300
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3918798597768795e-05,
+      "loss": 0.2171,
+      "step": 11301
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.390292712425989e-05,
+      "loss": 0.2159,
+      "step": 11302
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3887058606970327e-05,
+      "loss": 0.2191,
+      "step": 11303
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.387119304660988e-05,
+      "loss": 0.2156,
+      "step": 11304
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3855330443888114e-05,
+      "loss": 0.2228,
+      "step": 11305
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3839470799514526e-05,
+      "loss": 0.2252,
+      "step": 11306
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.382361411419842e-05,
+      "loss": 0.2205,
+      "step": 11307
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.380776038864899e-05,
+      "loss": 0.2264,
+      "step": 11308
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.379190962357528e-05,
+      "loss": 0.2291,
+      "step": 11309
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.377606181968627e-05,
+      "loss": 0.2116,
+      "step": 11310
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3760216977690706e-05,
+      "loss": 0.2222,
+      "step": 11311
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.374437509829732e-05,
+      "loss": 0.2054,
+      "step": 11312
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3728536182214574e-05,
+      "loss": 0.2135,
+      "step": 11313
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3712700230150954e-05,
+      "loss": 0.2302,
+      "step": 11314
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3696867242814634e-05,
+      "loss": 0.2069,
+      "step": 11315
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.368103722091385e-05,
+      "loss": 0.2191,
+      "step": 11316
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3665210165156515e-05,
+      "loss": 0.2195,
+      "step": 11317
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.364938607625057e-05,
+      "loss": 0.2094,
+      "step": 11318
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3633564954903695e-05,
+      "loss": 0.2248,
+      "step": 11319
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.361774680182357e-05,
+      "loss": 0.2301,
+      "step": 11320
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3601931617717594e-05,
+      "loss": 0.2139,
+      "step": 11321
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.358611940329317e-05,
+      "loss": 0.2289,
+      "step": 11322
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.357031015925745e-05,
+      "loss": 0.2125,
+      "step": 11323
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3554503886317556e-05,
+      "loss": 0.2254,
+      "step": 11324
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.353870058518038e-05,
+      "loss": 0.2161,
+      "step": 11325
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.352290025655279e-05,
+      "loss": 0.218,
+      "step": 11326
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.350710290114144e-05,
+      "loss": 0.2113,
+      "step": 11327
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3491308519652874e-05,
+      "loss": 0.2142,
+      "step": 11328
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.347551711279345e-05,
+      "loss": 0.216,
+      "step": 11329
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.345972868126953e-05,
+      "loss": 0.2206,
+      "step": 11330
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.344394322578718e-05,
+      "loss": 0.2189,
+      "step": 11331
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.342816074705248e-05,
+      "loss": 0.2027,
+      "step": 11332
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3412381245771265e-05,
+      "loss": 0.2167,
+      "step": 11333
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3396604722649306e-05,
+      "loss": 0.2184,
+      "step": 11334
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.338083117839217e-05,
+      "loss": 0.2192,
+      "step": 11335
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.33650606137054e-05,
+      "loss": 0.2137,
+      "step": 11336
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.334929302929428e-05,
+      "loss": 0.219,
+      "step": 11337
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.333352842586408e-05,
+      "loss": 0.217,
+      "step": 11338
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.331776680411982e-05,
+      "loss": 0.218,
+      "step": 11339
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.33020081647665e-05,
+      "loss": 0.2222,
+      "step": 11340
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.328625250850887e-05,
+      "loss": 0.2172,
+      "step": 11341
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3270499836051683e-05,
+      "loss": 0.2061,
+      "step": 11342
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.325475014809941e-05,
+      "loss": 0.2264,
+      "step": 11343
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.323900344535652e-05,
+      "loss": 0.2145,
+      "step": 11344
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.322325972852728e-05,
+      "loss": 0.2111,
+      "step": 11345
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3207518998315775e-05,
+      "loss": 0.2118,
+      "step": 11346
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.319178125542611e-05,
+      "loss": 0.2035,
+      "step": 11347
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.317604650056211e-05,
+      "loss": 0.2157,
+      "step": 11348
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3160314734427486e-05,
+      "loss": 0.2174,
+      "step": 11349
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.314458595772592e-05,
+      "loss": 0.2171,
+      "step": 11350
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.312886017116083e-05,
+      "loss": 0.2075,
+      "step": 11351
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.311313737543561e-05,
+      "loss": 0.2179,
+      "step": 11352
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.309741757125341e-05,
+      "loss": 0.2112,
+      "step": 11353
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.308170075931736e-05,
+      "loss": 0.217,
+      "step": 11354
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.306598694033034e-05,
+      "loss": 0.2194,
+      "step": 11355
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.305027611499524e-05,
+      "loss": 0.214,
+      "step": 11356
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.303456828401464e-05,
+      "loss": 0.2167,
+      "step": 11357
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3018863448091165e-05,
+      "loss": 0.2104,
+      "step": 11358
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.300316160792719e-05,
+      "loss": 0.2172,
+      "step": 11359
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.298746276422493e-05,
+      "loss": 0.2058,
+      "step": 11360
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.297176691768662e-05,
+      "loss": 0.2198,
+      "step": 11361
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.295607406901417e-05,
+      "loss": 0.2217,
+      "step": 11362
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.294038421890952e-05,
+      "loss": 0.2125,
+      "step": 11363
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.292469736807435e-05,
+      "loss": 0.2144,
+      "step": 11364
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.290901351721033e-05,
+      "loss": 0.2167,
+      "step": 11365
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.289333266701885e-05,
+      "loss": 0.2272,
+      "step": 11366
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.28776548182013e-05,
+      "loss": 0.2189,
+      "step": 11367
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.286197997145887e-05,
+      "loss": 0.2026,
+      "step": 11368
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.284630812749257e-05,
+      "loss": 0.2232,
+      "step": 11369
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.283063928700341e-05,
+      "loss": 0.2169,
+      "step": 11370
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2814973450692146e-05,
+      "loss": 0.2054,
+      "step": 11371
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2799310619259395e-05,
+      "loss": 0.2208,
+      "step": 11372
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.278365079340577e-05,
+      "loss": 0.2155,
+      "step": 11373
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.276799397383159e-05,
+      "loss": 0.212,
+      "step": 11374
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.275234016123717e-05,
+      "loss": 0.2057,
+      "step": 11375
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.273668935632258e-05,
+      "loss": 0.2233,
+      "step": 11376
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2721041559787865e-05,
+      "loss": 0.2131,
+      "step": 11377
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.270539677233283e-05,
+      "loss": 0.2102,
+      "step": 11378
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.268975499465724e-05,
+      "loss": 0.2093,
+      "step": 11379
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.267411622746062e-05,
+      "loss": 0.2285,
+      "step": 11380
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.265848047144249e-05,
+      "loss": 0.2243,
+      "step": 11381
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.26428477273021e-05,
+      "loss": 0.2154,
+      "step": 11382
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.26272179957387e-05,
+      "loss": 0.2176,
+      "step": 11383
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.261159127745126e-05,
+      "loss": 0.229,
+      "step": 11384
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2595967573138766e-05,
+      "loss": 0.2096,
+      "step": 11385
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.258034688349992e-05,
+      "loss": 0.2288,
+      "step": 11386
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.256472920923348e-05,
+      "loss": 0.2296,
+      "step": 11387
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2549114551037793e-05,
+      "loss": 0.2054,
+      "step": 11388
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.253350290961137e-05,
+      "loss": 0.2055,
+      "step": 11389
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.251789428565234e-05,
+      "loss": 0.2144,
+      "step": 11390
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.25022886798589e-05,
+      "loss": 0.2021,
+      "step": 11391
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2486686092928944e-05,
+      "loss": 0.2225,
+      "step": 11392
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2471086525560366e-05,
+      "loss": 0.2065,
+      "step": 11393
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2455489978450805e-05,
+      "loss": 0.2052,
+      "step": 11394
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.243989645229787e-05,
+      "loss": 0.2222,
+      "step": 11395
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.242430594779895e-05,
+      "loss": 0.216,
+      "step": 11396
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.240871846565138e-05,
+      "loss": 0.2173,
+      "step": 11397
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.239313400655226e-05,
+      "loss": 0.2147,
+      "step": 11398
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2377552571198676e-05,
+      "loss": 0.216,
+      "step": 11399
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2361974160287455e-05,
+      "loss": 0.2138,
+      "step": 11400
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.234639877451541e-05,
+      "loss": 0.1985,
+      "step": 11401
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.233082641457909e-05,
+      "loss": 0.2252,
+      "step": 11402
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2315257081175024e-05,
+      "loss": 0.2164,
+      "step": 11403
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2299690774999515e-05,
+      "loss": 0.2063,
+      "step": 11404
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.228412749674884e-05,
+      "loss": 0.2112,
+      "step": 11405
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2268567247118984e-05,
+      "loss": 0.221,
+      "step": 11406
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.225301002680601e-05,
+      "loss": 0.2267,
+      "step": 11407
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.223745583650557e-05,
+      "loss": 0.2236,
+      "step": 11408
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.222190467691344e-05,
+      "loss": 0.2218,
+      "step": 11409
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2206356548725084e-05,
+      "loss": 0.2083,
+      "step": 11410
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.219081145263597e-05,
+      "loss": 0.223,
+      "step": 11411
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.217526938934128e-05,
+      "loss": 0.2153,
+      "step": 11412
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2159730359536225e-05,
+      "loss": 0.2129,
+      "step": 11413
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2144194363915705e-05,
+      "loss": 0.214,
+      "step": 11414
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2128661403174655e-05,
+      "loss": 0.2158,
+      "step": 11415
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.211313147800773e-05,
+      "loss": 0.2105,
+      "step": 11416
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2097604589109565e-05,
+      "loss": 0.2091,
+      "step": 11417
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.208208073717456e-05,
+      "loss": 0.2152,
+      "step": 11418
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2066559922897064e-05,
+      "loss": 0.2116,
+      "step": 11419
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.205104214697121e-05,
+      "loss": 0.217,
+      "step": 11420
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.203552741009109e-05,
+      "loss": 0.2168,
+      "step": 11421
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2020015712950555e-05,
+      "loss": 0.2078,
+      "step": 11422
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.200450705624343e-05,
+      "loss": 0.2107,
+      "step": 11423
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.198900144066331e-05,
+      "loss": 0.219,
+      "step": 11424
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.197349886690366e-05,
+      "loss": 0.2167,
+      "step": 11425
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.19579993356579e-05,
+      "loss": 0.2147,
+      "step": 11426
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.194250284761923e-05,
+      "loss": 0.2109,
+      "step": 11427
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.192700940348069e-05,
+      "loss": 0.2209,
+      "step": 11428
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.191151900393532e-05,
+      "loss": 0.2296,
+      "step": 11429
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.189603164967584e-05,
+      "loss": 0.2199,
+      "step": 11430
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.188054734139502e-05,
+      "loss": 0.2178,
+      "step": 11431
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.186506607978532e-05,
+      "loss": 0.2141,
+      "step": 11432
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.184958786553922e-05,
+      "loss": 0.2251,
+      "step": 11433
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.183411269934892e-05,
+      "loss": 0.2144,
+      "step": 11434
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.181864058190661e-05,
+      "loss": 0.2279,
+      "step": 11435
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1803171513904276e-05,
+      "loss": 0.2186,
+      "step": 11436
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.178770549603372e-05,
+      "loss": 0.2116,
+      "step": 11437
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.177224252898676e-05,
+      "loss": 0.2025,
+      "step": 11438
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1756782613454896e-05,
+      "loss": 0.198,
+      "step": 11439
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.174132575012965e-05,
+      "loss": 0.2277,
+      "step": 11440
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1725871939702276e-05,
+      "loss": 0.2256,
+      "step": 11441
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1710421182864016e-05,
+      "loss": 0.2038,
+      "step": 11442
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1694973480305845e-05,
+      "loss": 0.2197,
+      "step": 11443
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.167952883271874e-05,
+      "loss": 0.2217,
+      "step": 11444
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1664087240793386e-05,
+      "loss": 0.2041,
+      "step": 11445
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1648648705220494e-05,
+      "loss": 0.2085,
+      "step": 11446
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.163321322669053e-05,
+      "loss": 0.2306,
+      "step": 11447
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.161778080589384e-05,
+      "loss": 0.2156,
+      "step": 11448
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.160235144352063e-05,
+      "loss": 0.2188,
+      "step": 11449
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.158692514026103e-05,
+      "loss": 0.2286,
+      "step": 11450
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1571501896804925e-05,
+      "loss": 0.2139,
+      "step": 11451
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.155608171384221e-05,
+      "loss": 0.2239,
+      "step": 11452
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1540664592062473e-05,
+      "loss": 0.2162,
+      "step": 11453
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1525250532155325e-05,
+      "loss": 0.2105,
+      "step": 11454
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.150983953481011e-05,
+      "loss": 0.2011,
+      "step": 11455
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1494431600716145e-05,
+      "loss": 0.2113,
+      "step": 11456
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.14790267305625e-05,
+      "loss": 0.2202,
+      "step": 11457
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.146362492503822e-05,
+      "loss": 0.2081,
+      "step": 11458
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.144822618483209e-05,
+      "loss": 0.2107,
+      "step": 11459
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1432830510632894e-05,
+      "loss": 0.2075,
+      "step": 11460
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.141743790312916e-05,
+      "loss": 0.2192,
+      "step": 11461
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.140204836300936e-05,
+      "loss": 0.1997,
+      "step": 11462
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1386661890961765e-05,
+      "loss": 0.2073,
+      "step": 11463
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.137127848767459e-05,
+      "loss": 0.2054,
+      "step": 11464
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135589815383582e-05,
+      "loss": 0.207,
+      "step": 11465
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134052089013338e-05,
+      "loss": 0.2043,
+      "step": 11466
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132514669725501e-05,
+      "loss": 0.2149,
+      "step": 11467
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.130977557588833e-05,
+      "loss": 0.2105,
+      "step": 11468
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.129440752672078e-05,
+      "loss": 0.2144,
+      "step": 11469
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.127904255043977e-05,
+      "loss": 0.2233,
+      "step": 11470
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126368064773244e-05,
+      "loss": 0.212,
+      "step": 11471
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.124832181928593e-05,
+      "loss": 0.2041,
+      "step": 11472
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1232966065787095e-05,
+      "loss": 0.21,
+      "step": 11473
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.121761338792279e-05,
+      "loss": 0.2149,
+      "step": 11474
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1202263786379605e-05,
+      "loss": 0.2195,
+      "step": 11475
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.118691726184414e-05,
+      "loss": 0.2091,
+      "step": 11476
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1171573815002685e-05,
+      "loss": 0.2187,
+      "step": 11477
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.115623344654157e-05,
+      "loss": 0.2067,
+      "step": 11478
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.114089615714681e-05,
+      "loss": 0.2228,
+      "step": 11479
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1125561947504464e-05,
+      "loss": 0.2176,
+      "step": 11480
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1110230818300276e-05,
+      "loss": 0.2215,
+      "step": 11481
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.109490277022003e-05,
+      "loss": 0.2087,
+      "step": 11482
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.107957780394917e-05,
+      "loss": 0.2035,
+      "step": 11483
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.10642559201732e-05,
+      "loss": 0.2192,
+      "step": 11484
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.104893711957734e-05,
+      "loss": 0.2062,
+      "step": 11485
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.103362140284678e-05,
+      "loss": 0.2196,
+      "step": 11486
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1018308770666506e-05,
+      "loss": 0.2113,
+      "step": 11487
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.100299922372136e-05,
+      "loss": 0.2129,
+      "step": 11488
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.098769276269605e-05,
+      "loss": 0.2197,
+      "step": 11489
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.097238938827523e-05,
+      "loss": 0.2125,
+      "step": 11490
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.095708910114329e-05,
+      "loss": 0.2121,
+      "step": 11491
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0941791901984594e-05,
+      "loss": 0.2062,
+      "step": 11492
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0926497791483254e-05,
+      "loss": 0.2216,
+      "step": 11493
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0911206770323366e-05,
+      "loss": 0.2214,
+      "step": 11494
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.089591883918878e-05,
+      "loss": 0.2226,
+      "step": 11495
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0880633998763295e-05,
+      "loss": 0.232,
+      "step": 11496
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0865352249730486e-05,
+      "loss": 0.2236,
+      "step": 11497
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.08500735927739e-05,
+      "loss": 0.2095,
+      "step": 11498
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0834798028576804e-05,
+      "loss": 0.2172,
+      "step": 11499
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.081952555782247e-05,
+      "loss": 0.213,
+      "step": 11500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0804256181193946e-05,
+      "loss": 0.2057,
+      "step": 11501
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.078898989937413e-05,
+      "loss": 0.2131,
+      "step": 11502
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.077372671304586e-05,
+      "loss": 0.2246,
+      "step": 11503
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.075846662289173e-05,
+      "loss": 0.208,
+      "step": 11504
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.074320962959434e-05,
+      "loss": 0.2079,
+      "step": 11505
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0727955733835964e-05,
+      "loss": 0.219,
+      "step": 11506
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.071270493629893e-05,
+      "loss": 0.2131,
+      "step": 11507
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.069745723766529e-05,
+      "loss": 0.211,
+      "step": 11508
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.068221263861699e-05,
+      "loss": 0.2118,
+      "step": 11509
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.06669711398359e-05,
+      "loss": 0.2282,
+      "step": 11510
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.065173274200364e-05,
+      "loss": 0.2176,
+      "step": 11511
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.063649744580182e-05,
+      "loss": 0.2123,
+      "step": 11512
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.062126525191177e-05,
+      "loss": 0.2225,
+      "step": 11513
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.060603616101484e-05,
+      "loss": 0.2146,
+      "step": 11514
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.059081017379212e-05,
+      "loss": 0.2034,
+      "step": 11515
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0575587290924555e-05,
+      "loss": 0.2103,
+      "step": 11516
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.056036751309307e-05,
+      "loss": 0.2147,
+      "step": 11517
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.054515084097831e-05,
+      "loss": 0.2127,
+      "step": 11518
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.052993727526091e-05,
+      "loss": 0.2122,
+      "step": 11519
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.051472681662124e-05,
+      "loss": 0.2124,
+      "step": 11520
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0499519465739657e-05,
+      "loss": 0.2189,
+      "step": 11521
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0484315223296245e-05,
+      "loss": 0.212,
+      "step": 11522
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0469114089971095e-05,
+      "loss": 0.2251,
+      "step": 11523
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0453916066444023e-05,
+      "loss": 0.225,
+      "step": 11524
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0438721153394822e-05,
+      "loss": 0.2152,
+      "step": 11525
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.042352935150302e-05,
+      "loss": 0.2145,
+      "step": 11526
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.04083406614482e-05,
+      "loss": 0.2162,
+      "step": 11527
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.039315508390952e-05,
+      "loss": 0.2144,
+      "step": 11528
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0377972619566274e-05,
+      "loss": 0.2054,
+      "step": 11529
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0362793269097445e-05,
+      "loss": 0.2116,
+      "step": 11530
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0347617033181998e-05,
+      "loss": 0.212,
+      "step": 11531
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0332443912498632e-05,
+      "loss": 0.2076,
+      "step": 11532
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.031727390772603e-05,
+      "loss": 0.2132,
+      "step": 11533
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0302107019542603e-05,
+      "loss": 0.2157,
+      "step": 11534
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0286943248626786e-05,
+      "loss": 0.2122,
+      "step": 11535
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0271782595656707e-05,
+      "loss": 0.2035,
+      "step": 11536
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0256625061310494e-05,
+      "loss": 0.2138,
+      "step": 11537
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0241470646266013e-05,
+      "loss": 0.2166,
+      "step": 11538
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0226319351201115e-05,
+      "loss": 0.211,
+      "step": 11539
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0211171176793395e-05,
+      "loss": 0.2103,
+      "step": 11540
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0196026123720412e-05,
+      "loss": 0.2062,
+      "step": 11541
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0180884192659464e-05,
+      "loss": 0.2209,
+      "step": 11542
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.016574538428787e-05,
+      "loss": 0.2117,
+      "step": 11543
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.015060969928263e-05,
+      "loss": 0.2128,
+      "step": 11544
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.013547713832077e-05,
+      "loss": 0.2206,
+      "step": 11545
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0120347702079032e-05,
+      "loss": 0.2178,
+      "step": 11546
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0105221391234183e-05,
+      "loss": 0.2189,
+      "step": 11547
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0090098206462637e-05,
+      "loss": 0.2152,
+      "step": 11548
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.007497814844087e-05,
+      "loss": 0.2082,
+      "step": 11549
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0059861217845063e-05,
+      "loss": 0.2109,
+      "step": 11550
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0044747415351404e-05,
+      "loss": 0.2112,
+      "step": 11551
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.002963674163579e-05,
+      "loss": 0.1975,
+      "step": 11552
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0014529197374118e-05,
+      "loss": 0.2187,
+      "step": 11553
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9999424783242024e-05,
+      "loss": 0.2206,
+      "step": 11554
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9984323499915125e-05,
+      "loss": 0.2238,
+      "step": 11555
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.996922534806874e-05,
+      "loss": 0.1962,
+      "step": 11556
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9954130328378238e-05,
+      "loss": 0.2239,
+      "step": 11557
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9939038441518675e-05,
+      "loss": 0.207,
+      "step": 11558
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.99239496881651e-05,
+      "loss": 0.202,
+      "step": 11559
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9908864068992303e-05,
+      "loss": 0.2187,
+      "step": 11560
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9893781584675052e-05,
+      "loss": 0.2228,
+      "step": 11561
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.987870223588787e-05,
+      "loss": 0.2117,
+      "step": 11562
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9863626023305235e-05,
+      "loss": 0.2201,
+      "step": 11563
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9848552947601392e-05,
+      "loss": 0.2301,
+      "step": 11564
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9833483009450535e-05,
+      "loss": 0.2114,
+      "step": 11565
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9818416209526613e-05,
+      "loss": 0.2121,
+      "step": 11566
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9803352548503593e-05,
+      "loss": 0.2099,
+      "step": 11567
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.978829202705509e-05,
+      "loss": 0.2069,
+      "step": 11568
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9773234645854763e-05,
+      "loss": 0.2174,
+      "step": 11569
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.975818040557602e-05,
+      "loss": 0.2068,
+      "step": 11570
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9743129306892203e-05,
+      "loss": 0.2081,
+      "step": 11571
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9728081350476445e-05,
+      "loss": 0.2165,
+      "step": 11572
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9713036537001826e-05,
+      "loss": 0.2315,
+      "step": 11573
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9697994867141153e-05,
+      "loss": 0.2242,
+      "step": 11574
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.968295634156726e-05,
+      "loss": 0.2135,
+      "step": 11575
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9667920960952667e-05,
+      "loss": 0.2115,
+      "step": 11576
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.965288872596991e-05,
+      "loss": 0.2237,
+      "step": 11577
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9637859637291243e-05,
+      "loss": 0.2211,
+      "step": 11578
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9622833695588915e-05,
+      "loss": 0.2155,
+      "step": 11579
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.960781090153495e-05,
+      "loss": 0.2188,
+      "step": 11580
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.959279125580119e-05,
+      "loss": 0.2221,
+      "step": 11581
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9577774759059484e-05,
+      "loss": 0.2212,
+      "step": 11582
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9562761411981366e-05,
+      "loss": 0.2231,
+      "step": 11583
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9547751215238395e-05,
+      "loss": 0.2032,
+      "step": 11584
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9532744169501826e-05,
+      "loss": 0.2189,
+      "step": 11585
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.951774027544294e-05,
+      "loss": 0.2211,
+      "step": 11586
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9502739533732748e-05,
+      "loss": 0.2074,
+      "step": 11587
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9487741945042148e-05,
+      "loss": 0.2196,
+      "step": 11588
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9472747510041955e-05,
+      "loss": 0.2106,
+      "step": 11589
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9457756229402744e-05,
+      "loss": 0.2124,
+      "step": 11590
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.944276810379507e-05,
+      "loss": 0.2122,
+      "step": 11591
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9427783133889265e-05,
+      "loss": 0.2133,
+      "step": 11592
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9412801320355498e-05,
+      "loss": 0.206,
+      "step": 11593
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9397822663863894e-05,
+      "loss": 0.2159,
+      "step": 11594
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9382847165084327e-05,
+      "loss": 0.2026,
+      "step": 11595
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9367874824686635e-05,
+      "loss": 0.2171,
+      "step": 11596
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.93529056433404e-05,
+      "loss": 0.2058,
+      "step": 11597
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9337939621715193e-05,
+      "loss": 0.2145,
+      "step": 11598
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9322976760480313e-05,
+      "loss": 0.2097,
+      "step": 11599
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9308017060305047e-05,
+      "loss": 0.2148,
+      "step": 11600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9293060521858408e-05,
+      "loss": 0.2127,
+      "step": 11601
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9278107145809386e-05,
+      "loss": 0.2196,
+      "step": 11602
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.926315693282672e-05,
+      "loss": 0.2132,
+      "step": 11603
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9248209883579136e-05,
+      "loss": 0.2078,
+      "step": 11604
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9233265998735082e-05,
+      "loss": 0.2265,
+      "step": 11605
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9218325278962976e-05,
+      "loss": 0.202,
+      "step": 11606
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.920338772493104e-05,
+      "loss": 0.1984,
+      "step": 11607
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9188453337307342e-05,
+      "loss": 0.2043,
+      "step": 11608
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.917352211675981e-05,
+      "loss": 0.2006,
+      "step": 11609
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9158594063956313e-05,
+      "loss": 0.2158,
+      "step": 11610
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9143669179564448e-05,
+      "loss": 0.2257,
+      "step": 11611
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9128747464251792e-05,
+      "loss": 0.2089,
+      "step": 11612
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.911382891868567e-05,
+      "loss": 0.208,
+      "step": 11613
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.909891354353339e-05,
+      "loss": 0.2176,
+      "step": 11614
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9084001339461963e-05,
+      "loss": 0.2006,
+      "step": 11615
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9069092307138433e-05,
+      "loss": 0.2106,
+      "step": 11616
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9054186447229536e-05,
+      "loss": 0.2189,
+      "step": 11617
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9039283760402004e-05,
+      "loss": 0.2038,
+      "step": 11618
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.902438424732231e-05,
+      "loss": 0.2082,
+      "step": 11619
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9009487908656897e-05,
+      "loss": 0.2155,
+      "step": 11620
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8994594745071958e-05,
+      "loss": 0.2125,
+      "step": 11621
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.897970475723364e-05,
+      "loss": 0.2037,
+      "step": 11622
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8964817945807855e-05,
+      "loss": 0.2138,
+      "step": 11623
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.894993431146049e-05,
+      "loss": 0.2125,
+      "step": 11624
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.893505385485714e-05,
+      "loss": 0.2058,
+      "step": 11625
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8920176576663428e-05,
+      "loss": 0.2097,
+      "step": 11626
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8905302477544692e-05,
+      "loss": 0.2043,
+      "step": 11627
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8890431558166196e-05,
+      "loss": 0.2092,
+      "step": 11628
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8875563819193008e-05,
+      "loss": 0.2136,
+      "step": 11629
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.886069926129016e-05,
+      "loss": 0.2132,
+      "step": 11630
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.884583788512242e-05,
+      "loss": 0.2182,
+      "step": 11631
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.883097969135452e-05,
+      "loss": 0.2059,
+      "step": 11632
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8816124680650935e-05,
+      "loss": 0.2159,
+      "step": 11633
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8801272853676144e-05,
+      "loss": 0.2108,
+      "step": 11634
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.878642421109432e-05,
+      "loss": 0.2207,
+      "step": 11635
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.877157875356964e-05,
+      "loss": 0.1991,
+      "step": 11636
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8756736481766022e-05,
+      "loss": 0.2172,
+      "step": 11637
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8741897396347338e-05,
+      "loss": 0.2087,
+      "step": 11638
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8727061497977227e-05,
+      "loss": 0.2002,
+      "step": 11639
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8712228787319272e-05,
+      "loss": 0.2235,
+      "step": 11640
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.869739926503684e-05,
+      "loss": 0.2156,
+      "step": 11641
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8682572931793216e-05,
+      "loss": 0.2191,
+      "step": 11642
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8667749788251487e-05,
+      "loss": 0.202,
+      "step": 11643
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.865292983507465e-05,
+      "loss": 0.2093,
+      "step": 11644
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8638113072925534e-05,
+      "loss": 0.2218,
+      "step": 11645
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8623299502466773e-05,
+      "loss": 0.2141,
+      "step": 11646
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8608489124360983e-05,
+      "loss": 0.2091,
+      "step": 11647
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8593681939270534e-05,
+      "loss": 0.2079,
+      "step": 11648
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.857887794785764e-05,
+      "loss": 0.2231,
+      "step": 11649
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8564077150784496e-05,
+      "loss": 0.2152,
+      "step": 11650
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8549279548713002e-05,
+      "loss": 0.2077,
+      "step": 11651
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8534485142305047e-05,
+      "loss": 0.2239,
+      "step": 11652
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8519693932222247e-05,
+      "loss": 0.2074,
+      "step": 11653
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8504905919126235e-05,
+      "loss": 0.2072,
+      "step": 11654
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8490121103678325e-05,
+      "loss": 0.2144,
+      "step": 11655
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8475339486539833e-05,
+      "loss": 0.2098,
+      "step": 11656
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.846056106837186e-05,
+      "loss": 0.2127,
+      "step": 11657
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.844578584983534e-05,
+      "loss": 0.2107,
+      "step": 11658
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.843101383159116e-05,
+      "loss": 0.2008,
+      "step": 11659
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8416245014299936e-05,
+      "loss": 0.2074,
+      "step": 11660
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.840147939862229e-05,
+      "loss": 0.209,
+      "step": 11661
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8386716985218544e-05,
+      "loss": 0.2143,
+      "step": 11662
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8371957774749013e-05,
+      "loss": 0.2144,
+      "step": 11663
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8357201767873754e-05,
+      "loss": 0.2202,
+      "step": 11664
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.83424489652528e-05,
+      "loss": 0.2062,
+      "step": 11665
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8327699367545913e-05,
+      "loss": 0.2215,
+      "step": 11666
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8312952975412822e-05,
+      "loss": 0.2125,
+      "step": 11667
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.829820978951305e-05,
+      "loss": 0.2078,
+      "step": 11668
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8283469810505993e-05,
+      "loss": 0.2138,
+      "step": 11669
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8268733039050865e-05,
+      "loss": 0.2197,
+      "step": 11670
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8253999475806848e-05,
+      "loss": 0.1986,
+      "step": 11671
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8239269121432833e-05,
+      "loss": 0.217,
+      "step": 11672
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8224541976587693e-05,
+      "loss": 0.2266,
+      "step": 11673
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.820981804193007e-05,
+      "loss": 0.2079,
+      "step": 11674
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8195097318118547e-05,
+      "loss": 0.2189,
+      "step": 11675
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.818037980581144e-05,
+      "loss": 0.2152,
+      "step": 11676
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8165665505667083e-05,
+      "loss": 0.2013,
+      "step": 11677
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.81509544183435e-05,
+      "loss": 0.2123,
+      "step": 11678
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8136246544498724e-05,
+      "loss": 0.1959,
+      "step": 11679
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8121541884790492e-05,
+      "loss": 0.1975,
+      "step": 11680
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.810684043987656e-05,
+      "loss": 0.2041,
+      "step": 11681
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.809214221041436e-05,
+      "loss": 0.2194,
+      "step": 11682
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.807744719706138e-05,
+      "loss": 0.209,
+      "step": 11683
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8062755400474784e-05,
+      "loss": 0.2024,
+      "step": 11684
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8048066821311715e-05,
+      "loss": 0.2251,
+      "step": 11685
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.803338146022908e-05,
+      "loss": 0.2195,
+      "step": 11686
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8018699317883778e-05,
+      "loss": 0.2182,
+      "step": 11687
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.800402039493234e-05,
+      "loss": 0.2115,
+      "step": 11688
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7989344692031393e-05,
+      "loss": 0.2023,
+      "step": 11689
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.797467220983725e-05,
+      "loss": 0.2119,
+      "step": 11690
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7960002949006202e-05,
+      "loss": 0.2093,
+      "step": 11691
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7945336910194277e-05,
+      "loss": 0.2044,
+      "step": 11692
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7930674094057475e-05,
+      "loss": 0.2057,
+      "step": 11693
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7916014501251554e-05,
+      "loss": 0.1949,
+      "step": 11694
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.790135813243221e-05,
+      "loss": 0.2166,
+      "step": 11695
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7886704988254908e-05,
+      "loss": 0.2107,
+      "step": 11696
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.787205506937507e-05,
+      "loss": 0.2109,
+      "step": 11697
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.785740837644787e-05,
+      "loss": 0.2108,
+      "step": 11698
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7842764910128427e-05,
+      "loss": 0.2096,
+      "step": 11699
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7828124671071632e-05,
+      "loss": 0.2055,
+      "step": 11700
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7813487659932336e-05,
+      "loss": 0.2073,
+      "step": 11701
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7798853877365116e-05,
+      "loss": 0.2055,
+      "step": 11702
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.778422332402455e-05,
+      "loss": 0.2168,
+      "step": 11703
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7769596000564925e-05,
+      "loss": 0.2058,
+      "step": 11704
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7754971907640504e-05,
+      "loss": 0.2147,
+      "step": 11705
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7740351045905312e-05,
+      "loss": 0.2166,
+      "step": 11706
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7725733416013366e-05,
+      "loss": 0.1987,
+      "step": 11707
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7711119018618304e-05,
+      "loss": 0.2213,
+      "step": 11708
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7696507854373878e-05,
+      "loss": 0.2071,
+      "step": 11709
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7681899923933497e-05,
+      "loss": 0.2218,
+      "step": 11710
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7667295227950574e-05,
+      "loss": 0.2173,
+      "step": 11711
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7652693767078253e-05,
+      "loss": 0.2181,
+      "step": 11712
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7638095541969646e-05,
+      "loss": 0.2161,
+      "step": 11713
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7623500553277602e-05,
+      "loss": 0.2085,
+      "step": 11714
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7608908801654964e-05,
+      "loss": 0.2069,
+      "step": 11715
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7594320287754283e-05,
+      "loss": 0.2063,
+      "step": 11716
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.757973501222809e-05,
+      "loss": 0.2066,
+      "step": 11717
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.756515297572867e-05,
+      "loss": 0.2238,
+      "step": 11718
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7550574178908262e-05,
+      "loss": 0.2169,
+      "step": 11719
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7535998622418858e-05,
+      "loss": 0.2195,
+      "step": 11720
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7521426306912402e-05,
+      "loss": 0.2062,
+      "step": 11721
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7506857233040595e-05,
+      "loss": 0.2149,
+      "step": 11722
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7492291401455107e-05,
+      "loss": 0.216,
+      "step": 11723
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7477728812807368e-05,
+      "loss": 0.2096,
+      "step": 11724
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7463169467748673e-05,
+      "loss": 0.2128,
+      "step": 11725
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7448613366930243e-05,
+      "loss": 0.2197,
+      "step": 11726
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.743406051100308e-05,
+      "loss": 0.2049,
+      "step": 11727
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.741951090061804e-05,
+      "loss": 0.2011,
+      "step": 11728
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7404964536425916e-05,
+      "loss": 0.2333,
+      "step": 11729
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.739042141907724e-05,
+      "loss": 0.218,
+      "step": 11730
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7375881549222527e-05,
+      "loss": 0.2075,
+      "step": 11731
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7361344927512013e-05,
+      "loss": 0.2073,
+      "step": 11732
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7346811554595907e-05,
+      "loss": 0.2139,
+      "step": 11733
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7332281431124175e-05,
+      "loss": 0.2323,
+      "step": 11734
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7317754557746732e-05,
+      "loss": 0.2082,
+      "step": 11735
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7303230935113276e-05,
+      "loss": 0.2063,
+      "step": 11736
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7288710563873356e-05,
+      "loss": 0.2087,
+      "step": 11737
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7274193444676442e-05,
+      "loss": 0.2145,
+      "step": 11738
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.725967957817177e-05,
+      "loss": 0.2147,
+      "step": 11739
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.724516896500855e-05,
+      "loss": 0.2139,
+      "step": 11740
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7230661605835694e-05,
+      "loss": 0.1964,
+      "step": 11741
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.721615750130213e-05,
+      "loss": 0.2115,
+      "step": 11742
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7201656652056483e-05,
+      "loss": 0.2072,
+      "step": 11743
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.718715905874738e-05,
+      "loss": 0.2163,
+      "step": 11744
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7172664722023166e-05,
+      "loss": 0.2129,
+      "step": 11745
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.715817364253217e-05,
+      "loss": 0.2139,
+      "step": 11746
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.714368582092248e-05,
+      "loss": 0.2144,
+      "step": 11747
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7129201257842074e-05,
+      "loss": 0.2039,
+      "step": 11748
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.711471995393875e-05,
+      "loss": 0.1989,
+      "step": 11749
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7100241909860234e-05,
+      "loss": 0.2154,
+      "step": 11750
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7085767126254036e-05,
+      "loss": 0.2081,
+      "step": 11751
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7071295603767567e-05,
+      "loss": 0.2147,
+      "step": 11752
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.705682734304804e-05,
+      "loss": 0.2164,
+      "step": 11753
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7042362344742613e-05,
+      "loss": 0.2078,
+      "step": 11754
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.702790060949816e-05,
+      "loss": 0.2098,
+      "step": 11755
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7013442137961565e-05,
+      "loss": 0.208,
+      "step": 11756
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6998986930779425e-05,
+      "loss": 0.1968,
+      "step": 11757
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.698453498859832e-05,
+      "loss": 0.2164,
+      "step": 11758
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6970086312064545e-05,
+      "loss": 0.2161,
+      "step": 11759
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6955640901824398e-05,
+      "loss": 0.2135,
+      "step": 11760
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.694119875852388e-05,
+      "loss": 0.2104,
+      "step": 11761
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6926759882809006e-05,
+      "loss": 0.2155,
+      "step": 11762
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6912324275325473e-05,
+      "loss": 0.213,
+      "step": 11763
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6897891936719e-05,
+      "loss": 0.2092,
+      "step": 11764
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.688346286763501e-05,
+      "loss": 0.1991,
+      "step": 11765
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6869037068718917e-05,
+      "loss": 0.2015,
+      "step": 11766
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6854614540615875e-05,
+      "loss": 0.2033,
+      "step": 11767
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.684019528397096e-05,
+      "loss": 0.2134,
+      "step": 11768
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6825779299429033e-05,
+      "loss": 0.2107,
+      "step": 11769
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6811366587634924e-05,
+      "loss": 0.2193,
+      "step": 11770
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6796957149233172e-05,
+      "loss": 0.2117,
+      "step": 11771
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6782550984868325e-05,
+      "loss": 0.223,
+      "step": 11772
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.676814809518463e-05,
+      "loss": 0.2158,
+      "step": 11773
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6753748480826336e-05,
+      "loss": 0.1997,
+      "step": 11774
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.67393521424374e-05,
+      "loss": 0.2191,
+      "step": 11775
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6724959080661772e-05,
+      "loss": 0.2083,
+      "step": 11776
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.671056929614313e-05,
+      "loss": 0.2207,
+      "step": 11777
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6696182789525103e-05,
+      "loss": 0.1991,
+      "step": 11778
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6681799561451093e-05,
+      "loss": 0.2066,
+      "step": 11779
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.666741961256446e-05,
+      "loss": 0.2086,
+      "step": 11780
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6653042943508277e-05,
+      "loss": 0.2168,
+      "step": 11781
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.663866955492561e-05,
+      "loss": 0.2192,
+      "step": 11782
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6624299447459255e-05,
+      "loss": 0.2096,
+      "step": 11783
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6609932621751997e-05,
+      "loss": 0.2187,
+      "step": 11784
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6595569078446325e-05,
+      "loss": 0.1994,
+      "step": 11785
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.658120881818471e-05,
+      "loss": 0.2041,
+      "step": 11786
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6566851841609396e-05,
+      "loss": 0.2031,
+      "step": 11787
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6552498149362505e-05,
+      "loss": 0.2157,
+      "step": 11788
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6538147742085983e-05,
+      "loss": 0.2193,
+      "step": 11789
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6523800620421725e-05,
+      "loss": 0.2066,
+      "step": 11790
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6509456785011332e-05,
+      "loss": 0.1995,
+      "step": 11791
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.649511623649642e-05,
+      "loss": 0.2081,
+      "step": 11792
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6480778975518293e-05,
+      "loss": 0.186,
+      "step": 11793
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6466445002718275e-05,
+      "loss": 0.2124,
+      "step": 11794
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6452114318737385e-05,
+      "loss": 0.2178,
+      "step": 11795
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6437786924216634e-05,
+      "loss": 0.2309,
+      "step": 11796
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.642346281979675e-05,
+      "loss": 0.2051,
+      "step": 11797
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.640914200611847e-05,
+      "loss": 0.2048,
+      "step": 11798
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.639482448382221e-05,
+      "loss": 0.22,
+      "step": 11799
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6380510253548396e-05,
+      "loss": 0.1959,
+      "step": 11800
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6366199315937224e-05,
+      "loss": 0.2149,
+      "step": 11801
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6351891671628703e-05,
+      "loss": 0.2077,
+      "step": 11802
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6337587321262813e-05,
+      "loss": 0.2137,
+      "step": 11803
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6323286265479286e-05,
+      "loss": 0.189,
+      "step": 11804
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6308988504917775e-05,
+      "loss": 0.2049,
+      "step": 11805
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6294694040217705e-05,
+      "loss": 0.2151,
+      "step": 11806
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.628040287201845e-05,
+      "loss": 0.22,
+      "step": 11807
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.626611500095918e-05,
+      "loss": 0.2175,
+      "step": 11808
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6251830427678882e-05,
+      "loss": 0.224,
+      "step": 11809
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.62375491528165e-05,
+      "loss": 0.2196,
+      "step": 11810
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6223271177010724e-05,
+      "loss": 0.2153,
+      "step": 11811
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6208996500900172e-05,
+      "loss": 0.2113,
+      "step": 11812
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6194725125123287e-05,
+      "loss": 0.2052,
+      "step": 11813
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6180457050318318e-05,
+      "loss": 0.1976,
+      "step": 11814
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.616619227712346e-05,
+      "loss": 0.2096,
+      "step": 11815
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6151930806176674e-05,
+      "loss": 0.1976,
+      "step": 11816
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.613767263811585e-05,
+      "loss": 0.2065,
+      "step": 11817
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.612341777357865e-05,
+      "loss": 0.2052,
+      "step": 11818
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6109166213202673e-05,
+      "loss": 0.1983,
+      "step": 11819
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6094917957625275e-05,
+      "loss": 0.2293,
+      "step": 11820
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.608067300748377e-05,
+      "loss": 0.2131,
+      "step": 11821
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.60664313634152e-05,
+      "loss": 0.2225,
+      "step": 11822
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6052193026056616e-05,
+      "loss": 0.2075,
+      "step": 11823
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6037957996044738e-05,
+      "loss": 0.2063,
+      "step": 11824
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6023726274016325e-05,
+      "loss": 0.2108,
+      "step": 11825
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6009497860607823e-05,
+      "loss": 0.2097,
+      "step": 11826
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.59952727564557e-05,
+      "loss": 0.2041,
+      "step": 11827
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5981050962196042e-05,
+      "loss": 0.1985,
+      "step": 11828
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5966832478465044e-05,
+      "loss": 0.2075,
+      "step": 11829
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5952617305898542e-05,
+      "loss": 0.2101,
+      "step": 11830
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5938405445132398e-05,
+      "loss": 0.2063,
+      "step": 11831
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5924196896802178e-05,
+      "loss": 0.2057,
+      "step": 11832
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.590999166154342e-05,
+      "loss": 0.1967,
+      "step": 11833
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5895789739991405e-05,
+      "loss": 0.1957,
+      "step": 11834
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.588159113278138e-05,
+      "loss": 0.2071,
+      "step": 11835
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5867395840548316e-05,
+      "loss": 0.216,
+      "step": 11836
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.585320386392717e-05,
+      "loss": 0.1936,
+      "step": 11837
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5839015203552617e-05,
+      "loss": 0.2138,
+      "step": 11838
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.582482986005933e-05,
+      "loss": 0.2146,
+      "step": 11839
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.581064783408167e-05,
+      "loss": 0.214,
+      "step": 11840
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.579646912625402e-05,
+      "loss": 0.2191,
+      "step": 11841
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.578229373721044e-05,
+      "loss": 0.2063,
+      "step": 11842
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.576812166758501e-05,
+      "loss": 0.1999,
+      "step": 11843
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5753952918011503e-05,
+      "loss": 0.2,
+      "step": 11844
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5739787489123714e-05,
+      "loss": 0.2073,
+      "step": 11845
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5725625381555107e-05,
+      "loss": 0.2071,
+      "step": 11846
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5711466595939195e-05,
+      "loss": 0.2105,
+      "step": 11847
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5697311132909107e-05,
+      "loss": 0.2085,
+      "step": 11848
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5683158993098043e-05,
+      "loss": 0.2071,
+      "step": 11849
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5669010177138918e-05,
+      "loss": 0.2086,
+      "step": 11850
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.565486468566458e-05,
+      "loss": 0.2112,
+      "step": 11851
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5640722519307657e-05,
+      "loss": 0.208,
+      "step": 11852
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.56265836787007e-05,
+      "loss": 0.213,
+      "step": 11853
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.561244816447603e-05,
+      "loss": 0.2149,
+      "step": 11854
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.559831597726592e-05,
+      "loss": 0.1968,
+      "step": 11855
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5584187117702363e-05,
+      "loss": 0.2049,
+      "step": 11856
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.557006158641737e-05,
+      "loss": 0.2033,
+      "step": 11857
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5555939384042628e-05,
+      "loss": 0.2054,
+      "step": 11858
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5541820511209813e-05,
+      "loss": 0.2185,
+      "step": 11859
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5527704968550358e-05,
+      "loss": 0.1956,
+      "step": 11860
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5513592756695637e-05,
+      "loss": 0.2095,
+      "step": 11861
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5499483876276765e-05,
+      "loss": 0.2216,
+      "step": 11862
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5485378327924826e-05,
+      "loss": 0.2121,
+      "step": 11863
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5471276112270625e-05,
+      "loss": 0.2049,
+      "step": 11864
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5457177229944974e-05,
+      "loss": 0.2084,
+      "step": 11865
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5443081681578405e-05,
+      "loss": 0.1938,
+      "step": 11866
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.542898946780137e-05,
+      "loss": 0.2006,
+      "step": 11867
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5414900589244083e-05,
+      "loss": 0.1991,
+      "step": 11868
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5400815046536753e-05,
+      "loss": 0.2056,
+      "step": 11869
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.53867328403093e-05,
+      "loss": 0.2152,
+      "step": 11870
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.537265397119163e-05,
+      "loss": 0.2088,
+      "step": 11871
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.535857843981335e-05,
+      "loss": 0.2042,
+      "step": 11872
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5344506246804056e-05,
+      "loss": 0.2042,
+      "step": 11873
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5330437392793083e-05,
+      "loss": 0.2155,
+      "step": 11874
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5316371878409705e-05,
+      "loss": 0.1961,
+      "step": 11875
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5302309704282977e-05,
+      "loss": 0.2067,
+      "step": 11876
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5288250871041876e-05,
+      "loss": 0.2083,
+      "step": 11877
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5274195379315126e-05,
+      "loss": 0.2174,
+      "step": 11878
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5260143229731436e-05,
+      "loss": 0.1951,
+      "step": 11879
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5246094422919264e-05,
+      "loss": 0.2059,
+      "step": 11880
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5232048959506925e-05,
+      "loss": 0.1988,
+      "step": 11881
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5218006840122644e-05,
+      "loss": 0.1928,
+      "step": 11882
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5203968065394424e-05,
+      "loss": 0.2166,
+      "step": 11883
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5189932635950207e-05,
+      "loss": 0.2116,
+      "step": 11884
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5175900552417674e-05,
+      "loss": 0.2149,
+      "step": 11885
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5161871815424464e-05,
+      "loss": 0.2182,
+      "step": 11886
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.514784642559801e-05,
+      "loss": 0.1964,
+      "step": 11887
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5133824383565564e-05,
+      "loss": 0.2112,
+      "step": 11888
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5119805689954323e-05,
+      "loss": 0.2054,
+      "step": 11889
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5105790345391222e-05,
+      "loss": 0.2058,
+      "step": 11890
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5091778350503158e-05,
+      "loss": 0.2117,
+      "step": 11891
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.507776970591681e-05,
+      "loss": 0.2156,
+      "step": 11892
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.506376441225867e-05,
+      "loss": 0.2061,
+      "step": 11893
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5049762470155214e-05,
+      "loss": 0.1966,
+      "step": 11894
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5035763880232597e-05,
+      "loss": 0.2092,
+      "step": 11895
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5021768643116993e-05,
+      "loss": 0.2138,
+      "step": 11896
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5007776759434276e-05,
+      "loss": 0.194,
+      "step": 11897
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.49937882298103e-05,
+      "loss": 0.2143,
+      "step": 11898
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4979803054870655e-05,
+      "loss": 0.206,
+      "step": 11899
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.496582123524087e-05,
+      "loss": 0.2042,
+      "step": 11900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4951842771546264e-05,
+      "loss": 0.2054,
+      "step": 11901
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.493786766441206e-05,
+      "loss": 0.2071,
+      "step": 11902
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4923895914463258e-05,
+      "loss": 0.2099,
+      "step": 11903
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4909927522324795e-05,
+      "loss": 0.2094,
+      "step": 11904
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.489596248862136e-05,
+      "loss": 0.2129,
+      "step": 11905
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4882000813977612e-05,
+      "loss": 0.2002,
+      "step": 11906
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4868042499017963e-05,
+      "loss": 0.2091,
+      "step": 11907
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.48540875443667e-05,
+      "loss": 0.1994,
+      "step": 11908
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4840135950647925e-05,
+      "loss": 0.2188,
+      "step": 11909
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4826187718485693e-05,
+      "loss": 0.1981,
+      "step": 11910
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4812242848503797e-05,
+      "loss": 0.2084,
+      "step": 11911
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.479830134132598e-05,
+      "loss": 0.2074,
+      "step": 11912
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4784363197575722e-05,
+      "loss": 0.201,
+      "step": 11913
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4770428417876467e-05,
+      "loss": 0.2057,
+      "step": 11914
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.47564970028514e-05,
+      "loss": 0.1989,
+      "step": 11915
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4742568953123656e-05,
+      "loss": 0.2054,
+      "step": 11916
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4728644269316138e-05,
+      "loss": 0.1953,
+      "step": 11917
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4714722952051673e-05,
+      "loss": 0.2033,
+      "step": 11918
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4700805001952854e-05,
+      "loss": 0.2023,
+      "step": 11919
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4686890419642204e-05,
+      "loss": 0.1955,
+      "step": 11920
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.467297920574202e-05,
+      "loss": 0.2115,
+      "step": 11921
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.465907136087453e-05,
+      "loss": 0.1889,
+      "step": 11922
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4645166885661732e-05,
+      "loss": 0.204,
+      "step": 11923
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4631265780725545e-05,
+      "loss": 0.2079,
+      "step": 11924
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.461736804668765e-05,
+      "loss": 0.2065,
+      "step": 11925
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4603473684169696e-05,
+      "loss": 0.2023,
+      "step": 11926
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.458958269379308e-05,
+      "loss": 0.2075,
+      "step": 11927
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4575695076179072e-05,
+      "loss": 0.2167,
+      "step": 11928
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4561810831948785e-05,
+      "loss": 0.2076,
+      "step": 11929
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.454792996172325e-05,
+      "loss": 0.2161,
+      "step": 11930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.453405246612325e-05,
+      "loss": 0.1997,
+      "step": 11931
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4520178345769497e-05,
+      "loss": 0.2026,
+      "step": 11932
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4506307601282474e-05,
+      "loss": 0.2066,
+      "step": 11933
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.449244023328261e-05,
+      "loss": 0.2111,
+      "step": 11934
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4478576242390073e-05,
+      "loss": 0.2085,
+      "step": 11935
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.446471562922499e-05,
+      "loss": 0.208,
+      "step": 11936
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4450858394407216e-05,
+      "loss": 0.1947,
+      "step": 11937
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4437004538556607e-05,
+      "loss": 0.2106,
+      "step": 11938
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4423154062292708e-05,
+      "loss": 0.2083,
+      "step": 11939
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4409306966235033e-05,
+      "loss": 0.2016,
+      "step": 11940
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.439546325100287e-05,
+      "loss": 0.2005,
+      "step": 11941
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4381622917215408e-05,
+      "loss": 0.1942,
+      "step": 11942
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.436778596549164e-05,
+      "loss": 0.2005,
+      "step": 11943
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4353952396450464e-05,
+      "loss": 0.2157,
+      "step": 11944
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.434012221071056e-05,
+      "loss": 0.1963,
+      "step": 11945
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4326295408890475e-05,
+      "loss": 0.2004,
+      "step": 11946
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4312471991608676e-05,
+      "loss": 0.197,
+      "step": 11947
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4298651959483375e-05,
+      "loss": 0.2084,
+      "step": 11948
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.428483531313267e-05,
+      "loss": 0.202,
+      "step": 11949
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4271022053174562e-05,
+      "loss": 0.2021,
+      "step": 11950
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4257212180226807e-05,
+      "loss": 0.1986,
+      "step": 11951
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4243405694907107e-05,
+      "loss": 0.1979,
+      "step": 11952
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.42296025978329e-05,
+      "loss": 0.2073,
+      "step": 11953
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.421580288962161e-05,
+      "loss": 0.2061,
+      "step": 11954
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4202006570890357e-05,
+      "loss": 0.2055,
+      "step": 11955
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4188213642256263e-05,
+      "loss": 0.2133,
+      "step": 11956
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.417442410433618e-05,
+      "loss": 0.1952,
+      "step": 11957
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4160637957746823e-05,
+      "loss": 0.2027,
+      "step": 11958
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4146855203104858e-05,
+      "loss": 0.2058,
+      "step": 11959
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4133075841026643e-05,
+      "loss": 0.198,
+      "step": 11960
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4119299872128543e-05,
+      "loss": 0.2008,
+      "step": 11961
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.410552729702662e-05,
+      "loss": 0.1978,
+      "step": 11962
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4091758116336926e-05,
+      "loss": 0.1929,
+      "step": 11963
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4077992330675236e-05,
+      "loss": 0.2048,
+      "step": 11964
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.406422994065729e-05,
+      "loss": 0.2023,
+      "step": 11965
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4050470946898553e-05,
+      "loss": 0.2033,
+      "step": 11966
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4036715350014462e-05,
+      "loss": 0.2036,
+      "step": 11967
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4022963150620204e-05,
+      "loss": 0.1934,
+      "step": 11968
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4009214349330867e-05,
+      "loss": 0.1958,
+      "step": 11969
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3995468946761336e-05,
+      "loss": 0.1982,
+      "step": 11970
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3981726943526428e-05,
+      "loss": 0.2074,
+      "step": 11971
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3967988340240723e-05,
+      "loss": 0.2014,
+      "step": 11972
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3954253137518733e-05,
+      "loss": 0.2213,
+      "step": 11973
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3940521335974707e-05,
+      "loss": 0.1957,
+      "step": 11974
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3926792936222875e-05,
+      "loss": 0.205,
+      "step": 11975
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3913067938877176e-05,
+      "loss": 0.201,
+      "step": 11976
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.389934634455153e-05,
+      "loss": 0.212,
+      "step": 11977
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3885628153859585e-05,
+      "loss": 0.2015,
+      "step": 11978
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3871913367414944e-05,
+      "loss": 0.203,
+      "step": 11979
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3858201985830965e-05,
+      "loss": 0.2094,
+      "step": 11980
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.384449400972093e-05,
+      "loss": 0.2023,
+      "step": 11981
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.38307894396979e-05,
+      "loss": 0.1962,
+      "step": 11982
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3817088276374843e-05,
+      "loss": 0.206,
+      "step": 11983
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3803390520364533e-05,
+      "loss": 0.2022,
+      "step": 11984
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.378969617227963e-05,
+      "loss": 0.2038,
+      "step": 11985
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.377600523273258e-05,
+      "loss": 0.1949,
+      "step": 11986
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.37623177023358e-05,
+      "loss": 0.1867,
+      "step": 11987
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3748633581701352e-05,
+      "loss": 0.225,
+      "step": 11988
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.373495287144135e-05,
+      "loss": 0.1956,
+      "step": 11989
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3721275572167623e-05,
+      "loss": 0.2146,
+      "step": 11990
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3707601684491942e-05,
+      "loss": 0.1992,
+      "step": 11991
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3693931209025812e-05,
+      "loss": 0.2119,
+      "step": 11992
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3680264146380716e-05,
+      "loss": 0.2032,
+      "step": 11993
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3666600497167867e-05,
+      "loss": 0.2091,
+      "step": 11994
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3652940261998435e-05,
+      "loss": 0.1999,
+      "step": 11995
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.363928344148332e-05,
+      "loss": 0.2185,
+      "step": 11996
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.362563003623337e-05,
+      "loss": 0.1965,
+      "step": 11997
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3611980046859207e-05,
+      "loss": 0.1985,
+      "step": 11998
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3598333473971378e-05,
+      "loss": 0.2049,
+      "step": 11999
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.358469031818017e-05,
+      "loss": 0.1998,
+      "step": 12000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3571050580095842e-05,
+      "loss": 0.2082,
+      "step": 12001
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.355741426032838e-05,
+      "loss": 0.2208,
+      "step": 12002
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3543781359487736e-05,
+      "loss": 0.1955,
+      "step": 12003
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3530151878183583e-05,
+      "loss": 0.1983,
+      "step": 12004
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3516525817025558e-05,
+      "loss": 0.2045,
+      "step": 12005
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.350290317662305e-05,
+      "loss": 0.2101,
+      "step": 12006
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3489283957585406e-05,
+      "loss": 0.2028,
+      "step": 12007
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3475668160521656e-05,
+      "loss": 0.1977,
+      "step": 12008
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3462055786040836e-05,
+      "loss": 0.2119,
+      "step": 12009
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3448446834751725e-05,
+      "loss": 0.2067,
+      "step": 12010
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3434841307263045e-05,
+      "loss": 0.2119,
+      "step": 12011
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3421239204183233e-05,
+      "loss": 0.2096,
+      "step": 12012
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3407640526120734e-05,
+      "loss": 0.1912,
+      "step": 12013
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3394045273683674e-05,
+      "loss": 0.2113,
+      "step": 12014
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3380453447480176e-05,
+      "loss": 0.1923,
+      "step": 12015
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.336686504811809e-05,
+      "loss": 0.2024,
+      "step": 12016
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3353280076205197e-05,
+      "loss": 0.2099,
+      "step": 12017
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.333969853234905e-05,
+      "loss": 0.2105,
+      "step": 12018
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.332612041715715e-05,
+      "loss": 0.2078,
+      "step": 12019
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3312545731236713e-05,
+      "loss": 0.2014,
+      "step": 12020
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3298974475194947e-05,
+      "loss": 0.2035,
+      "step": 12021
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3285406649638775e-05,
+      "loss": 0.2012,
+      "step": 12022
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3271842255175026e-05,
+      "loss": 0.1936,
+      "step": 12023
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3258281292410412e-05,
+      "loss": 0.2159,
+      "step": 12024
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3244723761951402e-05,
+      "loss": 0.2165,
+      "step": 12025
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3231169664404428e-05,
+      "loss": 0.2056,
+      "step": 12026
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3217619000375657e-05,
+      "loss": 0.2115,
+      "step": 12027
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3204071770471124e-05,
+      "loss": 0.2,
+      "step": 12028
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3190527975296807e-05,
+      "loss": 0.1996,
+      "step": 12029
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3176987615458378e-05,
+      "loss": 0.2058,
+      "step": 12030
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3163450691561506e-05,
+      "loss": 0.2018,
+      "step": 12031
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.314991720421157e-05,
+      "loss": 0.2017,
+      "step": 12032
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.313638715401393e-05,
+      "loss": 0.2135,
+      "step": 12033
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.312286054157369e-05,
+      "loss": 0.201,
+      "step": 12034
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.31093373674958e-05,
+      "loss": 0.2097,
+      "step": 12035
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3095817632385152e-05,
+      "loss": 0.1929,
+      "step": 12036
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.308230133684636e-05,
+      "loss": 0.2122,
+      "step": 12037
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3068788481484006e-05,
+      "loss": 0.2294,
+      "step": 12038
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.30552790669024e-05,
+      "loss": 0.2021,
+      "step": 12039
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.304177309370582e-05,
+      "loss": 0.2048,
+      "step": 12040
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3028270562498255e-05,
+      "loss": 0.2074,
+      "step": 12041
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3014771473883678e-05,
+      "loss": 0.1999,
+      "step": 12042
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3001275828465796e-05,
+      "loss": 0.2169,
+      "step": 12043
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2987783626848237e-05,
+      "loss": 0.2027,
+      "step": 12044
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2974294869634417e-05,
+      "loss": 0.2061,
+      "step": 12045
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2960809557427653e-05,
+      "loss": 0.1985,
+      "step": 12046
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2947327690831076e-05,
+      "loss": 0.1984,
+      "step": 12047
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.293384927044767e-05,
+      "loss": 0.2044,
+      "step": 12048
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.292037429688021e-05,
+      "loss": 0.1945,
+      "step": 12049
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.290690277073145e-05,
+      "loss": 0.201,
+      "step": 12050
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2893434692603855e-05,
+      "loss": 0.2076,
+      "step": 12051
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.287997006309983e-05,
+      "loss": 0.1996,
+      "step": 12052
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2866508882821537e-05,
+      "loss": 0.202,
+      "step": 12053
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2853051152371086e-05,
+      "loss": 0.217,
+      "step": 12054
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.283959687235033e-05,
+      "loss": 0.2193,
+      "step": 12055
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2826146043361074e-05,
+      "loss": 0.1989,
+      "step": 12056
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.281269866600485e-05,
+      "loss": 0.2076,
+      "step": 12057
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.279925474088316e-05,
+      "loss": 0.1927,
+      "step": 12058
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2785814268597217e-05,
+      "loss": 0.2058,
+      "step": 12059
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2772377249748233e-05,
+      "loss": 0.2049,
+      "step": 12060
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2758943684937107e-05,
+      "loss": 0.2004,
+      "step": 12061
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2745513574764733e-05,
+      "loss": 0.2073,
+      "step": 12062
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.273208691983171e-05,
+      "loss": 0.1978,
+      "step": 12063
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.271866372073861e-05,
+      "loss": 0.1994,
+      "step": 12064
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2705243978085734e-05,
+      "loss": 0.2032,
+      "step": 12065
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2691827692473343e-05,
+      "loss": 0.207,
+      "step": 12066
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2678414864501475e-05,
+      "loss": 0.2085,
+      "step": 12067
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2665005494769997e-05,
+      "loss": 0.2061,
+      "step": 12068
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2651599583878624e-05,
+      "loss": 0.2137,
+      "step": 12069
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.263819713242702e-05,
+      "loss": 0.2035,
+      "step": 12070
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2624798141014535e-05,
+      "loss": 0.2032,
+      "step": 12071
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2611402610240517e-05,
+      "loss": 0.2014,
+      "step": 12072
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2598010540704017e-05,
+      "loss": 0.2064,
+      "step": 12073
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.258462193300407e-05,
+      "loss": 0.1974,
+      "step": 12074
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2571236787739425e-05,
+      "loss": 0.2019,
+      "step": 12075
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2557855105508795e-05,
+      "loss": 0.199,
+      "step": 12076
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2544476886910616e-05,
+      "loss": 0.196,
+      "step": 12077
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2531102132543303e-05,
+      "loss": 0.2051,
+      "step": 12078
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2517730843004992e-05,
+      "loss": 0.204,
+      "step": 12079
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.250436301889377e-05,
+      "loss": 0.1944,
+      "step": 12080
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.249099866080746e-05,
+      "loss": 0.1935,
+      "step": 12081
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.247763776934384e-05,
+      "loss": 0.2015,
+      "step": 12082
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2464280345100453e-05,
+      "loss": 0.2065,
+      "step": 12083
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2450926388674732e-05,
+      "loss": 0.1962,
+      "step": 12084
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.243757590066392e-05,
+      "loss": 0.1958,
+      "step": 12085
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2424228881665144e-05,
+      "loss": 0.1963,
+      "step": 12086
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2410885332275357e-05,
+      "loss": 0.1882,
+      "step": 12087
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2397545253091344e-05,
+      "loss": 0.2005,
+      "step": 12088
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2384208644709702e-05,
+      "loss": 0.1996,
+      "step": 12089
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2370875507726996e-05,
+      "loss": 0.197,
+      "step": 12090
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.235754584273949e-05,
+      "loss": 0.1952,
+      "step": 12091
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2344219650343424e-05,
+      "loss": 0.2083,
+      "step": 12092
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2330896931134736e-05,
+      "loss": 0.1963,
+      "step": 12093
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2317577685709378e-05,
+      "loss": 0.2158,
+      "step": 12094
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2304261914662984e-05,
+      "loss": 0.201,
+      "step": 12095
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.229094961859116e-05,
+      "loss": 0.2009,
+      "step": 12096
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.227764079808926e-05,
+      "loss": 0.2046,
+      "step": 12097
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2264335453752584e-05,
+      "loss": 0.195,
+      "step": 12098
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2251033586176152e-05,
+      "loss": 0.2152,
+      "step": 12099
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2237735195954955e-05,
+      "loss": 0.1995,
+      "step": 12100
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2224440283683744e-05,
+      "loss": 0.1931,
+      "step": 12101
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2211148849957108e-05,
+      "loss": 0.1987,
+      "step": 12102
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.219786089536958e-05,
+      "loss": 0.1808,
+      "step": 12103
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.21845764205154e-05,
+      "loss": 0.2052,
+      "step": 12104
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2171295425988782e-05,
+      "loss": 0.2007,
+      "step": 12105
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2158017912383653e-05,
+      "loss": 0.2041,
+      "step": 12106
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.214474388029395e-05,
+      "loss": 0.2033,
+      "step": 12107
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.213147333031329e-05,
+      "loss": 0.1991,
+      "step": 12108
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.21182062630352e-05,
+      "loss": 0.2029,
+      "step": 12109
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2104942679053108e-05,
+      "loss": 0.2036,
+      "step": 12110
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.209168257896017e-05,
+      "loss": 0.2004,
+      "step": 12111
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2078425963349504e-05,
+      "loss": 0.2031,
+      "step": 12112
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.206517283281401e-05,
+      "loss": 0.191,
+      "step": 12113
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.205192318794639e-05,
+      "loss": 0.2253,
+      "step": 12114
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2038677029339293e-05,
+      "loss": 0.2011,
+      "step": 12115
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2025434357585127e-05,
+      "loss": 0.2037,
+      "step": 12116
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2012195173276218e-05,
+      "loss": 0.2001,
+      "step": 12117
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1998959477004634e-05,
+      "loss": 0.2062,
+      "step": 12118
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.19857272693624e-05,
+      "loss": 0.2113,
+      "step": 12119
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.197249855094129e-05,
+      "loss": 0.2006,
+      "step": 12120
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.195927332233302e-05,
+      "loss": 0.2017,
+      "step": 12121
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1946051584129024e-05,
+      "loss": 0.2,
+      "step": 12122
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1932833336920712e-05,
+      "loss": 0.2067,
+      "step": 12123
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.191961858129923e-05,
+      "loss": 0.198,
+      "step": 12124
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.190640731785565e-05,
+      "loss": 0.2046,
+      "step": 12125
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.189319954718081e-05,
+      "loss": 0.1941,
+      "step": 12126
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1879995269865507e-05,
+      "loss": 0.1882,
+      "step": 12127
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.186679448650021e-05,
+      "loss": 0.193,
+      "step": 12128
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.18535971976754e-05,
+      "loss": 0.1981,
+      "step": 12129
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1840403403981278e-05,
+      "loss": 0.2068,
+      "step": 12130
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1827213106008015e-05,
+      "loss": 0.2034,
+      "step": 12131
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.181402630434547e-05,
+      "loss": 0.2037,
+      "step": 12132
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1800842999583494e-05,
+      "loss": 0.193,
+      "step": 12133
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.178766319231167e-05,
+      "loss": 0.1974,
+      "step": 12134
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1774486883119517e-05,
+      "loss": 0.1904,
+      "step": 12135
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1761314072596297e-05,
+      "loss": 0.201,
+      "step": 12136
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1748144761331214e-05,
+      "loss": 0.2031,
+      "step": 12137
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.173497894991323e-05,
+      "loss": 0.2016,
+      "step": 12138
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.172181663893126e-05,
+      "loss": 0.2003,
+      "step": 12139
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.17086578289739e-05,
+      "loss": 0.2022,
+      "step": 12140
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1695502520629775e-05,
+      "loss": 0.1993,
+      "step": 12141
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1682350714487186e-05,
+      "loss": 0.1935,
+      "step": 12142
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1669202411134415e-05,
+      "loss": 0.1889,
+      "step": 12143
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1656057611159462e-05,
+      "loss": 0.1998,
+      "step": 12144
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1642916315150307e-05,
+      "loss": 0.2089,
+      "step": 12145
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1629778523694634e-05,
+      "loss": 0.2013,
+      "step": 12146
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1616644237380112e-05,
+      "loss": 0.1911,
+      "step": 12147
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1603513456794077e-05,
+      "loss": 0.2028,
+      "step": 12148
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.159038618252389e-05,
+      "loss": 0.1881,
+      "step": 12149
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1577262415156618e-05,
+      "loss": 0.2032,
+      "step": 12150
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1564142155279277e-05,
+      "loss": 0.1955,
+      "step": 12151
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.155102540347862e-05,
+      "loss": 0.2096,
+      "step": 12152
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1537912160341367e-05,
+      "loss": 0.1934,
+      "step": 12153
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.152480242645396e-05,
+      "loss": 0.1955,
+      "step": 12154
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1511696202402763e-05,
+      "loss": 0.1929,
+      "step": 12155
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1498593488773944e-05,
+      "loss": 0.1987,
+      "step": 12156
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.148549428615355e-05,
+      "loss": 0.2012,
+      "step": 12157
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1472398595127408e-05,
+      "loss": 0.1986,
+      "step": 12158
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1459306416281277e-05,
+      "loss": 0.1898,
+      "step": 12159
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.144621775020066e-05,
+      "loss": 0.2056,
+      "step": 12160
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1433132597470994e-05,
+      "loss": 0.204,
+      "step": 12161
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1420050958677485e-05,
+      "loss": 0.2052,
+      "step": 12162
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.140697283440526e-05,
+      "loss": 0.1948,
+      "step": 12163
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1393898225239196e-05,
+      "loss": 0.2019,
+      "step": 12164
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.138082713176409e-05,
+      "loss": 0.1987,
+      "step": 12165
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.136775955456456e-05,
+      "loss": 0.2089,
+      "step": 12166
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.135469549422503e-05,
+      "loss": 0.1981,
+      "step": 12167
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.134163495132978e-05,
+      "loss": 0.2082,
+      "step": 12168
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1328577926463013e-05,
+      "loss": 0.2017,
+      "step": 12169
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1315524420208632e-05,
+      "loss": 0.2095,
+      "step": 12170
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1302474433150532e-05,
+      "loss": 0.1948,
+      "step": 12171
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.128942796587232e-05,
+      "loss": 0.1927,
+      "step": 12172
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1276385018957567e-05,
+      "loss": 0.1918,
+      "step": 12173
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.126334559298956e-05,
+      "loss": 0.194,
+      "step": 12174
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.125030968855155e-05,
+      "loss": 0.1991,
+      "step": 12175
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.123727730622651e-05,
+      "loss": 0.2104,
+      "step": 12176
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1224248446597393e-05,
+      "loss": 0.2154,
+      "step": 12177
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1211223110246868e-05,
+      "loss": 0.2143,
+      "step": 12178
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.11982012977575e-05,
+      "loss": 0.1973,
+      "step": 12179
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.118518300971173e-05,
+      "loss": 0.2053,
+      "step": 12180
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.117216824669176e-05,
+      "loss": 0.1955,
+      "step": 12181
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1159157009279718e-05,
+      "loss": 0.2025,
+      "step": 12182
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1146149298057516e-05,
+      "loss": 0.2017,
+      "step": 12183
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1133145113606945e-05,
+      "loss": 0.2054,
+      "step": 12184
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1120144456509605e-05,
+      "loss": 0.1934,
+      "step": 12185
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1107147327346977e-05,
+      "loss": 0.2033,
+      "step": 12186
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.109415372670036e-05,
+      "loss": 0.2022,
+      "step": 12187
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.108116365515087e-05,
+      "loss": 0.197,
+      "step": 12188
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.106817711327953e-05,
+      "loss": 0.198,
+      "step": 12189
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1055194101667154e-05,
+      "loss": 0.1997,
+      "step": 12190
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1042214620894386e-05,
+      "loss": 0.1868,
+      "step": 12191
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1029238671541794e-05,
+      "loss": 0.1794,
+      "step": 12192
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1016266254189666e-05,
+      "loss": 0.2039,
+      "step": 12193
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1003297369418263e-05,
+      "loss": 0.2063,
+      "step": 12194
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.099033201780758e-05,
+      "loss": 0.2007,
+      "step": 12195
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.097737019993753e-05,
+      "loss": 0.1973,
+      "step": 12196
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0964411916387794e-05,
+      "loss": 0.1977,
+      "step": 12197
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0951457167737985e-05,
+      "loss": 0.2062,
+      "step": 12198
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0938505954567466e-05,
+      "loss": 0.1903,
+      "step": 12199
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0925558277455535e-05,
+      "loss": 0.1903,
+      "step": 12200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0912614136981225e-05,
+      "loss": 0.1907,
+      "step": 12201
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.089967353372352e-05,
+      "loss": 0.1924,
+      "step": 12202
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0886736468261138e-05,
+      "loss": 0.2045,
+      "step": 12203
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.087380294117276e-05,
+      "loss": 0.2025,
+      "step": 12204
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0860872953036792e-05,
+      "loss": 0.2093,
+      "step": 12205
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.084794650443157e-05,
+      "loss": 0.1978,
+      "step": 12206
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.083502359593521e-05,
+      "loss": 0.1893,
+      "step": 12207
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0822104228125715e-05,
+      "loss": 0.2023,
+      "step": 12208
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.080918840158086e-05,
+      "loss": 0.1994,
+      "step": 12209
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0796276116878377e-05,
+      "loss": 0.211,
+      "step": 12210
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0783367374595708e-05,
+      "loss": 0.2028,
+      "step": 12211
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0770462175310267e-05,
+      "loss": 0.2033,
+      "step": 12212
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0757560519599185e-05,
+      "loss": 0.1996,
+      "step": 12213
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.074466240803956e-05,
+      "loss": 0.2004,
+      "step": 12214
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0731767841208193e-05,
+      "loss": 0.2012,
+      "step": 12215
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0718876819681854e-05,
+      "loss": 0.194,
+      "step": 12216
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.070598934403706e-05,
+      "loss": 0.1961,
+      "step": 12217
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0693105414850256e-05,
+      "loss": 0.1982,
+      "step": 12218
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.068022503269763e-05,
+      "loss": 0.2056,
+      "step": 12219
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.06673481981553e-05,
+      "loss": 0.2087,
+      "step": 12220
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.065447491179916e-05,
+      "loss": 0.1922,
+      "step": 12221
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0641605174205014e-05,
+      "loss": 0.2074,
+      "step": 12222
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0628738985948402e-05,
+      "loss": 0.2014,
+      "step": 12223
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0615876347604833e-05,
+      "loss": 0.1933,
+      "step": 12224
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0603017259749557e-05,
+      "loss": 0.195,
+      "step": 12225
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0590161722957723e-05,
+      "loss": 0.1968,
+      "step": 12226
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0577309737804296e-05,
+      "loss": 0.1992,
+      "step": 12227
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.056446130486408e-05,
+      "loss": 0.2043,
+      "step": 12228
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0551616424711707e-05,
+      "loss": 0.2091,
+      "step": 12229
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0538775097921714e-05,
+      "loss": 0.1939,
+      "step": 12230
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.052593732506838e-05,
+      "loss": 0.1966,
+      "step": 12231
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0513103106725938e-05,
+      "loss": 0.1896,
+      "step": 12232
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.050027244346835e-05,
+      "loss": 0.2091,
+      "step": 12233
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0487445335869525e-05,
+      "loss": 0.1957,
+      "step": 12234
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.047462178450311e-05,
+      "loss": 0.194,
+      "step": 12235
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.046180178994269e-05,
+      "loss": 0.2029,
+      "step": 12236
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.044898535276161e-05,
+      "loss": 0.1934,
+      "step": 12237
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.043617247353311e-05,
+      "loss": 0.1869,
+      "step": 12238
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0423363152830233e-05,
+      "loss": 0.1956,
+      "step": 12239
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0410557391225914e-05,
+      "loss": 0.2104,
+      "step": 12240
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0397755189292856e-05,
+      "loss": 0.2106,
+      "step": 12241
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0384956547603685e-05,
+      "loss": 0.205,
+      "step": 12242
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0372161466730778e-05,
+      "loss": 0.202,
+      "step": 12243
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0359369947246444e-05,
+      "loss": 0.1988,
+      "step": 12244
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.034658198972278e-05,
+      "loss": 0.1873,
+      "step": 12245
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0333797594731685e-05,
+      "loss": 0.1964,
+      "step": 12246
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0321016762845026e-05,
+      "loss": 0.1964,
+      "step": 12247
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0308239494634383e-05,
+      "loss": 0.1961,
+      "step": 12248
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.02954657906712e-05,
+      "loss": 0.2086,
+      "step": 12249
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.028269565152685e-05,
+      "loss": 0.2039,
+      "step": 12250
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.026992907777242e-05,
+      "loss": 0.1957,
+      "step": 12251
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0257166069978962e-05,
+      "loss": 0.2026,
+      "step": 12252
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.024440662871725e-05,
+      "loss": 0.1846,
+      "step": 12253
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0231650754558008e-05,
+      "loss": 0.1991,
+      "step": 12254
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.021889844807169e-05,
+      "loss": 0.196,
+      "step": 12255
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0206149709828704e-05,
+      "loss": 0.1944,
+      "step": 12256
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0193404540399217e-05,
+      "loss": 0.2025,
+      "step": 12257
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0180662940353244e-05,
+      "loss": 0.1986,
+      "step": 12258
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0167924910260704e-05,
+      "loss": 0.2045,
+      "step": 12259
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.015519045069125e-05,
+      "loss": 0.2148,
+      "step": 12260
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.01424595622145e-05,
+      "loss": 0.2002,
+      "step": 12261
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0129732245399803e-05,
+      "loss": 0.1983,
+      "step": 12262
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0117008500816425e-05,
+      "loss": 0.1872,
+      "step": 12263
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0104288329033395e-05,
+      "loss": 0.2095,
+      "step": 12264
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0091571730619687e-05,
+      "loss": 0.1976,
+      "step": 12265
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.007885870614401e-05,
+      "loss": 0.1971,
+      "step": 12266
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0066149256174993e-05,
+      "loss": 0.2041,
+      "step": 12267
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0053443381281057e-05,
+      "loss": 0.1996,
+      "step": 12268
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0040741082030477e-05,
+      "loss": 0.2035,
+      "step": 12269
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.002804235899134e-05,
+      "loss": 0.2147,
+      "step": 12270
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.001534721273165e-05,
+      "loss": 0.1918,
+      "step": 12271
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0002655643819166e-05,
+      "loss": 0.2086,
+      "step": 12272
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.998996765282156e-05,
+      "loss": 0.1839,
+      "step": 12273
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9977283240306265e-05,
+      "loss": 0.2551,
+      "step": 12274
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.996460240684064e-05,
+      "loss": 0.1958,
+      "step": 12275
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9951925152991802e-05,
+      "loss": 0.1927,
+      "step": 12276
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.993925147932678e-05,
+      "loss": 0.2016,
+      "step": 12277
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.992658138641237e-05,
+      "loss": 0.1946,
+      "step": 12278
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9913914874815297e-05,
+      "loss": 0.2091,
+      "step": 12279
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9901251945102027e-05,
+      "loss": 0.1961,
+      "step": 12280
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.988859259783896e-05,
+      "loss": 0.2004,
+      "step": 12281
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9875936833592244e-05,
+      "loss": 0.2063,
+      "step": 12282
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.986328465292797e-05,
+      "loss": 0.1971,
+      "step": 12283
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.985063605641195e-05,
+      "loss": 0.2098,
+      "step": 12284
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.983799104460995e-05,
+      "loss": 0.2023,
+      "step": 12285
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9825349618087476e-05,
+      "loss": 0.1914,
+      "step": 12286
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9812711777409998e-05,
+      "loss": 0.2066,
+      "step": 12287
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9800077523142647e-05,
+      "loss": 0.207,
+      "step": 12288
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9787446855850566e-05,
+      "loss": 0.1985,
+      "step": 12289
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9774819776098628e-05,
+      "loss": 0.1956,
+      "step": 12290
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.976219628445163e-05,
+      "loss": 0.1944,
+      "step": 12291
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.97495763814741e-05,
+      "loss": 0.1956,
+      "step": 12292
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9736960067730538e-05,
+      "loss": 0.1958,
+      "step": 12293
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9724347343785142e-05,
+      "loss": 0.1963,
+      "step": 12294
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9711738210202092e-05,
+      "loss": 0.196,
+      "step": 12295
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9699132667545283e-05,
+      "loss": 0.195,
+      "step": 12296
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.968653071637855e-05,
+      "loss": 0.1957,
+      "step": 12297
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9673932357265455e-05,
+      "loss": 0.1922,
+      "step": 12298
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9661337590769545e-05,
+      "loss": 0.1913,
+      "step": 12299
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9648746417454055e-05,
+      "loss": 0.1967,
+      "step": 12300
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.963615883788219e-05,
+      "loss": 0.2046,
+      "step": 12301
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9623574852616878e-05,
+      "loss": 0.1963,
+      "step": 12302
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9610994462221e-05,
+      "loss": 0.188,
+      "step": 12303
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9598417667257162e-05,
+      "loss": 0.2173,
+      "step": 12304
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.958584446828793e-05,
+      "loss": 0.1893,
+      "step": 12305
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9573274865875568e-05,
+      "loss": 0.1901,
+      "step": 12306
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9560708860582367e-05,
+      "loss": 0.1982,
+      "step": 12307
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9548146452970217e-05,
+      "loss": 0.1872,
+      "step": 12308
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9535587643601072e-05,
+      "loss": 0.1854,
+      "step": 12309
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.952303243303657e-05,
+      "loss": 0.2063,
+      "step": 12310
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9510480821838295e-05,
+      "loss": 0.1929,
+      "step": 12311
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.949793281056759e-05,
+      "loss": 0.2019,
+      "step": 12312
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.94853883997857e-05,
+      "loss": 0.1998,
+      "step": 12313
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9472847590053633e-05,
+      "loss": 0.2057,
+      "step": 12314
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9460310381932345e-05,
+      "loss": 0.2039,
+      "step": 12315
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9447776775982507e-05,
+      "loss": 0.1987,
+      "step": 12316
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.943524677276474e-05,
+      "loss": 0.1955,
+      "step": 12317
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9422720372839408e-05,
+      "loss": 0.1901,
+      "step": 12318
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.941019757676681e-05,
+      "loss": 0.197,
+      "step": 12319
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9397678385106977e-05,
+      "loss": 0.1992,
+      "step": 12320
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9385162798419888e-05,
+      "loss": 0.1899,
+      "step": 12321
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.937265081726528e-05,
+      "loss": 0.2036,
+      "step": 12322
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9360142442202745e-05,
+      "loss": 0.2026,
+      "step": 12323
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9347637673791764e-05,
+      "loss": 0.1905,
+      "step": 12324
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.933513651259157e-05,
+      "loss": 0.1913,
+      "step": 12325
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.932263895916132e-05,
+      "loss": 0.1956,
+      "step": 12326
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9310145014059978e-05,
+      "loss": 0.2034,
+      "step": 12327
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.929765467784629e-05,
+      "loss": 0.1956,
+      "step": 12328
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9285167951078943e-05,
+      "loss": 0.2078,
+      "step": 12329
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.927268483431637e-05,
+      "loss": 0.2145,
+      "step": 12330
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.926020532811693e-05,
+      "loss": 0.1997,
+      "step": 12331
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.924772943303873e-05,
+      "loss": 0.1818,
+      "step": 12332
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9235257149639806e-05,
+      "loss": 0.1872,
+      "step": 12333
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.922278847847796e-05,
+      "loss": 0.2034,
+      "step": 12334
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9210323420110833e-05,
+      "loss": 0.2023,
+      "step": 12335
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9197861975095975e-05,
+      "loss": 0.1922,
+      "step": 12336
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9185404143990683e-05,
+      "loss": 0.1824,
+      "step": 12337
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.91729499273522e-05,
+      "loss": 0.1924,
+      "step": 12338
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9160499325737478e-05,
+      "loss": 0.2026,
+      "step": 12339
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9148052339703437e-05,
+      "loss": 0.2014,
+      "step": 12340
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9135608969806716e-05,
+      "loss": 0.2073,
+      "step": 12341
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.912316921660392e-05,
+      "loss": 0.2031,
+      "step": 12342
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9110733080651343e-05,
+      "loss": 0.202,
+      "step": 12343
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.1977,
+      "step": 12344
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.908587166272168e-05,
+      "loss": 0.196,
+      "step": 12345
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9073446381856553e-05,
+      "loss": 0.2081,
+      "step": 12346
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9061024720465505e-05,
+      "loss": 0.2045,
+      "step": 12347
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9048606679104175e-05,
+      "loss": 0.1913,
+      "step": 12348
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9036192258327933e-05,
+      "loss": 0.2019,
+      "step": 12349
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9023781458692048e-05,
+      "loss": 0.1993,
+      "step": 12350
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9011374280751548e-05,
+      "loss": 0.1946,
+      "step": 12351
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8998970725061416e-05,
+      "loss": 0.1841,
+      "step": 12352
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.898657079217634e-05,
+      "loss": 0.1945,
+      "step": 12353
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8974174482650965e-05,
+      "loss": 0.2009,
+      "step": 12354
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.896178179703968e-05,
+      "loss": 0.196,
+      "step": 12355
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8949392735896797e-05,
+      "loss": 0.2009,
+      "step": 12356
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8937007299776376e-05,
+      "loss": 0.1893,
+      "step": 12357
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8924625489232407e-05,
+      "loss": 0.201,
+      "step": 12358
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.891224730481862e-05,
+      "loss": 0.1937,
+      "step": 12359
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8899872747088686e-05,
+      "loss": 0.1962,
+      "step": 12360
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.888750181659602e-05,
+      "loss": 0.1918,
+      "step": 12361
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8875134513893955e-05,
+      "loss": 0.1987,
+      "step": 12362
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.886277083953558e-05,
+      "loss": 0.195,
+      "step": 12363
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8850410794073915e-05,
+      "loss": 0.1996,
+      "step": 12364
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.883805437806172e-05,
+      "loss": 0.1882,
+      "step": 12365
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8825701592051715e-05,
+      "loss": 0.1956,
+      "step": 12366
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8813352436596276e-05,
+      "loss": 0.1985,
+      "step": 12367
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.880100691224781e-05,
+      "loss": 0.1895,
+      "step": 12368
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8788665019558417e-05,
+      "loss": 0.1907,
+      "step": 12369
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8776326759080152e-05,
+      "loss": 0.1956,
+      "step": 12370
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8763992131364795e-05,
+      "loss": 0.1995,
+      "step": 12371
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.875166113696406e-05,
+      "loss": 0.1937,
+      "step": 12372
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8739333776429423e-05,
+      "loss": 0.1942,
+      "step": 12373
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8727010050312266e-05,
+      "loss": 0.1956,
+      "step": 12374
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8714689959163723e-05,
+      "loss": 0.1965,
+      "step": 12375
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8702373503534875e-05,
+      "loss": 0.2079,
+      "step": 12376
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.869006068397652e-05,
+      "loss": 0.2042,
+      "step": 12377
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8677751501039408e-05,
+      "loss": 0.1959,
+      "step": 12378
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.866544595527403e-05,
+      "loss": 0.1865,
+      "step": 12379
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8653144047230797e-05,
+      "loss": 0.195,
+      "step": 12380
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.864084577745987e-05,
+      "loss": 0.199,
+      "step": 12381
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8628551146511352e-05,
+      "loss": 0.1924,
+      "step": 12382
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8616260154935063e-05,
+      "loss": 0.1922,
+      "step": 12383
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.860397280328077e-05,
+      "loss": 0.2025,
+      "step": 12384
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8591689092097987e-05,
+      "loss": 0.1952,
+      "step": 12385
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.85794090219362e-05,
+      "loss": 0.2025,
+      "step": 12386
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8567132593344504e-05,
+      "loss": 0.1932,
+      "step": 12387
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.855485980687208e-05,
+      "loss": 0.1829,
+      "step": 12388
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.854259066306776e-05,
+      "loss": 0.1927,
+      "step": 12389
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8530325162480333e-05,
+      "loss": 0.1973,
+      "step": 12390
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.851806330565835e-05,
+      "loss": 0.1938,
+      "step": 12391
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8505805093150265e-05,
+      "loss": 0.1979,
+      "step": 12392
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8493550525504277e-05,
+      "loss": 0.1948,
+      "step": 12393
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8481299603268533e-05,
+      "loss": 0.1915,
+      "step": 12394
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.846905232699092e-05,
+      "loss": 0.2061,
+      "step": 12395
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.845680869721923e-05,
+      "loss": 0.1938,
+      "step": 12396
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.844456871450103e-05,
+      "loss": 0.1986,
+      "step": 12397
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.843233237938381e-05,
+      "loss": 0.1965,
+      "step": 12398
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8420099692414815e-05,
+      "loss": 0.1922,
+      "step": 12399
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8407870654141134e-05,
+      "loss": 0.2027,
+      "step": 12400
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8395645265109763e-05,
+      "loss": 0.1971,
+      "step": 12401
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.838342352586744e-05,
+      "loss": 0.2147,
+      "step": 12402
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8371205436960836e-05,
+      "loss": 0.1919,
+      "step": 12403
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8358990998936354e-05,
+      "loss": 0.1885,
+      "step": 12404
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8346780212340352e-05,
+      "loss": 0.2062,
+      "step": 12405
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.833457307771894e-05,
+      "loss": 0.1978,
+      "step": 12406
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.832236959561805e-05,
+      "loss": 0.1926,
+      "step": 12407
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.831016976658354e-05,
+      "loss": 0.1968,
+      "step": 12408
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.829797359116101e-05,
+      "loss": 0.195,
+      "step": 12409
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8285781069895978e-05,
+      "loss": 0.194,
+      "step": 12410
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.827359220333371e-05,
+      "loss": 0.1938,
+      "step": 12411
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8261406992019425e-05,
+      "loss": 0.1984,
+      "step": 12412
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8249225436498073e-05,
+      "loss": 0.1843,
+      "step": 12413
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8237047537314456e-05,
+      "loss": 0.198,
+      "step": 12414
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.822487329501329e-05,
+      "loss": 0.2028,
+      "step": 12415
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8212702710139018e-05,
+      "loss": 0.1947,
+      "step": 12416
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8200535783236027e-05,
+      "loss": 0.2058,
+      "step": 12417
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8188372514848428e-05,
+      "loss": 0.1987,
+      "step": 12418
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8176212905520294e-05,
+      "loss": 0.2007,
+      "step": 12419
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8164056955795417e-05,
+      "loss": 0.1887,
+      "step": 12420
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.815190466621751e-05,
+      "loss": 0.1979,
+      "step": 12421
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8139756037330058e-05,
+      "loss": 0.194,
+      "step": 12422
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8127611069676455e-05,
+      "loss": 0.1991,
+      "step": 12423
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8115469763799842e-05,
+      "loss": 0.1921,
+      "step": 12424
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.81033321202433e-05,
+      "loss": 0.187,
+      "step": 12425
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.809119813954965e-05,
+      "loss": 0.1961,
+      "step": 12426
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8079067822261598e-05,
+      "loss": 0.1968,
+      "step": 12427
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.806694116892166e-05,
+      "loss": 0.1862,
+      "step": 12428
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8054818180072243e-05,
+      "loss": 0.1929,
+      "step": 12429
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8042698856255514e-05,
+      "loss": 0.1879,
+      "step": 12430
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8030583198013552e-05,
+      "loss": 0.191,
+      "step": 12431
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8018471205888188e-05,
+      "loss": 0.1955,
+      "step": 12432
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.80063628804212e-05,
+      "loss": 0.1905,
+      "step": 12433
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.799425822215407e-05,
+      "loss": 0.1747,
+      "step": 12434
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.798215723162825e-05,
+      "loss": 0.2066,
+      "step": 12435
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7970059909384896e-05,
+      "loss": 0.1931,
+      "step": 12436
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.795796625596512e-05,
+      "loss": 0.2062,
+      "step": 12437
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7945876271909778e-05,
+      "loss": 0.1859,
+      "step": 12438
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.793378995775964e-05,
+      "loss": 0.2073,
+      "step": 12439
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7921707314055213e-05,
+      "loss": 0.2069,
+      "step": 12440
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.790962834133697e-05,
+      "loss": 0.1997,
+      "step": 12441
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7897553040145077e-05,
+      "loss": 0.1891,
+      "step": 12442
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7885481411019666e-05,
+      "loss": 0.2122,
+      "step": 12443
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.78734134545006e-05,
+      "loss": 0.1978,
+      "step": 12444
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.786134917112766e-05,
+      "loss": 0.198,
+      "step": 12445
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7849288561440426e-05,
+      "loss": 0.18,
+      "step": 12446
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.783723162597828e-05,
+      "loss": 0.1994,
+      "step": 12447
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.782517836528047e-05,
+      "loss": 0.1931,
+      "step": 12448
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7813128779886135e-05,
+      "loss": 0.1864,
+      "step": 12449
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.780108287033414e-05,
+      "loss": 0.1993,
+      "step": 12450
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.77890406371633e-05,
+      "loss": 0.1873,
+      "step": 12451
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7777002080912153e-05,
+      "loss": 0.1888,
+      "step": 12452
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7764967202119175e-05,
+      "loss": 0.1934,
+      "step": 12453
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7752936001322584e-05,
+      "loss": 0.1874,
+      "step": 12454
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7740908479060537e-05,
+      "loss": 0.2016,
+      "step": 12455
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7728884635870912e-05,
+      "loss": 0.1925,
+      "step": 12456
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.771686447229154e-05,
+      "loss": 0.2043,
+      "step": 12457
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7704847988859963e-05,
+      "loss": 0.1915,
+      "step": 12458
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7692835186113688e-05,
+      "loss": 0.1914,
+      "step": 12459
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7680826064589938e-05,
+      "loss": 0.1945,
+      "step": 12460
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7668820624825875e-05,
+      "loss": 0.2044,
+      "step": 12461
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7656818867358393e-05,
+      "loss": 0.2028,
+      "step": 12462
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7644820792724324e-05,
+      "loss": 0.192,
+      "step": 12463
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7632826401460257e-05,
+      "loss": 0.1979,
+      "step": 12464
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7620835694102668e-05,
+      "loss": 0.1963,
+      "step": 12465
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7608848671187848e-05,
+      "loss": 0.1988,
+      "step": 12466
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.75968653332519e-05,
+      "loss": 0.2056,
+      "step": 12467
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7584885680830765e-05,
+      "loss": 0.1984,
+      "step": 12468
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7572909714460294e-05,
+      "loss": 0.2002,
+      "step": 12469
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7560937434676073e-05,
+      "loss": 0.1888,
+      "step": 12470
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7548968842013603e-05,
+      "loss": 0.1906,
+      "step": 12471
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7537003937008134e-05,
+      "loss": 0.1937,
+      "step": 12472
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7525042720194863e-05,
+      "loss": 0.2006,
+      "step": 12473
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.751308519210869e-05,
+      "loss": 0.1917,
+      "step": 12474
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7501131353284505e-05,
+      "loss": 0.188,
+      "step": 12475
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7489181204256865e-05,
+      "loss": 0.1883,
+      "step": 12476
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.747723474556031e-05,
+      "loss": 0.1912,
+      "step": 12477
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7465291977729127e-05,
+      "loss": 0.1847,
+      "step": 12478
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.745335290129744e-05,
+      "loss": 0.2038,
+      "step": 12479
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7441417516799252e-05,
+      "loss": 0.1875,
+      "step": 12480
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7429485824768366e-05,
+      "loss": 0.1921,
+      "step": 12481
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.741755782573846e-05,
+      "loss": 0.197,
+      "step": 12482
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.740563352024297e-05,
+      "loss": 0.1956,
+      "step": 12483
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7393712908815275e-05,
+      "loss": 0.1914,
+      "step": 12484
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.738179599198847e-05,
+      "loss": 0.183,
+      "step": 12485
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.73698827702956e-05,
+      "loss": 0.1838,
+      "step": 12486
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7357973244269453e-05,
+      "loss": 0.1973,
+      "step": 12487
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.734606741444268e-05,
+      "loss": 0.1988,
+      "step": 12488
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7334165281347813e-05,
+      "loss": 0.2028,
+      "step": 12489
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7322266845517153e-05,
+      "loss": 0.1884,
+      "step": 12490
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7310372107482843e-05,
+      "loss": 0.1974,
+      "step": 12491
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.729848106777693e-05,
+      "loss": 0.2002,
+      "step": 12492
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.72865937269312e-05,
+      "loss": 0.1915,
+      "step": 12493
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7274710085477354e-05,
+      "loss": 0.187,
+      "step": 12494
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.726283014394686e-05,
+      "loss": 0.1981,
+      "step": 12495
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7250953902871093e-05,
+      "loss": 0.1901,
+      "step": 12496
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7239081362781173e-05,
+      "loss": 0.2032,
+      "step": 12497
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.722721252420816e-05,
+      "loss": 0.1949,
+      "step": 12498
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7215347387682836e-05,
+      "loss": 0.1857,
+      "step": 12499
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7203485953735932e-05,
+      "loss": 0.196,
+      "step": 12500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7191628222897906e-05,
+      "loss": 0.1938,
+      "step": 12501
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.717977419569914e-05,
+      "loss": 0.202,
+      "step": 12502
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7167923872669767e-05,
+      "loss": 0.1891,
+      "step": 12503
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.715607725433984e-05,
+      "loss": 0.1938,
+      "step": 12504
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7144234341239173e-05,
+      "loss": 0.1945,
+      "step": 12505
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7132395133897506e-05,
+      "loss": 0.1946,
+      "step": 12506
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7120559632844245e-05,
+      "loss": 0.1881,
+      "step": 12507
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.710872783860883e-05,
+      "loss": 0.1876,
+      "step": 12508
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7096899751720385e-05,
+      "loss": 0.1915,
+      "step": 12509
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.708507537270797e-05,
+      "loss": 0.2073,
+      "step": 12510
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7073254702100405e-05,
+      "loss": 0.201,
+      "step": 12511
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.70614377404264e-05,
+      "loss": 0.1891,
+      "step": 12512
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.704962448821443e-05,
+      "loss": 0.1906,
+      "step": 12513
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7037814945992914e-05,
+      "loss": 0.215,
+      "step": 12514
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7026009114289977e-05,
+      "loss": 0.1973,
+      "step": 12515
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7014206993633686e-05,
+      "loss": 0.2004,
+      "step": 12516
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7002408584551854e-05,
+      "loss": 0.1901,
+      "step": 12517
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6990613887572217e-05,
+      "loss": 0.2,
+      "step": 12518
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6978822903222234e-05,
+      "loss": 0.2027,
+      "step": 12519
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.696703563202934e-05,
+      "loss": 0.1865,
+      "step": 12520
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6955252074520643e-05,
+      "loss": 0.1993,
+      "step": 12521
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6943472231223245e-05,
+      "loss": 0.1811,
+      "step": 12522
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6931696102663942e-05,
+      "loss": 0.191,
+      "step": 12523
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6919923689369477e-05,
+      "loss": 0.1905,
+      "step": 12524
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.690815499186633e-05,
+      "loss": 0.186,
+      "step": 12525
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.689639001068093e-05,
+      "loss": 0.1893,
+      "step": 12526
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6884628746339382e-05,
+      "loss": 0.1883,
+      "step": 12527
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6872871199367767e-05,
+      "loss": 0.2026,
+      "step": 12528
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6861117370291922e-05,
+      "loss": 0.1874,
+      "step": 12529
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6849367259637572e-05,
+      "loss": 0.1877,
+      "step": 12530
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6837620867930203e-05,
+      "loss": 0.196,
+      "step": 12531
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6825878195695234e-05,
+      "loss": 0.1938,
+      "step": 12532
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6814139243457803e-05,
+      "loss": 0.1883,
+      "step": 12533
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6802404011742988e-05,
+      "loss": 0.2059,
+      "step": 12534
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.67906725010756e-05,
+      "loss": 0.2007,
+      "step": 12535
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6778944711980405e-05,
+      "loss": 0.1943,
+      "step": 12536
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6767220644981853e-05,
+      "loss": 0.1916,
+      "step": 12537
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.675550030060439e-05,
+      "loss": 0.1929,
+      "step": 12538
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6743783679372137e-05,
+      "loss": 0.1853,
+      "step": 12539
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6732070781809184e-05,
+      "loss": 0.1799,
+      "step": 12540
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.672036160843935e-05,
+      "loss": 0.184,
+      "step": 12541
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6708656159786383e-05,
+      "loss": 0.1861,
+      "step": 12542
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.669695443637378e-05,
+      "loss": 0.1985,
+      "step": 12543
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6685256438724893e-05,
+      "loss": 0.1997,
+      "step": 12544
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6673562167362955e-05,
+      "loss": 0.1909,
+      "step": 12545
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6661871622810988e-05,
+      "loss": 0.1806,
+      "step": 12546
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6650184805591818e-05,
+      "loss": 0.1987,
+      "step": 12547
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6638501716228193e-05,
+      "loss": 0.1898,
+      "step": 12548
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.662682235524261e-05,
+      "loss": 0.1925,
+      "step": 12549
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.661514672315746e-05,
+      "loss": 0.1912,
+      "step": 12550
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6603474820494914e-05,
+      "loss": 0.2084,
+      "step": 12551
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.659180664777704e-05,
+      "loss": 0.1882,
+      "step": 12552
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6580142205525652e-05,
+      "loss": 0.1986,
+      "step": 12553
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.65684814942625e-05,
+      "loss": 0.1969,
+      "step": 12554
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6556824514509082e-05,
+      "loss": 0.1894,
+      "step": 12555
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.654517126678675e-05,
+      "loss": 0.1929,
+      "step": 12556
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6533521751616743e-05,
+      "loss": 0.1955,
+      "step": 12557
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.652187596952004e-05,
+      "loss": 0.198,
+      "step": 12558
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.651023392101755e-05,
+      "loss": 0.2086,
+      "step": 12559
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.649859560662993e-05,
+      "loss": 0.2012,
+      "step": 12560
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6486961026877734e-05,
+      "loss": 0.1992,
+      "step": 12561
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.647533018228131e-05,
+      "loss": 0.1962,
+      "step": 12562
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.646370307336087e-05,
+      "loss": 0.1899,
+      "step": 12563
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6452079700636415e-05,
+      "loss": 0.1918,
+      "step": 12564
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6440460064627838e-05,
+      "loss": 0.1897,
+      "step": 12565
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6428844165854805e-05,
+      "loss": 0.2011,
+      "step": 12566
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6417232004836836e-05,
+      "loss": 0.1959,
+      "step": 12567
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6405623582093334e-05,
+      "loss": 0.198,
+      "step": 12568
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.639401889814345e-05,
+      "loss": 0.1936,
+      "step": 12569
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6382417953506202e-05,
+      "loss": 0.1963,
+      "step": 12570
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6370820748700476e-05,
+      "loss": 0.1963,
+      "step": 12571
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.635922728424494e-05,
+      "loss": 0.1978,
+      "step": 12572
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6347637560658145e-05,
+      "loss": 0.19,
+      "step": 12573
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.63360515784584e-05,
+      "loss": 0.2056,
+      "step": 12574
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.632446933816395e-05,
+      "loss": 0.1859,
+      "step": 12575
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.631289084029276e-05,
+      "loss": 0.1977,
+      "step": 12576
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.630131608536274e-05,
+      "loss": 0.1995,
+      "step": 12577
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6289745073891516e-05,
+      "loss": 0.1861,
+      "step": 12578
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6278177806396667e-05,
+      "loss": 0.1957,
+      "step": 12579
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6266614283395475e-05,
+      "loss": 0.1986,
+      "step": 12580
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6255054505405197e-05,
+      "loss": 0.1859,
+      "step": 12581
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6243498472942787e-05,
+      "loss": 0.1909,
+      "step": 12582
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6231946186525138e-05,
+      "loss": 0.1884,
+      "step": 12583
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6220397646668896e-05,
+      "loss": 0.1874,
+      "step": 12584
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.620885285389061e-05,
+      "loss": 0.1909,
+      "step": 12585
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6197311808706605e-05,
+      "loss": 0.1851,
+      "step": 12586
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6185774511633056e-05,
+      "loss": 0.1943,
+      "step": 12587
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.617424096318596e-05,
+      "loss": 0.196,
+      "step": 12588
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6162711163881204e-05,
+      "loss": 0.1964,
+      "step": 12589
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6151185114234412e-05,
+      "loss": 0.1938,
+      "step": 12590
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6139662814761138e-05,
+      "loss": 0.1831,
+      "step": 12591
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6128144265976676e-05,
+      "loss": 0.1861,
+      "step": 12592
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.611662946839625e-05,
+      "loss": 0.1905,
+      "step": 12593
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6105118422534805e-05,
+      "loss": 0.1939,
+      "step": 12594
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.609361112890724e-05,
+      "loss": 0.1908,
+      "step": 12595
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.608210758802816e-05,
+      "loss": 0.1865,
+      "step": 12596
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.607060780041212e-05,
+      "loss": 0.2008,
+      "step": 12597
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.605911176657341e-05,
+      "loss": 0.1887,
+      "step": 12598
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6047619487026246e-05,
+      "loss": 0.1879,
+      "step": 12599
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.603613096228457e-05,
+      "loss": 0.1939,
+      "step": 12600
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6024646192862248e-05,
+      "loss": 0.1826,
+      "step": 12601
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6013165179272916e-05,
+      "loss": 0.1769,
+      "step": 12602
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6001687922030096e-05,
+      "loss": 0.1983,
+      "step": 12603
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5990214421647075e-05,
+      "loss": 0.1949,
+      "step": 12604
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5978744678637058e-05,
+      "loss": 0.193,
+      "step": 12605
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.596727869351301e-05,
+      "loss": 0.1985,
+      "step": 12606
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5955816466787742e-05,
+      "loss": 0.1849,
+      "step": 12607
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5944357998973902e-05,
+      "loss": 0.1871,
+      "step": 12608
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5932903290584e-05,
+      "loss": 0.1946,
+      "step": 12609
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5921452342130328e-05,
+      "loss": 0.1809,
+      "step": 12610
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5910005154125064e-05,
+      "loss": 0.1918,
+      "step": 12611
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5898561727080153e-05,
+      "loss": 0.1984,
+      "step": 12612
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5887122061507442e-05,
+      "loss": 0.1941,
+      "step": 12613
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5875686157918545e-05,
+      "loss": 0.1984,
+      "step": 12614
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5864254016824965e-05,
+      "loss": 0.187,
+      "step": 12615
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5852825638737978e-05,
+      "loss": 0.2007,
+      "step": 12616
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5841401024168757e-05,
+      "loss": 0.1965,
+      "step": 12617
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5829980173628233e-05,
+      "loss": 0.2027,
+      "step": 12618
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.581856308762726e-05,
+      "loss": 0.1853,
+      "step": 12619
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5807149766676423e-05,
+      "loss": 0.1932,
+      "step": 12620
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5795740211286214e-05,
+      "loss": 0.1864,
+      "step": 12621
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.578433442196694e-05,
+      "loss": 0.1916,
+      "step": 12622
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5772932399228692e-05,
+      "loss": 0.1888,
+      "step": 12623
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5761534143581457e-05,
+      "loss": 0.1892,
+      "step": 12624
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5750139655535014e-05,
+      "loss": 0.1776,
+      "step": 12625
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5738748935599014e-05,
+      "loss": 0.1913,
+      "step": 12626
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5727361984282894e-05,
+      "loss": 0.1972,
+      "step": 12627
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.571597880209591e-05,
+      "loss": 0.1851,
+      "step": 12628
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5704599389547236e-05,
+      "loss": 0.1887,
+      "step": 12629
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5693223747145767e-05,
+      "loss": 0.1888,
+      "step": 12630
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5681851875400334e-05,
+      "loss": 0.1809,
+      "step": 12631
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5670483774819487e-05,
+      "loss": 0.1992,
+      "step": 12632
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.565911944591174e-05,
+      "loss": 0.1996,
+      "step": 12633
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5647758889185337e-05,
+      "loss": 0.1861,
+      "step": 12634
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.563640210514835e-05,
+      "loss": 0.2033,
+      "step": 12635
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5625049094308775e-05,
+      "loss": 0.1973,
+      "step": 12636
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5613699857174324e-05,
+      "loss": 0.1931,
+      "step": 12637
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.560235439425264e-05,
+      "loss": 0.2064,
+      "step": 12638
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5591012706051123e-05,
+      "loss": 0.1856,
+      "step": 12639
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5579674793077072e-05,
+      "loss": 0.1933,
+      "step": 12640
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.556834065583752e-05,
+      "loss": 0.1967,
+      "step": 12641
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5557010294839468e-05,
+      "loss": 0.1915,
+      "step": 12642
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5545683710589596e-05,
+      "loss": 0.208,
+      "step": 12643
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5534360903594557e-05,
+      "loss": 0.1871,
+      "step": 12644
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.552304187436071e-05,
+      "loss": 0.1978,
+      "step": 12645
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5511726623394372e-05,
+      "loss": 0.1875,
+      "step": 12646
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5500415151201543e-05,
+      "loss": 0.19,
+      "step": 12647
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5489107458288175e-05,
+      "loss": 0.1861,
+      "step": 12648
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5477803545159998e-05,
+      "loss": 0.1984,
+      "step": 12649
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5466503412322608e-05,
+      "loss": 0.1901,
+      "step": 12650
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5455207060281373e-05,
+      "loss": 0.1833,
+      "step": 12651
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5443914489541568e-05,
+      "loss": 0.2057,
+      "step": 12652
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5432625700608205e-05,
+      "loss": 0.1896,
+      "step": 12653
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.542134069398624e-05,
+      "loss": 0.194,
+      "step": 12654
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.541005947018035e-05,
+      "loss": 0.1816,
+      "step": 12655
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5398782029695125e-05,
+      "loss": 0.1877,
+      "step": 12656
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.538750837303492e-05,
+      "loss": 0.1907,
+      "step": 12657
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5376238500704e-05,
+      "loss": 0.199,
+      "step": 12658
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5364972413206368e-05,
+      "loss": 0.1917,
+      "step": 12659
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.535371011104595e-05,
+      "loss": 0.2036,
+      "step": 12660
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5342451594726403e-05,
+      "loss": 0.1866,
+      "step": 12661
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.533119686475133e-05,
+      "loss": 0.1989,
+      "step": 12662
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5319945921624046e-05,
+      "loss": 0.1942,
+      "step": 12663
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5308698765847796e-05,
+      "loss": 0.2004,
+      "step": 12664
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5297455397925586e-05,
+      "loss": 0.1939,
+      "step": 12665
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5286215818360327e-05,
+      "loss": 0.1939,
+      "step": 12666
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.527498002765465e-05,
+      "loss": 0.1828,
+      "step": 12667
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.526374802631112e-05,
+      "loss": 0.1851,
+      "step": 12668
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5252519814832067e-05,
+      "loss": 0.1916,
+      "step": 12669
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5241295393719712e-05,
+      "loss": 0.1789,
+      "step": 12670
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5230074763476033e-05,
+      "loss": 0.193,
+      "step": 12671
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5218857924602925e-05,
+      "loss": 0.1861,
+      "step": 12672
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5207644877602e-05,
+      "loss": 0.185,
+      "step": 12673
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5196435622974847e-05,
+      "loss": 0.1974,
+      "step": 12674
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.518523016122273e-05,
+      "loss": 0.1817,
+      "step": 12675
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.517402849284687e-05,
+      "loss": 0.1925,
+      "step": 12676
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5162830618348235e-05,
+      "loss": 0.1925,
+      "step": 12677
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5151636538227687e-05,
+      "loss": 0.1922,
+      "step": 12678
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.514044625298583e-05,
+      "loss": 0.1907,
+      "step": 12679
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5129259763123226e-05,
+      "loss": 0.1943,
+      "step": 12680
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5118077069140136e-05,
+      "loss": 0.1879,
+      "step": 12681
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5106898171536754e-05,
+      "loss": 0.1897,
+      "step": 12682
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5095723070813016e-05,
+      "loss": 0.1864,
+      "step": 12683
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.508455176746878e-05,
+      "loss": 0.1899,
+      "step": 12684
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5073384262003643e-05,
+      "loss": 0.1884,
+      "step": 12685
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.506222055491714e-05,
+      "loss": 0.1738,
+      "step": 12686
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5051060646708493e-05,
+      "loss": 0.1994,
+      "step": 12687
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5039904537876891e-05,
+      "loss": 0.1888,
+      "step": 12688
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.502875222892125e-05,
+      "loss": 0.1921,
+      "step": 12689
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5017603720340423e-05,
+      "loss": 0.1854,
+      "step": 12690
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5006459012632957e-05,
+      "loss": 0.1825,
+      "step": 12691
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4995318106297374e-05,
+      "loss": 0.1943,
+      "step": 12692
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.498418100183191e-05,
+      "loss": 0.1919,
+      "step": 12693
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.49730476997347e-05,
+      "loss": 0.1957,
+      "step": 12694
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4961918200503667e-05,
+      "loss": 0.1842,
+      "step": 12695
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4950792504636613e-05,
+      "loss": 0.1901,
+      "step": 12696
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4939670612631085e-05,
+      "loss": 0.1785,
+      "step": 12697
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4928552524984585e-05,
+      "loss": 0.1943,
+      "step": 12698
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4917438242194337e-05,
+      "loss": 0.1877,
+      "step": 12699
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.49063277647574e-05,
+      "loss": 0.1773,
+      "step": 12700
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4895221093170763e-05,
+      "loss": 0.1863,
+      "step": 12701
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4884118227931109e-05,
+      "loss": 0.1838,
+      "step": 12702
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4873019169535074e-05,
+      "loss": 0.1918,
+      "step": 12703
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4861923918479026e-05,
+      "loss": 0.1896,
+      "step": 12704
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4850832475259246e-05,
+      "loss": 0.1829,
+      "step": 12705
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4839744840371772e-05,
+      "loss": 0.1881,
+      "step": 12706
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4828661014312494e-05,
+      "loss": 0.1854,
+      "step": 12707
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4817580997577186e-05,
+      "loss": 0.1933,
+      "step": 12708
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4806504790661357e-05,
+      "loss": 0.1891,
+      "step": 12709
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4795432394060439e-05,
+      "loss": 0.1836,
+      "step": 12710
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4784363808269618e-05,
+      "loss": 0.1955,
+      "step": 12711
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4773299033783938e-05,
+      "loss": 0.1835,
+      "step": 12712
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.47622380710983e-05,
+      "loss": 0.1973,
+      "step": 12713
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4751180920707386e-05,
+      "loss": 0.1952,
+      "step": 12714
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4740127583105767e-05,
+      "loss": 0.1748,
+      "step": 12715
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.472907805878776e-05,
+      "loss": 0.1858,
+      "step": 12716
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4718032348247613e-05,
+      "loss": 0.1813,
+      "step": 12717
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4706990451979285e-05,
+      "loss": 0.196,
+      "step": 12718
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4695952370476707e-05,
+      "loss": 0.1837,
+      "step": 12719
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.468491810423348e-05,
+      "loss": 0.1926,
+      "step": 12720
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4673887653743201e-05,
+      "loss": 0.1909,
+      "step": 12721
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.466286101949913e-05,
+      "loss": 0.1856,
+      "step": 12722
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4651838201994505e-05,
+      "loss": 0.1849,
+      "step": 12723
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4640819201722278e-05,
+      "loss": 0.1933,
+      "step": 12724
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4629804019175309e-05,
+      "loss": 0.1921,
+      "step": 12725
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4618792654846258e-05,
+      "loss": 0.1943,
+      "step": 12726
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.460778510922759e-05,
+      "loss": 0.1853,
+      "step": 12727
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.459678138281162e-05,
+      "loss": 0.1878,
+      "step": 12728
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4585781476090521e-05,
+      "loss": 0.1948,
+      "step": 12729
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4574785389556233e-05,
+      "loss": 0.1896,
+      "step": 12730
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4563793123700609e-05,
+      "loss": 0.1903,
+      "step": 12731
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4552804679015219e-05,
+      "loss": 0.1943,
+      "step": 12732
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4541820055991595e-05,
+      "loss": 0.1963,
+      "step": 12733
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4530839255120975e-05,
+      "loss": 0.1903,
+      "step": 12734
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4519862276894524e-05,
+      "loss": 0.2004,
+      "step": 12735
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4508889121803149e-05,
+      "loss": 0.1967,
+      "step": 12736
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4497919790337667e-05,
+      "loss": 0.188,
+      "step": 12737
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.448695428298864e-05,
+      "loss": 0.19,
+      "step": 12738
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4475992600246557e-05,
+      "loss": 0.1968,
+      "step": 12739
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4465034742601646e-05,
+      "loss": 0.1923,
+      "step": 12740
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4454080710544026e-05,
+      "loss": 0.1982,
+      "step": 12741
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4443130504563596e-05,
+      "loss": 0.1814,
+      "step": 12742
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.443218412515014e-05,
+      "loss": 0.187,
+      "step": 12743
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4421241572793209e-05,
+      "loss": 0.1932,
+      "step": 12744
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4410302847982238e-05,
+      "loss": 0.1808,
+      "step": 12745
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4399367951206466e-05,
+      "loss": 0.1737,
+      "step": 12746
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4388436882954936e-05,
+      "loss": 0.2029,
+      "step": 12747
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4377509643716546e-05,
+      "loss": 0.1791,
+      "step": 12748
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.436658623398005e-05,
+      "loss": 0.1971,
+      "step": 12749
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4355666654233968e-05,
+      "loss": 0.1941,
+      "step": 12750
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.434475090496672e-05,
+      "loss": 0.1839,
+      "step": 12751
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.433383898666647e-05,
+      "loss": 0.1828,
+      "step": 12752
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4322930899821308e-05,
+      "loss": 0.191,
+      "step": 12753
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4312026644919064e-05,
+      "loss": 0.1922,
+      "step": 12754
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.430112622244747e-05,
+      "loss": 0.199,
+      "step": 12755
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4290229632894014e-05,
+      "loss": 0.1861,
+      "step": 12756
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4279336876746096e-05,
+      "loss": 0.1924,
+      "step": 12757
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.426844795449085e-05,
+      "loss": 0.1892,
+      "step": 12758
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4257562866615338e-05,
+      "loss": 0.1849,
+      "step": 12759
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4246681613606339e-05,
+      "loss": 0.1838,
+      "step": 12760
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4235804195950585e-05,
+      "loss": 0.1922,
+      "step": 12761
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4224930614134523e-05,
+      "loss": 0.1875,
+      "step": 12762
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4214060868644519e-05,
+      "loss": 0.196,
+      "step": 12763
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4203194959966703e-05,
+      "loss": 0.1871,
+      "step": 12764
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4192332888587046e-05,
+      "loss": 0.1944,
+      "step": 12765
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.41814746549914e-05,
+      "loss": 0.1937,
+      "step": 12766
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.417062025966538e-05,
+      "loss": 0.2015,
+      "step": 12767
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4159769703094417e-05,
+      "loss": 0.1841,
+      "step": 12768
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4148922985763869e-05,
+      "loss": 0.1891,
+      "step": 12769
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4138080108158814e-05,
+      "loss": 0.1912,
+      "step": 12770
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.412724107076424e-05,
+      "loss": 0.1946,
+      "step": 12771
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4116405874064897e-05,
+      "loss": 0.1951,
+      "step": 12772
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4105574518545428e-05,
+      "loss": 0.186,
+      "step": 12773
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4094747004690222e-05,
+      "loss": 0.1823,
+      "step": 12774
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4083923332983595e-05,
+      "loss": 0.1788,
+      "step": 12775
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4073103503909591e-05,
+      "loss": 0.1928,
+      "step": 12776
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4062287517952177e-05,
+      "loss": 0.2011,
+      "step": 12777
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.405147537559509e-05,
+      "loss": 0.1802,
+      "step": 12778
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.404066707732188e-05,
+      "loss": 0.1914,
+      "step": 12779
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4029862623615985e-05,
+      "loss": 0.1818,
+      "step": 12780
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4019062014960615e-05,
+      "loss": 0.1916,
+      "step": 12781
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4008265251838859e-05,
+      "loss": 0.1925,
+      "step": 12782
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3997472334733563e-05,
+      "loss": 0.1876,
+      "step": 12783
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3986683264127509e-05,
+      "loss": 0.1888,
+      "step": 12784
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3975898040503177e-05,
+      "loss": 0.1822,
+      "step": 12785
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3965116664342992e-05,
+      "loss": 0.1894,
+      "step": 12786
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3954339136129124e-05,
+      "loss": 0.1865,
+      "step": 12787
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3943565456343599e-05,
+      "loss": 0.186,
+      "step": 12788
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.393279562546832e-05,
+      "loss": 0.1903,
+      "step": 12789
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3922029643984924e-05,
+      "loss": 0.1875,
+      "step": 12790
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3911267512374926e-05,
+      "loss": 0.1848,
+      "step": 12791
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3900509231119695e-05,
+      "loss": 0.1859,
+      "step": 12792
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3889754800700372e-05,
+      "loss": 0.1809,
+      "step": 12793
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.387900422159798e-05,
+      "loss": 0.1806,
+      "step": 12794
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3868257494293313e-05,
+      "loss": 0.1869,
+      "step": 12795
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3857514619267064e-05,
+      "loss": 0.1845,
+      "step": 12796
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.384677559699966e-05,
+      "loss": 0.1841,
+      "step": 12797
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3836040427971474e-05,
+      "loss": 0.1814,
+      "step": 12798
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3825309112662566e-05,
+      "loss": 0.1985,
+      "step": 12799
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.381458165155296e-05,
+      "loss": 0.1897,
+      "step": 12800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.380385804512241e-05,
+      "loss": 0.1974,
+      "step": 12801
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.379313829385056e-05,
+      "loss": 0.1938,
+      "step": 12802
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3782422398216821e-05,
+      "loss": 0.187,
+      "step": 12803
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3771710358700508e-05,
+      "loss": 0.1849,
+      "step": 12804
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.376100217578068e-05,
+      "loss": 0.1935,
+      "step": 12805
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3750297849936322e-05,
+      "loss": 0.1912,
+      "step": 12806
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3739597381646119e-05,
+      "loss": 0.1845,
+      "step": 12807
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3728900771388697e-05,
+      "loss": 0.1862,
+      "step": 12808
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3718208019642442e-05,
+      "loss": 0.1909,
+      "step": 12809
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3707519126885626e-05,
+      "loss": 0.1916,
+      "step": 12810
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3696834093596279e-05,
+      "loss": 0.1922,
+      "step": 12811
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3686152920252327e-05,
+      "loss": 0.1936,
+      "step": 12812
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3675475607331456e-05,
+      "loss": 0.1957,
+      "step": 12813
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.366480215531124e-05,
+      "loss": 0.1847,
+      "step": 12814
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3654132564669032e-05,
+      "loss": 0.1941,
+      "step": 12815
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3643466835882068e-05,
+      "loss": 0.196,
+      "step": 12816
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3632804969427326e-05,
+      "loss": 0.1843,
+      "step": 12817
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3622146965781712e-05,
+      "loss": 0.184,
+      "step": 12818
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3611492825421867e-05,
+      "loss": 0.1838,
+      "step": 12819
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3600842548824355e-05,
+      "loss": 0.1871,
+      "step": 12820
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3590196136465449e-05,
+      "loss": 0.1879,
+      "step": 12821
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3579553588821382e-05,
+      "loss": 0.1781,
+      "step": 12822
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.356891490636808e-05,
+      "loss": 0.1901,
+      "step": 12823
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.355828008958142e-05,
+      "loss": 0.191,
+      "step": 12824
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3547649138937001e-05,
+      "loss": 0.1858,
+      "step": 12825
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3537022054910364e-05,
+      "loss": 0.1812,
+      "step": 12826
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3526398837976718e-05,
+      "loss": 0.1825,
+      "step": 12827
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3515779488611268e-05,
+      "loss": 0.1926,
+      "step": 12828
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3505164007288907e-05,
+      "loss": 0.1969,
+      "step": 12829
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3494552394484472e-05,
+      "loss": 0.1976,
+      "step": 12830
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3483944650672531e-05,
+      "loss": 0.2045,
+      "step": 12831
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.347334077632756e-05,
+      "loss": 0.1859,
+      "step": 12832
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3462740771923776e-05,
+      "loss": 0.179,
+      "step": 12833
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3452144637935315e-05,
+      "loss": 0.1709,
+      "step": 12834
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3441552374836042e-05,
+      "loss": 0.1774,
+      "step": 12835
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3430963983099754e-05,
+      "loss": 0.189,
+      "step": 12836
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3420379463199973e-05,
+      "loss": 0.194,
+      "step": 12837
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3409798815610142e-05,
+      "loss": 0.1902,
+      "step": 12838
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3399222040803438e-05,
+      "loss": 0.1969,
+      "step": 12839
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3388649139252962e-05,
+      "loss": 0.194,
+      "step": 12840
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3378080111431545e-05,
+      "loss": 0.1844,
+      "step": 12841
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3367514957811921e-05,
+      "loss": 0.1948,
+      "step": 12842
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3356953678866623e-05,
+      "loss": 0.201,
+      "step": 12843
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3346396275067974e-05,
+      "loss": 0.1867,
+      "step": 12844
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3335842746888195e-05,
+      "loss": 0.1873,
+      "step": 12845
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3325293094799296e-05,
+      "loss": 0.1885,
+      "step": 12846
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3314747319273069e-05,
+      "loss": 0.1939,
+      "step": 12847
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3304205420781236e-05,
+      "loss": 0.1837,
+      "step": 12848
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.329366739979524e-05,
+      "loss": 0.1837,
+      "step": 12849
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3283133256786451e-05,
+      "loss": 0.1854,
+      "step": 12850
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3272602992225946e-05,
+      "loss": 0.1902,
+      "step": 12851
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3262076606584772e-05,
+      "loss": 0.1831,
+      "step": 12852
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3251554100333652e-05,
+      "loss": 0.1935,
+      "step": 12853
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3241035473943276e-05,
+      "loss": 0.1889,
+      "step": 12854
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3230520727884054e-05,
+      "loss": 0.1909,
+      "step": 12855
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3220009862626248e-05,
+      "loss": 0.1806,
+      "step": 12856
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3209502878640012e-05,
+      "loss": 0.1915,
+      "step": 12857
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3198999776395227e-05,
+      "loss": 0.1955,
+      "step": 12858
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3188500556361683e-05,
+      "loss": 0.1914,
+      "step": 12859
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3178005219008927e-05,
+      "loss": 0.1728,
+      "step": 12860
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3167513764806417e-05,
+      "loss": 0.1843,
+      "step": 12861
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.315702619422332e-05,
+      "loss": 0.1793,
+      "step": 12862
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3146542507728776e-05,
+      "loss": 0.1958,
+      "step": 12863
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3136062705791596e-05,
+      "loss": 0.1973,
+      "step": 12864
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3125586788880561e-05,
+      "loss": 0.1883,
+      "step": 12865
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3115114757464176e-05,
+      "loss": 0.1898,
+      "step": 12866
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3104646612010817e-05,
+      "loss": 0.1852,
+      "step": 12867
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3094182352988638e-05,
+      "loss": 0.1782,
+      "step": 12868
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3083721980865715e-05,
+      "loss": 0.1788,
+      "step": 12869
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3073265496109843e-05,
+      "loss": 0.1717,
+      "step": 12870
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3062812899188736e-05,
+      "loss": 0.1874,
+      "step": 12871
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3052364190569854e-05,
+      "loss": 0.1794,
+      "step": 12872
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3041919370720556e-05,
+      "loss": 0.1894,
+      "step": 12873
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3031478440107959e-05,
+      "loss": 0.1885,
+      "step": 12874
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3021041399199062e-05,
+      "loss": 0.1846,
+      "step": 12875
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3010608248460632e-05,
+      "loss": 0.1941,
+      "step": 12876
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3000178988359345e-05,
+      "loss": 0.1903,
+      "step": 12877
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.29897536193616e-05,
+      "loss": 0.1809,
+      "step": 12878
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2979332141933732e-05,
+      "loss": 0.1954,
+      "step": 12879
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.29689145565418e-05,
+      "loss": 0.1868,
+      "step": 12880
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2958500863651767e-05,
+      "loss": 0.196,
+      "step": 12881
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.294809106372935e-05,
+      "loss": 0.1902,
+      "step": 12882
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2937685157240186e-05,
+      "loss": 0.183,
+      "step": 12883
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2927283144649638e-05,
+      "loss": 0.1919,
+      "step": 12884
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2916885026422965e-05,
+      "loss": 0.1866,
+      "step": 12885
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2906490803025229e-05,
+      "loss": 0.1889,
+      "step": 12886
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2896100474921313e-05,
+      "loss": 0.1878,
+      "step": 12887
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2885714042575892e-05,
+      "loss": 0.1857,
+      "step": 12888
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2875331506453547e-05,
+      "loss": 0.1845,
+      "step": 12889
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.286495286701862e-05,
+      "loss": 0.1844,
+      "step": 12890
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2854578124735317e-05,
+      "loss": 0.188,
+      "step": 12891
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2844207280067633e-05,
+      "loss": 0.2001,
+      "step": 12892
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2833840333479441e-05,
+      "loss": 0.1763,
+      "step": 12893
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2823477285434359e-05,
+      "loss": 0.1848,
+      "step": 12894
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.281311813639593e-05,
+      "loss": 0.1785,
+      "step": 12895
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2802762886827412e-05,
+      "loss": 0.1895,
+      "step": 12896
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2792411537192017e-05,
+      "loss": 0.1855,
+      "step": 12897
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2782064087952661e-05,
+      "loss": 0.1843,
+      "step": 12898
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2771720539572162e-05,
+      "loss": 0.1823,
+      "step": 12899
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2761380892513119e-05,
+      "loss": 0.1863,
+      "step": 12900
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2751045147238016e-05,
+      "loss": 0.1905,
+      "step": 12901
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2740713304209073e-05,
+      "loss": 0.1813,
+      "step": 12902
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.273038536388843e-05,
+      "loss": 0.1967,
+      "step": 12903
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2720061326737976e-05,
+      "loss": 0.1796,
+      "step": 12904
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2709741193219493e-05,
+      "loss": 0.1808,
+      "step": 12905
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2699424963794526e-05,
+      "loss": 0.1948,
+      "step": 12906
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2689112638924484e-05,
+      "loss": 0.1819,
+      "step": 12907
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2678804219070561e-05,
+      "loss": 0.1872,
+      "step": 12908
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2668499704693848e-05,
+      "loss": 0.1847,
+      "step": 12909
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2658199096255185e-05,
+      "loss": 0.1891,
+      "step": 12910
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2647902394215305e-05,
+      "loss": 0.1823,
+      "step": 12911
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2637609599034694e-05,
+      "loss": 0.1933,
+      "step": 12912
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2627320711173752e-05,
+      "loss": 0.1871,
+      "step": 12913
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2617035731092586e-05,
+      "loss": 0.1775,
+      "step": 12914
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2606754659251262e-05,
+      "loss": 0.1951,
+      "step": 12915
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.259647749610955e-05,
+      "loss": 0.1824,
+      "step": 12916
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2586204242127154e-05,
+      "loss": 0.1832,
+      "step": 12917
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2575934897763498e-05,
+      "loss": 0.1903,
+      "step": 12918
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2565669463477925e-05,
+      "loss": 0.1953,
+      "step": 12919
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2555407939729547e-05,
+      "loss": 0.1755,
+      "step": 12920
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2545150326977296e-05,
+      "loss": 0.1892,
+      "step": 12921
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.253489662567997e-05,
+      "loss": 0.1754,
+      "step": 12922
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.252464683629615e-05,
+      "loss": 0.1871,
+      "step": 12923
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.25144009592843e-05,
+      "loss": 0.1796,
+      "step": 12924
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2504158995102622e-05,
+      "loss": 0.1928,
+      "step": 12925
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2493920944209236e-05,
+      "loss": 0.1754,
+      "step": 12926
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2483686807062034e-05,
+      "loss": 0.1782,
+      "step": 12927
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2473456584118692e-05,
+      "loss": 0.1882,
+      "step": 12928
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2463230275836835e-05,
+      "loss": 0.1868,
+      "step": 12929
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2453007882673785e-05,
+      "loss": 0.1854,
+      "step": 12930
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2442789405086774e-05,
+      "loss": 0.1842,
+      "step": 12931
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2432574843532806e-05,
+      "loss": 0.1927,
+      "step": 12932
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.242236419846876e-05,
+      "loss": 0.1872,
+      "step": 12933
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2412157470351294e-05,
+      "loss": 0.1833,
+      "step": 12934
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2401954659636872e-05,
+      "loss": 0.1917,
+      "step": 12935
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2391755766781887e-05,
+      "loss": 0.1915,
+      "step": 12936
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2381560792242442e-05,
+      "loss": 0.1925,
+      "step": 12937
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2371369736474536e-05,
+      "loss": 0.1889,
+      "step": 12938
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.236118259993394e-05,
+      "loss": 0.1829,
+      "step": 12939
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2350999383076322e-05,
+      "loss": 0.1789,
+      "step": 12940
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2340820086357074e-05,
+      "loss": 0.1767,
+      "step": 12941
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2330644710231531e-05,
+      "loss": 0.1876,
+      "step": 12942
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2320473255154741e-05,
+      "loss": 0.1833,
+      "step": 12943
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.231030572158166e-05,
+      "loss": 0.1928,
+      "step": 12944
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2300142109966995e-05,
+      "loss": 0.1831,
+      "step": 12945
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2289982420765389e-05,
+      "loss": 0.1864,
+      "step": 12946
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2279826654431158e-05,
+      "loss": 0.1984,
+      "step": 12947
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2269674811418574e-05,
+      "loss": 0.1872,
+      "step": 12948
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.225952689218164e-05,
+      "loss": 0.1822,
+      "step": 12949
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2249382897174278e-05,
+      "loss": 0.1889,
+      "step": 12950
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2239242826850139e-05,
+      "loss": 0.1933,
+      "step": 12951
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2229106681662772e-05,
+      "loss": 0.1869,
+      "step": 12952
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2218974462065492e-05,
+      "loss": 0.1942,
+      "step": 12953
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2208846168511513e-05,
+      "loss": 0.1749,
+      "step": 12954
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.219872180145376e-05,
+      "loss": 0.186,
+      "step": 12955
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.218860136134512e-05,
+      "loss": 0.1884,
+      "step": 12956
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2178484848638171e-05,
+      "loss": 0.1874,
+      "step": 12957
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2168372263785432e-05,
+      "loss": 0.1875,
+      "step": 12958
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.215826360723915e-05,
+      "loss": 0.1885,
+      "step": 12959
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2148158879451477e-05,
+      "loss": 0.1843,
+      "step": 12960
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2138058080874304e-05,
+      "loss": 0.18,
+      "step": 12961
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.212796121195945e-05,
+      "loss": 0.1893,
+      "step": 12962
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2117868273158451e-05,
+      "loss": 0.1832,
+      "step": 12963
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2107779264922758e-05,
+      "loss": 0.1886,
+      "step": 12964
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2097694187703568e-05,
+      "loss": 0.1764,
+      "step": 12965
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2087613041951995e-05,
+      "loss": 0.1846,
+      "step": 12966
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2077535828118847e-05,
+      "loss": 0.1908,
+      "step": 12967
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2067462546654907e-05,
+      "loss": 0.1987,
+      "step": 12968
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2057393198010636e-05,
+      "loss": 0.1927,
+      "step": 12969
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2047327782636452e-05,
+      "loss": 0.1969,
+      "step": 12970
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2037266300982486e-05,
+      "loss": 0.178,
+      "step": 12971
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2027208753498786e-05,
+      "loss": 0.1877,
+      "step": 12972
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.201715514063514e-05,
+      "loss": 0.1874,
+      "step": 12973
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.200710546284124e-05,
+      "loss": 0.1904,
+      "step": 12974
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1997059720566517e-05,
+      "loss": 0.1755,
+      "step": 12975
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1987017914260323e-05,
+      "loss": 0.1999,
+      "step": 12976
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1976980044371722e-05,
+      "loss": 0.1811,
+      "step": 12977
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.196694611134972e-05,
+      "loss": 0.197,
+      "step": 12978
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1956916115643046e-05,
+      "loss": 0.1877,
+      "step": 12979
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1946890057700321e-05,
+      "loss": 0.182,
+      "step": 12980
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1936867937969953e-05,
+      "loss": 0.1923,
+      "step": 12981
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1926849756900193e-05,
+      "loss": 0.1889,
+      "step": 12982
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1916835514939096e-05,
+      "loss": 0.1875,
+      "step": 12983
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1906825212534577e-05,
+      "loss": 0.1818,
+      "step": 12984
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1896818850134306e-05,
+      "loss": 0.1804,
+      "step": 12985
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.188681642818591e-05,
+      "loss": 0.1827,
+      "step": 12986
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1876817947136632e-05,
+      "loss": 0.19,
+      "step": 12987
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1866823407433746e-05,
+      "loss": 0.1791,
+      "step": 12988
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1856832809524222e-05,
+      "loss": 0.1834,
+      "step": 12989
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1846846153854918e-05,
+      "loss": 0.1818,
+      "step": 12990
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.183686344087247e-05,
+      "loss": 0.1842,
+      "step": 12991
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1826884671023375e-05,
+      "loss": 0.1848,
+      "step": 12992
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.181690984475392e-05,
+      "loss": 0.1907,
+      "step": 12993
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1806938962510261e-05,
+      "loss": 0.178,
+      "step": 12994
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1796972024738318e-05,
+      "loss": 0.186,
+      "step": 12995
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.17870090318839e-05,
+      "loss": 0.1832,
+      "step": 12996
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1777049984392574e-05,
+      "loss": 0.1868,
+      "step": 12997
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1767094882709783e-05,
+      "loss": 0.1773,
+      "step": 12998
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1757143727280783e-05,
+      "loss": 0.1933,
+      "step": 12999
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1747196518550596e-05,
+      "loss": 0.1935,
+      "step": 13000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1737253256964165e-05,
+      "loss": 0.186,
+      "step": 13001
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1727313942966179e-05,
+      "loss": 0.1737,
+      "step": 13002
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1717378577001204e-05,
+      "loss": 0.1768,
+      "step": 13003
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1707447159513562e-05,
+      "loss": 0.191,
+      "step": 13004
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1697519690947489e-05,
+      "loss": 0.1893,
+      "step": 13005
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1687596171746962e-05,
+      "loss": 0.1903,
+      "step": 13006
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1677676602355814e-05,
+      "loss": 0.188,
+      "step": 13007
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1667760983217723e-05,
+      "loss": 0.1897,
+      "step": 13008
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1657849314776137e-05,
+      "loss": 0.1924,
+      "step": 13009
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.164794159747441e-05,
+      "loss": 0.1964,
+      "step": 13010
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1638037831755633e-05,
+      "loss": 0.1914,
+      "step": 13011
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1628138018062739e-05,
+      "loss": 0.1824,
+      "step": 13012
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1618242156838543e-05,
+      "loss": 0.184,
+      "step": 13013
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1608350248525601e-05,
+      "loss": 0.1819,
+      "step": 13014
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1598462293566382e-05,
+      "loss": 0.1815,
+      "step": 13015
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1588578292403084e-05,
+      "loss": 0.1814,
+      "step": 13016
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1578698245477803e-05,
+      "loss": 0.1842,
+      "step": 13017
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1568822153232396e-05,
+      "loss": 0.1856,
+      "step": 13018
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1558950016108616e-05,
+      "loss": 0.19,
+      "step": 13019
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1549081834547958e-05,
+      "loss": 0.1876,
+      "step": 13020
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1539217608991826e-05,
+      "loss": 0.186,
+      "step": 13021
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.152935733988134e-05,
+      "loss": 0.191,
+      "step": 13022
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1519501027657575e-05,
+      "loss": 0.194,
+      "step": 13023
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1509648672761297e-05,
+      "loss": 0.1761,
+      "step": 13024
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.149980027563321e-05,
+      "loss": 0.1761,
+      "step": 13025
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1489955836713762e-05,
+      "loss": 0.1878,
+      "step": 13026
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1480115356443245e-05,
+      "loss": 0.1804,
+      "step": 13027
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.147027883526176e-05,
+      "loss": 0.189,
+      "step": 13028
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1460446273609305e-05,
+      "loss": 0.1904,
+      "step": 13029
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1450617671925579e-05,
+      "loss": 0.1859,
+      "step": 13030
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1440793030650233e-05,
+      "loss": 0.1841,
+      "step": 13031
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1430972350222624e-05,
+      "loss": 0.1819,
+      "step": 13032
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1421155631082026e-05,
+      "loss": 0.1953,
+      "step": 13033
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1411342873667475e-05,
+      "loss": 0.179,
+      "step": 13034
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1401534078417864e-05,
+      "loss": 0.186,
+      "step": 13035
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1391729245771876e-05,
+      "loss": 0.183,
+      "step": 13036
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1381928376168072e-05,
+      "loss": 0.1784,
+      "step": 13037
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1372131470044756e-05,
+      "loss": 0.1857,
+      "step": 13038
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1362338527840133e-05,
+      "loss": 0.1759,
+      "step": 13039
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1352549549992165e-05,
+      "loss": 0.182,
+      "step": 13040
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1342764536938711e-05,
+      "loss": 0.1797,
+      "step": 13041
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1332983489117354e-05,
+      "loss": 0.1942,
+      "step": 13042
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1323206406965614e-05,
+      "loss": 0.1906,
+      "step": 13043
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1313433290920728e-05,
+      "loss": 0.1753,
+      "step": 13044
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1303664141419845e-05,
+      "loss": 0.1754,
+      "step": 13045
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1293898958899874e-05,
+      "loss": 0.1869,
+      "step": 13046
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1284137743797551e-05,
+      "loss": 0.1765,
+      "step": 13047
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1274380496549452e-05,
+      "loss": 0.1974,
+      "step": 13048
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1264627217592005e-05,
+      "loss": 0.183,
+      "step": 13049
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1254877907361393e-05,
+      "loss": 0.2008,
+      "step": 13050
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1245132566293703e-05,
+      "loss": 0.1793,
+      "step": 13051
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.123539119482474e-05,
+      "loss": 0.1765,
+      "step": 13052
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1225653793390245e-05,
+      "loss": 0.1825,
+      "step": 13053
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1215920362425681e-05,
+      "loss": 0.1828,
+      "step": 13054
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1206190902366432e-05,
+      "loss": 0.1902,
+      "step": 13055
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1196465413647606e-05,
+      "loss": 0.1771,
+      "step": 13056
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.118674389670421e-05,
+      "loss": 0.1861,
+      "step": 13057
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.117702635197101e-05,
+      "loss": 0.1806,
+      "step": 13058
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1167312779882677e-05,
+      "loss": 0.1854,
+      "step": 13059
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1157603180873599e-05,
+      "loss": 0.1843,
+      "step": 13060
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1147897555378094e-05,
+      "loss": 0.1765,
+      "step": 13061
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1138195903830195e-05,
+      "loss": 0.1897,
+      "step": 13062
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1128498226663875e-05,
+      "loss": 0.1875,
+      "step": 13063
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.111880452431282e-05,
+      "loss": 0.1873,
+      "step": 13064
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1109114797210585e-05,
+      "loss": 0.1886,
+      "step": 13065
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1099429045790588e-05,
+      "loss": 0.1918,
+      "step": 13066
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1089747270485996e-05,
+      "loss": 0.1913,
+      "step": 13067
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1080069471729815e-05,
+      "loss": 0.1916,
+      "step": 13068
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1070395649954923e-05,
+      "loss": 0.1742,
+      "step": 13069
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1060725805593963e-05,
+      "loss": 0.1925,
+      "step": 13070
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1051059939079444e-05,
+      "loss": 0.1929,
+      "step": 13071
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1041398050843642e-05,
+      "loss": 0.1911,
+      "step": 13072
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1031740141318736e-05,
+      "loss": 0.1956,
+      "step": 13073
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1022086210936633e-05,
+      "loss": 0.1857,
+      "step": 13074
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1012436260129156e-05,
+      "loss": 0.1836,
+      "step": 13075
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.100279028932787e-05,
+      "loss": 0.1813,
+      "step": 13076
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0993148298964185e-05,
+      "loss": 0.1785,
+      "step": 13077
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.098351028946939e-05,
+      "loss": 0.1777,
+      "step": 13078
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0973876261274497e-05,
+      "loss": 0.1884,
+      "step": 13079
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0964246214810425e-05,
+      "loss": 0.1926,
+      "step": 13080
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0954620150507854e-05,
+      "loss": 0.1816,
+      "step": 13081
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0944998068797363e-05,
+      "loss": 0.18,
+      "step": 13082
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.093537997010925e-05,
+      "loss": 0.183,
+      "step": 13083
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0925765854873727e-05,
+      "loss": 0.195,
+      "step": 13084
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0916155723520748e-05,
+      "loss": 0.1911,
+      "step": 13085
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0906549576480185e-05,
+      "loss": 0.1855,
+      "step": 13086
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0896947414181646e-05,
+      "loss": 0.1891,
+      "step": 13087
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.088734923705459e-05,
+      "loss": 0.1827,
+      "step": 13088
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0877755045528281e-05,
+      "loss": 0.1856,
+      "step": 13089
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.086816484003187e-05,
+      "loss": 0.1891,
+      "step": 13090
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.085857862099423e-05,
+      "loss": 0.1893,
+      "step": 13091
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0848996388844169e-05,
+      "loss": 0.1805,
+      "step": 13092
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0839418144010195e-05,
+      "loss": 0.1823,
+      "step": 13093
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0829843886920754e-05,
+      "loss": 0.1796,
+      "step": 13094
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0820273618004007e-05,
+      "loss": 0.1917,
+      "step": 13095
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.081070733768803e-05,
+      "loss": 0.1792,
+      "step": 13096
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0801145046400652e-05,
+      "loss": 0.1764,
+      "step": 13097
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0791586744569583e-05,
+      "loss": 0.1863,
+      "step": 13098
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.078203243262228e-05,
+      "loss": 0.1732,
+      "step": 13099
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0772482110986093e-05,
+      "loss": 0.1997,
+      "step": 13100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0762935780088146e-05,
+      "loss": 0.1865,
+      "step": 13101
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.075339344035543e-05,
+      "loss": 0.1911,
+      "step": 13102
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0743855092214694e-05,
+      "loss": 0.196,
+      "step": 13103
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0734320736092573e-05,
+      "loss": 0.1734,
+      "step": 13104
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0724790372415473e-05,
+      "loss": 0.1861,
+      "step": 13105
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0715264001609693e-05,
+      "loss": 0.1895,
+      "step": 13106
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0705741624101228e-05,
+      "loss": 0.1725,
+      "step": 13107
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0696223240316028e-05,
+      "loss": 0.1842,
+      "step": 13108
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.068670885067976e-05,
+      "loss": 0.1903,
+      "step": 13109
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.067719845561802e-05,
+      "loss": 0.1837,
+      "step": 13110
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.06676920555561e-05,
+      "loss": 0.1826,
+      "step": 13111
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0658189650919237e-05,
+      "loss": 0.1801,
+      "step": 13112
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.064869124213237e-05,
+      "loss": 0.1902,
+      "step": 13113
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.063919682962038e-05,
+      "loss": 0.1831,
+      "step": 13114
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.062970641380786e-05,
+      "loss": 0.167,
+      "step": 13115
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0620219995119306e-05,
+      "loss": 0.1852,
+      "step": 13116
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.061073757397898e-05,
+      "loss": 0.1808,
+      "step": 13117
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0601259150811005e-05,
+      "loss": 0.1762,
+      "step": 13118
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0591784726039278e-05,
+      "loss": 0.187,
+      "step": 13119
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0582314300087592e-05,
+      "loss": 0.1783,
+      "step": 13120
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0572847873379465e-05,
+      "loss": 0.1799,
+      "step": 13121
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0563385446338337e-05,
+      "loss": 0.1859,
+      "step": 13122
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0553927019387378e-05,
+      "loss": 0.1853,
+      "step": 13123
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.054447259294965e-05,
+      "loss": 0.1858,
+      "step": 13124
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0535022167447972e-05,
+      "loss": 0.1926,
+      "step": 13125
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0525575743305082e-05,
+      "loss": 0.1775,
+      "step": 13126
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.05161333209434e-05,
+      "loss": 0.1918,
+      "step": 13127
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0506694900785286e-05,
+      "loss": 0.1944,
+      "step": 13128
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0497260483252846e-05,
+      "loss": 0.1737,
+      "step": 13129
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.048783006876808e-05,
+      "loss": 0.1779,
+      "step": 13130
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0478403657752723e-05,
+      "loss": 0.1762,
+      "step": 13131
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0468981250628418e-05,
+      "loss": 0.1882,
+      "step": 13132
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0459562847816551e-05,
+      "loss": 0.1865,
+      "step": 13133
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0450148449738394e-05,
+      "loss": 0.1854,
+      "step": 13134
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.044073805681497e-05,
+      "loss": 0.1792,
+      "step": 13135
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0431331669467214e-05,
+      "loss": 0.1895,
+      "step": 13136
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0421929288115784e-05,
+      "loss": 0.1892,
+      "step": 13137
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0412530913181251e-05,
+      "loss": 0.1841,
+      "step": 13138
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0403136545083903e-05,
+      "loss": 0.1817,
+      "step": 13139
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0393746184243968e-05,
+      "loss": 0.1889,
+      "step": 13140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0384359831081392e-05,
+      "loss": 0.1792,
+      "step": 13141
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0374977486016013e-05,
+      "loss": 0.1748,
+      "step": 13142
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0365599149467442e-05,
+      "loss": 0.1881,
+      "step": 13143
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.035622482185511e-05,
+      "loss": 0.1838,
+      "step": 13144
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0346854503598324e-05,
+      "loss": 0.1942,
+      "step": 13145
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0337488195116173e-05,
+      "loss": 0.1842,
+      "step": 13146
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0328125896827534e-05,
+      "loss": 0.1727,
+      "step": 13147
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0318767609151181e-05,
+      "loss": 0.1855,
+      "step": 13148
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0309413332505625e-05,
+      "loss": 0.1772,
+      "step": 13149
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0300063067309296e-05,
+      "loss": 0.1711,
+      "step": 13150
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0290716813980327e-05,
+      "loss": 0.181,
+      "step": 13151
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0281374572936785e-05,
+      "loss": 0.193,
+      "step": 13152
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0272036344596458e-05,
+      "loss": 0.1938,
+      "step": 13153
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0262702129377055e-05,
+      "loss": 0.1843,
+      "step": 13154
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.025337192769602e-05,
+      "loss": 0.1906,
+      "step": 13155
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0244045739970643e-05,
+      "loss": 0.1841,
+      "step": 13156
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0234723566618065e-05,
+      "loss": 0.1744,
+      "step": 13157
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0225405408055199e-05,
+      "loss": 0.1797,
+      "step": 13158
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0216091264698835e-05,
+      "loss": 0.1846,
+      "step": 13159
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0206781136965526e-05,
+      "loss": 0.1705,
+      "step": 13160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0197475025271686e-05,
+      "loss": 0.1924,
+      "step": 13161
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0188172930033513e-05,
+      "loss": 0.1918,
+      "step": 13162
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0178874851667085e-05,
+      "loss": 0.2021,
+      "step": 13163
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0169580790588219e-05,
+      "loss": 0.1915,
+      "step": 13164
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0160290747212631e-05,
+      "loss": 0.1881,
+      "step": 13165
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0151004721955814e-05,
+      "loss": 0.1842,
+      "step": 13166
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0141722715233093e-05,
+      "loss": 0.1861,
+      "step": 13167
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0132444727459566e-05,
+      "loss": 0.1785,
+      "step": 13168
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.012317075905026e-05,
+      "loss": 0.1725,
+      "step": 13169
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0113900810419895e-05,
+      "loss": 0.1775,
+      "step": 13170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0104634881983132e-05,
+      "loss": 0.1826,
+      "step": 13171
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.009537297415435e-05,
+      "loss": 0.1769,
+      "step": 13172
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0086115087347826e-05,
+      "loss": 0.1841,
+      "step": 13173
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0076861221977584e-05,
+      "loss": 0.1926,
+      "step": 13174
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0067611378457541e-05,
+      "loss": 0.1735,
+      "step": 13175
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0058365557201377e-05,
+      "loss": 0.1839,
+      "step": 13176
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0049123758622636e-05,
+      "loss": 0.1801,
+      "step": 13177
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0039885983134644e-05,
+      "loss": 0.168,
+      "step": 13178
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0030652231150584e-05,
+      "loss": 0.1807,
+      "step": 13179
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.00214225030834e-05,
+      "loss": 0.1871,
+      "step": 13180
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0012196799345952e-05,
+      "loss": 0.1829,
+      "step": 13181
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.000297512035081e-05,
+      "loss": 0.1718,
+      "step": 13182
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.993757466510456e-06,
+      "loss": 0.1849,
+      "step": 13183
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.984543838237127e-06,
+      "loss": 0.1836,
+      "step": 13184
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.975334235942924e-06,
+      "loss": 0.184,
+      "step": 13185
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.966128660039752e-06,
+      "loss": 0.1781,
+      "step": 13186
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.956927110939329e-06,
+      "loss": 0.1784,
+      "step": 13187
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.947729589053179e-06,
+      "loss": 0.1945,
+      "step": 13188
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.938536094792694e-06,
+      "loss": 0.1833,
+      "step": 13189
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.929346628569026e-06,
+      "loss": 0.1861,
+      "step": 13190
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.92016119079322e-06,
+      "loss": 0.1919,
+      "step": 13191
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.91097978187605e-06,
+      "loss": 0.188,
+      "step": 13192
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.90180240222821e-06,
+      "loss": 0.1831,
+      "step": 13193
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.892629052260105e-06,
+      "loss": 0.1859,
+      "step": 13194
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.883459732382072e-06,
+      "loss": 0.1834,
+      "step": 13195
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.874294443004172e-06,
+      "loss": 0.1802,
+      "step": 13196
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.865133184536356e-06,
+      "loss": 0.1848,
+      "step": 13197
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.855975957388342e-06,
+      "loss": 0.1835,
+      "step": 13198
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.84682276196971e-06,
+      "loss": 0.1803,
+      "step": 13199
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.837673598689811e-06,
+      "loss": 0.189,
+      "step": 13200
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.828528467957888e-06,
+      "loss": 0.1799,
+      "step": 13201
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.81938737018292e-06,
+      "loss": 0.1744,
+      "step": 13202
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.810250305773783e-06,
+      "loss": 0.1836,
+      "step": 13203
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.801117275139093e-06,
+      "loss": 0.1835,
+      "step": 13204
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.79198827868737e-06,
+      "loss": 0.1837,
+      "step": 13205
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.782863316826907e-06,
+      "loss": 0.1818,
+      "step": 13206
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.773742389965801e-06,
+      "loss": 0.1862,
+      "step": 13207
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.764625498511981e-06,
+      "loss": 0.1855,
+      "step": 13208
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.755512642873244e-06,
+      "loss": 0.1898,
+      "step": 13209
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.746403823457117e-06,
+      "loss": 0.1928,
+      "step": 13210
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.737299040671044e-06,
+      "loss": 0.1806,
+      "step": 13211
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.728198294922197e-06,
+      "loss": 0.179,
+      "step": 13212
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.719101586617663e-06,
+      "loss": 0.181,
+      "step": 13213
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.710008916164238e-06,
+      "loss": 0.1822,
+      "step": 13214
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.70092028396864e-06,
+      "loss": 0.1777,
+      "step": 13215
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.691835690437323e-06,
+      "loss": 0.189,
+      "step": 13216
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.68275513597664e-06,
+      "loss": 0.1875,
+      "step": 13217
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.673678620992688e-06,
+      "loss": 0.1935,
+      "step": 13218
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.664606145891453e-06,
+      "loss": 0.1881,
+      "step": 13219
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.655537711078688e-06,
+      "loss": 0.1903,
+      "step": 13220
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.646473316959947e-06,
+      "loss": 0.1861,
+      "step": 13221
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.637412963940705e-06,
+      "loss": 0.1834,
+      "step": 13222
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.62835665242614e-06,
+      "loss": 0.1803,
+      "step": 13223
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.619304382821325e-06,
+      "loss": 0.1902,
+      "step": 13224
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.610256155531105e-06,
+      "loss": 0.1833,
+      "step": 13225
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.6012119709602e-06,
+      "loss": 0.179,
+      "step": 13226
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.592171829513086e-06,
+      "loss": 0.1724,
+      "step": 13227
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.583135731594083e-06,
+      "loss": 0.1941,
+      "step": 13228
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.57410367760736e-06,
+      "loss": 0.1856,
+      "step": 13229
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.565075667956846e-06,
+      "loss": 0.1783,
+      "step": 13230
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.556051703046365e-06,
+      "loss": 0.1964,
+      "step": 13231
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.547031783279491e-06,
+      "loss": 0.1784,
+      "step": 13232
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.538015909059627e-06,
+      "loss": 0.1723,
+      "step": 13233
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.52900408079006e-06,
+      "loss": 0.1821,
+      "step": 13234
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.51999629887379e-06,
+      "loss": 0.1812,
+      "step": 13235
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.51099256371375e-06,
+      "loss": 0.1793,
+      "step": 13236
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.501992875712595e-06,
+      "loss": 0.1854,
+      "step": 13237
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.49299723527286e-06,
+      "loss": 0.1829,
+      "step": 13238
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.484005642796868e-06,
+      "loss": 0.1847,
+      "step": 13239
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.475018098686783e-06,
+      "loss": 0.1867,
+      "step": 13240
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.466034603344565e-06,
+      "loss": 0.179,
+      "step": 13241
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.457055157172023e-06,
+      "loss": 0.1956,
+      "step": 13242
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.448079760570727e-06,
+      "loss": 0.1748,
+      "step": 13243
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.439108413942166e-06,
+      "loss": 0.1823,
+      "step": 13244
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.430141117687519e-06,
+      "loss": 0.1752,
+      "step": 13245
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.421177872207943e-06,
+      "loss": 0.1769,
+      "step": 13246
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.412218677904217e-06,
+      "loss": 0.186,
+      "step": 13247
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.40326353517711e-06,
+      "loss": 0.1843,
+      "step": 13248
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.394312444427112e-06,
+      "loss": 0.1816,
+      "step": 13249
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.385365406054602e-06,
+      "loss": 0.1785,
+      "step": 13250
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.376422420459697e-06,
+      "loss": 0.1809,
+      "step": 13251
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.367483488042429e-06,
+      "loss": 0.1862,
+      "step": 13252
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.358548609202533e-06,
+      "loss": 0.1785,
+      "step": 13253
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.34961778433967e-06,
+      "loss": 0.1806,
+      "step": 13254
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.340691013853253e-06,
+      "loss": 0.1882,
+      "step": 13255
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.331768298142562e-06,
+      "loss": 0.1822,
+      "step": 13256
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.322849637606635e-06,
+      "loss": 0.1721,
+      "step": 13257
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.313935032644394e-06,
+      "loss": 0.1834,
+      "step": 13258
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.305024483654512e-06,
+      "loss": 0.1821,
+      "step": 13259
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.296117991035557e-06,
+      "loss": 0.1874,
+      "step": 13260
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.287215555185847e-06,
+      "loss": 0.1833,
+      "step": 13261
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.27831717650357e-06,
+      "loss": 0.1854,
+      "step": 13262
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.269422855386677e-06,
+      "loss": 0.1873,
+      "step": 13263
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.260532592233017e-06,
+      "loss": 0.1824,
+      "step": 13264
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.25164638744016e-06,
+      "loss": 0.1903,
+      "step": 13265
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.24276424140561e-06,
+      "loss": 0.179,
+      "step": 13266
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.233886154526538e-06,
+      "loss": 0.1845,
+      "step": 13267
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.225012127200095e-06,
+      "loss": 0.1832,
+      "step": 13268
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.216142159823116e-06,
+      "loss": 0.1821,
+      "step": 13269
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.207276252792374e-06,
+      "loss": 0.1894,
+      "step": 13270
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.198414406504352e-06,
+      "loss": 0.1793,
+      "step": 13271
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.189556621355432e-06,
+      "loss": 0.1816,
+      "step": 13272
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.180702897741756e-06,
+      "loss": 0.1892,
+      "step": 13273
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.171853236059335e-06,
+      "loss": 0.185,
+      "step": 13274
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.163007636703958e-06,
+      "loss": 0.1919,
+      "step": 13275
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.154166100071259e-06,
+      "loss": 0.1843,
+      "step": 13276
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.14532862655666e-06,
+      "loss": 0.1769,
+      "step": 13277
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.136495216555451e-06,
+      "loss": 0.1885,
+      "step": 13278
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.127665870462676e-06,
+      "loss": 0.1763,
+      "step": 13279
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.118840588673261e-06,
+      "loss": 0.1796,
+      "step": 13280
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.11001937158189e-06,
+      "loss": 0.1842,
+      "step": 13281
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.101202219583138e-06,
+      "loss": 0.1809,
+      "step": 13282
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.09238913307131e-06,
+      "loss": 0.1785,
+      "step": 13283
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.083580112440614e-06,
+      "loss": 0.1896,
+      "step": 13284
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.074775158085013e-06,
+      "loss": 0.1903,
+      "step": 13285
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.065974270398326e-06,
+      "loss": 0.1848,
+      "step": 13286
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.057177449774157e-06,
+      "loss": 0.1773,
+      "step": 13287
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.04838469660596e-06,
+      "loss": 0.1828,
+      "step": 13288
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.039596011286988e-06,
+      "loss": 0.1921,
+      "step": 13289
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.030811394210348e-06,
+      "loss": 0.1812,
+      "step": 13290
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.02203084576888e-06,
+      "loss": 0.1832,
+      "step": 13291
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.013254366355351e-06,
+      "loss": 0.2014,
+      "step": 13292
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.004481956362266e-06,
+      "loss": 0.1805,
+      "step": 13293
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.99571361618199e-06,
+      "loss": 0.191,
+      "step": 13294
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.986949346206653e-06,
+      "loss": 0.189,
+      "step": 13295
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.978189146828298e-06,
+      "loss": 0.1802,
+      "step": 13296
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.969433018438677e-06,
+      "loss": 0.1863,
+      "step": 13297
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.960680961429457e-06,
+      "loss": 0.182,
+      "step": 13298
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.951932976192057e-06,
+      "loss": 0.1769,
+      "step": 13299
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.943189063117708e-06,
+      "loss": 0.1806,
+      "step": 13300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.934449222597529e-06,
+      "loss": 0.1798,
+      "step": 13301
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.925713455022378e-06,
+      "loss": 0.1878,
+      "step": 13302
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.916981760782995e-06,
+      "loss": 0.1866,
+      "step": 13303
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.908254140269867e-06,
+      "loss": 0.1827,
+      "step": 13304
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.899530593873396e-06,
+      "loss": 0.1801,
+      "step": 13305
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.890811121983722e-06,
+      "loss": 0.1821,
+      "step": 13306
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.882095724990803e-06,
+      "loss": 0.1859,
+      "step": 13307
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.873384403284468e-06,
+      "loss": 0.1805,
+      "step": 13308
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.86467715725432e-06,
+      "loss": 0.1911,
+      "step": 13309
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.855973987289812e-06,
+      "loss": 0.1822,
+      "step": 13310
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.847274893780177e-06,
+      "loss": 0.1826,
+      "step": 13311
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.838579877114483e-06,
+      "loss": 0.1871,
+      "step": 13312
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.829888937681651e-06,
+      "loss": 0.1734,
+      "step": 13313
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.821202075870339e-06,
+      "loss": 0.1823,
+      "step": 13314
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.812519292069121e-06,
+      "loss": 0.1819,
+      "step": 13315
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.80384058666629e-06,
+      "loss": 0.1875,
+      "step": 13316
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.795165960050056e-06,
+      "loss": 0.1819,
+      "step": 13317
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.786495412608342e-06,
+      "loss": 0.1797,
+      "step": 13318
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.777828944728983e-06,
+      "loss": 0.1832,
+      "step": 13319
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.769166556799557e-06,
+      "loss": 0.1803,
+      "step": 13320
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.760508249207532e-06,
+      "loss": 0.1769,
+      "step": 13321
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.751854022340112e-06,
+      "loss": 0.1797,
+      "step": 13322
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.743203876584405e-06,
+      "loss": 0.1793,
+      "step": 13323
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.734557812327238e-06,
+      "loss": 0.1828,
+      "step": 13324
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.725915829955377e-06,
+      "loss": 0.1798,
+      "step": 13325
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.717277929855294e-06,
+      "loss": 0.1865,
+      "step": 13326
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.708644112413334e-06,
+      "loss": 0.1802,
+      "step": 13327
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.70001437801562e-06,
+      "loss": 0.1836,
+      "step": 13328
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.691388727048167e-06,
+      "loss": 0.1831,
+      "step": 13329
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.682767159896733e-06,
+      "loss": 0.1764,
+      "step": 13330
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.67414967694693e-06,
+      "loss": 0.1801,
+      "step": 13331
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.665536278584163e-06,
+      "loss": 0.18,
+      "step": 13332
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.656926965193713e-06,
+      "loss": 0.1849,
+      "step": 13333
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.648321737160569e-06,
+      "loss": 0.1839,
+      "step": 13334
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.639720594869671e-06,
+      "loss": 0.1812,
+      "step": 13335
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.631123538705654e-06,
+      "loss": 0.1951,
+      "step": 13336
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.622530569053078e-06,
+      "loss": 0.1771,
+      "step": 13337
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.6139416862962e-06,
+      "loss": 0.1864,
+      "step": 13338
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.605356890819238e-06,
+      "loss": 0.1768,
+      "step": 13339
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.596776183006083e-06,
+      "loss": 0.1694,
+      "step": 13340
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.588199563240551e-06,
+      "loss": 0.1739,
+      "step": 13341
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.579627031906212e-06,
+      "loss": 0.1893,
+      "step": 13342
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.571058589386516e-06,
+      "loss": 0.1829,
+      "step": 13343
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.562494236064633e-06,
+      "loss": 0.1857,
+      "step": 13344
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.553933972323647e-06,
+      "loss": 0.1879,
+      "step": 13345
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.545377798546416e-06,
+      "loss": 0.2006,
+      "step": 13346
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.536825715115615e-06,
+      "loss": 0.1822,
+      "step": 13347
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.528277722413714e-06,
+      "loss": 0.1823,
+      "step": 13348
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.519733820823062e-06,
+      "loss": 0.1763,
+      "step": 13349
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.511194010725764e-06,
+      "loss": 0.1822,
+      "step": 13350
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.502658292503785e-06,
+      "loss": 0.1743,
+      "step": 13351
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.494126666538871e-06,
+      "loss": 0.1873,
+      "step": 13352
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.485599133212619e-06,
+      "loss": 0.1861,
+      "step": 13353
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.47707569290641e-06,
+      "loss": 0.1915,
+      "step": 13354
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.468556346001477e-06,
+      "loss": 0.1897,
+      "step": 13355
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.460041092878834e-06,
+      "loss": 0.1816,
+      "step": 13356
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.451529933919344e-06,
+      "loss": 0.1796,
+      "step": 13357
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.443022869503658e-06,
+      "loss": 0.1821,
+      "step": 13358
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.434519900012272e-06,
+      "loss": 0.1824,
+      "step": 13359
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.42602102582547e-06,
+      "loss": 0.1678,
+      "step": 13360
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.417526247323381e-06,
+      "loss": 0.1784,
+      "step": 13361
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.409035564885925e-06,
+      "loss": 0.1871,
+      "step": 13362
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.400548978892875e-06,
+      "loss": 0.1843,
+      "step": 13363
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.392066489723782e-06,
+      "loss": 0.1776,
+      "step": 13364
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.38358809775801e-06,
+      "loss": 0.1884,
+      "step": 13365
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.37511380337479e-06,
+      "loss": 0.1872,
+      "step": 13366
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.366643606953129e-06,
+      "loss": 0.1765,
+      "step": 13367
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.358177508871834e-06,
+      "loss": 0.1721,
+      "step": 13368
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.349715509509593e-06,
+      "loss": 0.182,
+      "step": 13369
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.341257609244846e-06,
+      "loss": 0.1795,
+      "step": 13370
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.332803808455891e-06,
+      "loss": 0.1821,
+      "step": 13371
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.324354107520816e-06,
+      "loss": 0.1872,
+      "step": 13372
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.315908506817561e-06,
+      "loss": 0.1792,
+      "step": 13373
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.307467006723823e-06,
+      "loss": 0.1808,
+      "step": 13374
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.299029607617193e-06,
+      "loss": 0.1797,
+      "step": 13375
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.290596309875009e-06,
+      "loss": 0.1806,
+      "step": 13376
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.28216711387444e-06,
+      "loss": 0.1837,
+      "step": 13377
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.273742019992525e-06,
+      "loss": 0.187,
+      "step": 13378
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.265321028606054e-06,
+      "loss": 0.178,
+      "step": 13379
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.25690414009167e-06,
+      "loss": 0.1866,
+      "step": 13380
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.248491354825804e-06,
+      "loss": 0.1859,
+      "step": 13381
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.240082673184758e-06,
+      "loss": 0.1769,
+      "step": 13382
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.23167809554457e-06,
+      "loss": 0.181,
+      "step": 13383
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.223277622281168e-06,
+      "loss": 0.1741,
+      "step": 13384
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.214881253770257e-06,
+      "loss": 0.1816,
+      "step": 13385
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.206488990387374e-06,
+      "loss": 0.1876,
+      "step": 13386
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.19810083250786e-06,
+      "loss": 0.1757,
+      "step": 13387
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.189716780506884e-06,
+      "loss": 0.1817,
+      "step": 13388
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.181336834759412e-06,
+      "loss": 0.1876,
+      "step": 13389
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.172960995640255e-06,
+      "loss": 0.1863,
+      "step": 13390
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.164589263524015e-06,
+      "loss": 0.1817,
+      "step": 13391
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.156221638785144e-06,
+      "loss": 0.1811,
+      "step": 13392
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.147858121797846e-06,
+      "loss": 0.1892,
+      "step": 13393
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.139498712936232e-06,
+      "loss": 0.169,
+      "step": 13394
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.131143412574138e-06,
+      "loss": 0.1876,
+      "step": 13395
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.122792221085307e-06,
+      "loss": 0.1884,
+      "step": 13396
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.114445138843186e-06,
+      "loss": 0.1842,
+      "step": 13397
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.106102166221163e-06,
+      "loss": 0.1847,
+      "step": 13398
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.09776330359232e-06,
+      "loss": 0.1903,
+      "step": 13399
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.08942855132968e-06,
+      "loss": 0.1766,
+      "step": 13400
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.081097909805967e-06,
+      "loss": 0.1784,
+      "step": 13401
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.072771379393818e-06,
+      "loss": 0.1792,
+      "step": 13402
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.064448960465587e-06,
+      "loss": 0.174,
+      "step": 13403
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.056130653393556e-06,
+      "loss": 0.1711,
+      "step": 13404
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.047816458549707e-06,
+      "loss": 0.1854,
+      "step": 13405
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.03950637630596e-06,
+      "loss": 0.1908,
+      "step": 13406
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.031200407033923e-06,
+      "loss": 0.1873,
+      "step": 13407
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.022898551105118e-06,
+      "loss": 0.1895,
+      "step": 13408
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.014600808890838e-06,
+      "loss": 0.1864,
+      "step": 13409
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.006307180762219e-06,
+      "loss": 0.1738,
+      "step": 13410
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.998017667090163e-06,
+      "loss": 0.1759,
+      "step": 13411
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.989732268245475e-06,
+      "loss": 0.1924,
+      "step": 13412
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.98145098459867e-06,
+      "loss": 0.1805,
+      "step": 13413
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.973173816520164e-06,
+      "loss": 0.1782,
+      "step": 13414
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.964900764380145e-06,
+      "loss": 0.1864,
+      "step": 13415
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.956631828548633e-06,
+      "loss": 0.1827,
+      "step": 13416
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.948367009395452e-06,
+      "loss": 0.1784,
+      "step": 13417
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.940106307290274e-06,
+      "loss": 0.1841,
+      "step": 13418
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.931849722602513e-06,
+      "loss": 0.1739,
+      "step": 13419
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.92359725570151e-06,
+      "loss": 0.1883,
+      "step": 13420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.91534890695631e-06,
+      "loss": 0.1697,
+      "step": 13421
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.907104676735855e-06,
+      "loss": 0.1918,
+      "step": 13422
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.898864565408848e-06,
+      "loss": 0.1773,
+      "step": 13423
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.89062857334385e-06,
+      "loss": 0.1896,
+      "step": 13424
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.8823967009092e-06,
+      "loss": 0.1814,
+      "step": 13425
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.874168948473126e-06,
+      "loss": 0.177,
+      "step": 13426
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.865945316403533e-06,
+      "loss": 0.1769,
+      "step": 13427
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.857725805068295e-06,
+      "loss": 0.1883,
+      "step": 13428
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.849510414834981e-06,
+      "loss": 0.1852,
+      "step": 13429
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.841299146071079e-06,
+      "loss": 0.1845,
+      "step": 13430
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.833091999143793e-06,
+      "loss": 0.1829,
+      "step": 13431
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.82488897442023e-06,
+      "loss": 0.1813,
+      "step": 13432
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.816690072267252e-06,
+      "loss": 0.1823,
+      "step": 13433
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.808495293051576e-06,
+      "loss": 0.1837,
+      "step": 13434
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.800304637139689e-06,
+      "loss": 0.1839,
+      "step": 13435
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.79211810489795e-06,
+      "loss": 0.1722,
+      "step": 13436
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.783935696692491e-06,
+      "loss": 0.1807,
+      "step": 13437
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.775757412889285e-06,
+      "loss": 0.1766,
+      "step": 13438
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.767583253854072e-06,
+      "loss": 0.1811,
+      "step": 13439
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.759413219952504e-06,
+      "loss": 0.1829,
+      "step": 13440
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.751247311549958e-06,
+      "loss": 0.1838,
+      "step": 13441
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.74308552901163e-06,
+      "loss": 0.1863,
+      "step": 13442
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.734927872702613e-06,
+      "loss": 0.1859,
+      "step": 13443
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.726774342987719e-06,
+      "loss": 0.1767,
+      "step": 13444
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.718624940231645e-06,
+      "loss": 0.1843,
+      "step": 13445
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.710479664798875e-06,
+      "loss": 0.1828,
+      "step": 13446
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.702338517053676e-06,
+      "loss": 0.1912,
+      "step": 13447
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.694201497360209e-06,
+      "loss": 0.1861,
+      "step": 13448
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.686068606082364e-06,
+      "loss": 0.1734,
+      "step": 13449
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.677939843583937e-06,
+      "loss": 0.1827,
+      "step": 13450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.669815210228438e-06,
+      "loss": 0.1891,
+      "step": 13451
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.661694706379296e-06,
+      "loss": 0.1875,
+      "step": 13452
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.653578332399648e-06,
+      "loss": 0.1851,
+      "step": 13453
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.645466088652554e-06,
+      "loss": 0.1754,
+      "step": 13454
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.637357975500825e-06,
+      "loss": 0.1772,
+      "step": 13455
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.6292539933070615e-06,
+      "loss": 0.183,
+      "step": 13456
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.621154142433762e-06,
+      "loss": 0.1893,
+      "step": 13457
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.613058423243169e-06,
+      "loss": 0.177,
+      "step": 13458
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.604966836097394e-06,
+      "loss": 0.1799,
+      "step": 13459
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.596879381358301e-06,
+      "loss": 0.1855,
+      "step": 13460
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.588796059387648e-06,
+      "loss": 0.178,
+      "step": 13461
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.580716870546922e-06,
+      "loss": 0.1826,
+      "step": 13462
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.572641815197501e-06,
+      "loss": 0.185,
+      "step": 13463
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.564570893700518e-06,
+      "loss": 0.1714,
+      "step": 13464
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.556504106416973e-06,
+      "loss": 0.1935,
+      "step": 13465
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.548441453707655e-06,
+      "loss": 0.1799,
+      "step": 13466
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.5403829359331545e-06,
+      "loss": 0.1814,
+      "step": 13467
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.532328553453882e-06,
+      "loss": 0.1768,
+      "step": 13468
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.524278306630106e-06,
+      "loss": 0.162,
+      "step": 13469
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.516232195821837e-06,
+      "loss": 0.1699,
+      "step": 13470
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.508190221388988e-06,
+      "loss": 0.1867,
+      "step": 13471
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.5001523836911945e-06,
+      "loss": 0.1915,
+      "step": 13472
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.49211868308799e-06,
+      "loss": 0.1834,
+      "step": 13473
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.484089119938653e-06,
+      "loss": 0.1751,
+      "step": 13474
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.476063694602342e-06,
+      "loss": 0.1747,
+      "step": 13475
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.46804240743797e-06,
+      "loss": 0.1739,
+      "step": 13476
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.460025258804304e-06,
+      "loss": 0.183,
+      "step": 13477
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.452012249059903e-06,
+      "loss": 0.1793,
+      "step": 13478
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.44400337856318e-06,
+      "loss": 0.1844,
+      "step": 13479
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.435998647672304e-06,
+      "loss": 0.1811,
+      "step": 13480
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.4279980567453115e-06,
+      "loss": 0.1809,
+      "step": 13481
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.420001606140004e-06,
+      "loss": 0.178,
+      "step": 13482
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.412009296214073e-06,
+      "loss": 0.1778,
+      "step": 13483
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.4040211273249225e-06,
+      "loss": 0.1763,
+      "step": 13484
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.396037099829878e-06,
+      "loss": 0.1726,
+      "step": 13485
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.388057214086008e-06,
+      "loss": 0.1799,
+      "step": 13486
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.380081470450217e-06,
+      "loss": 0.1861,
+      "step": 13487
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.3721098692791975e-06,
+      "loss": 0.1783,
+      "step": 13488
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.36414241092952e-06,
+      "loss": 0.1881,
+      "step": 13489
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.356179095757498e-06,
+      "loss": 0.1794,
+      "step": 13490
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.348219924119337e-06,
+      "loss": 0.1795,
+      "step": 13491
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.340264896370974e-06,
+      "loss": 0.1753,
+      "step": 13492
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.3323140128682355e-06,
+      "loss": 0.1819,
+      "step": 13493
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.324367273966692e-06,
+      "loss": 0.1771,
+      "step": 13494
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.316424680021794e-06,
+      "loss": 0.1759,
+      "step": 13495
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.308486231388756e-06,
+      "loss": 0.1836,
+      "step": 13496
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.300551928422661e-06,
+      "loss": 0.1681,
+      "step": 13497
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.292621771478336e-06,
+      "loss": 0.1701,
+      "step": 13498
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.284695760910487e-06,
+      "loss": 0.1679,
+      "step": 13499
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.276773897073585e-06,
+      "loss": 0.1761,
+      "step": 13500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.268856180321959e-06,
+      "loss": 0.1989,
+      "step": 13501
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.260942611009714e-06,
+      "loss": 0.1797,
+      "step": 13502
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.253033189490821e-06,
+      "loss": 0.1861,
+      "step": 13503
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.245127916118988e-06,
+      "loss": 0.1739,
+      "step": 13504
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.23722679124782e-06,
+      "loss": 0.1803,
+      "step": 13505
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.22932981523069e-06,
+      "loss": 0.1785,
+      "step": 13506
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.221436988420771e-06,
+      "loss": 0.1927,
+      "step": 13507
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.213548311171081e-06,
+      "loss": 0.1723,
+      "step": 13508
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.205663783834471e-06,
+      "loss": 0.1697,
+      "step": 13509
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.197783406763537e-06,
+      "loss": 0.1771,
+      "step": 13510
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.189907180310762e-06,
+      "loss": 0.1797,
+      "step": 13511
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.182035104828399e-06,
+      "loss": 0.1821,
+      "step": 13512
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.174167180668556e-06,
+      "loss": 0.1722,
+      "step": 13513
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.166303408183084e-06,
+      "loss": 0.1796,
+      "step": 13514
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.158443787723745e-06,
+      "loss": 0.1805,
+      "step": 13515
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.150588319642027e-06,
+      "loss": 0.1803,
+      "step": 13516
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.142737004289291e-06,
+      "loss": 0.1734,
+      "step": 13517
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.134889842016656e-06,
+      "loss": 0.1752,
+      "step": 13518
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.127046833175144e-06,
+      "loss": 0.1794,
+      "step": 13519
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.119207978115494e-06,
+      "loss": 0.1759,
+      "step": 13520
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.1113732771883045e-06,
+      "loss": 0.1755,
+      "step": 13521
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.103542730744006e-06,
+      "loss": 0.1798,
+      "step": 13522
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.095716339132808e-06,
+      "loss": 0.1758,
+      "step": 13523
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.087894102704762e-06,
+      "loss": 0.1864,
+      "step": 13524
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.0800760218097e-06,
+      "loss": 0.1806,
+      "step": 13525
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.072262096797322e-06,
+      "loss": 0.1766,
+      "step": 13526
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.064452328017079e-06,
+      "loss": 0.177,
+      "step": 13527
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.056646715818272e-06,
+      "loss": 0.1836,
+      "step": 13528
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.048845260550019e-06,
+      "loss": 0.1808,
+      "step": 13529
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.0410479625612315e-06,
+      "loss": 0.1826,
+      "step": 13530
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.033254822200664e-06,
+      "loss": 0.172,
+      "step": 13531
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.0254658398168694e-06,
+      "loss": 0.1781,
+      "step": 13532
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.017681015758171e-06,
+      "loss": 0.1693,
+      "step": 13533
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.0099003503728e-06,
+      "loss": 0.177,
+      "step": 13534
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.0021238440087235e-06,
+      "loss": 0.1772,
+      "step": 13535
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.994351497013763e-06,
+      "loss": 0.1689,
+      "step": 13536
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.986583309735517e-06,
+      "loss": 0.1855,
+      "step": 13537
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.978819282521443e-06,
+      "loss": 0.1828,
+      "step": 13538
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.9710594157187726e-06,
+      "loss": 0.1806,
+      "step": 13539
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.963303709674606e-06,
+      "loss": 0.1843,
+      "step": 13540
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.955552164735768e-06,
+      "loss": 0.1857,
+      "step": 13541
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.94780478124899e-06,
+      "loss": 0.1806,
+      "step": 13542
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.940061559560751e-06,
+      "loss": 0.1837,
+      "step": 13543
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.932322500017397e-06,
+      "loss": 0.1781,
+      "step": 13544
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.924587602965027e-06,
+      "loss": 0.1779,
+      "step": 13545
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.916856868749644e-06,
+      "loss": 0.1777,
+      "step": 13546
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.909130297716937e-06,
+      "loss": 0.1903,
+      "step": 13547
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.901407890212519e-06,
+      "loss": 0.1766,
+      "step": 13548
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.893689646581769e-06,
+      "loss": 0.1765,
+      "step": 13549
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.8859755671699e-06,
+      "loss": 0.1795,
+      "step": 13550
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.878265652321902e-06,
+      "loss": 0.1771,
+      "step": 13551
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.870559902382645e-06,
+      "loss": 0.1773,
+      "step": 13552
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.86285831769673e-06,
+      "loss": 0.1668,
+      "step": 13553
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.855160898608648e-06,
+      "loss": 0.1798,
+      "step": 13554
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.847467645462635e-06,
+      "loss": 0.1831,
+      "step": 13555
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.839778558602816e-06,
+      "loss": 0.1787,
+      "step": 13556
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.832093638373039e-06,
+      "loss": 0.1827,
+      "step": 13557
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.82441288511707e-06,
+      "loss": 0.1759,
+      "step": 13558
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.816736299178383e-06,
+      "loss": 0.1692,
+      "step": 13559
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.809063880900357e-06,
+      "loss": 0.1895,
+      "step": 13560
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.801395630626106e-06,
+      "loss": 0.1844,
+      "step": 13561
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.793731548698634e-06,
+      "loss": 0.179,
+      "step": 13562
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.7860716354607e-06,
+      "loss": 0.1806,
+      "step": 13563
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.778415891254897e-06,
+      "loss": 0.1766,
+      "step": 13564
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.770764316423628e-06,
+      "loss": 0.171,
+      "step": 13565
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.763116911309153e-06,
+      "loss": 0.1838,
+      "step": 13566
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.755473676253432e-06,
+      "loss": 0.1721,
+      "step": 13567
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.74783461159838e-06,
+      "loss": 0.1831,
+      "step": 13568
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.740199717685602e-06,
+      "loss": 0.187,
+      "step": 13569
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.732568994856614e-06,
+      "loss": 0.1818,
+      "step": 13570
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.724942443452675e-06,
+      "loss": 0.1849,
+      "step": 13571
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.717320063814914e-06,
+      "loss": 0.184,
+      "step": 13572
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.709701856284212e-06,
+      "loss": 0.1806,
+      "step": 13573
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.702087821201331e-06,
+      "loss": 0.1786,
+      "step": 13574
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.694477958906786e-06,
+      "loss": 0.1786,
+      "step": 13575
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.686872269740952e-06,
+      "loss": 0.1805,
+      "step": 13576
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.6792707540439646e-06,
+      "loss": 0.1796,
+      "step": 13577
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.671673412155844e-06,
+      "loss": 0.1734,
+      "step": 13578
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.664080244416348e-06,
+      "loss": 0.1848,
+      "step": 13579
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.65649125116512e-06,
+      "loss": 0.1775,
+      "step": 13580
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.648906432741541e-06,
+      "loss": 0.1817,
+      "step": 13581
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.641325789484898e-06,
+      "loss": 0.1822,
+      "step": 13582
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.633749321734173e-06,
+      "loss": 0.1766,
+      "step": 13583
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.6261770298282955e-06,
+      "loss": 0.1803,
+      "step": 13584
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.618608914105895e-06,
+      "loss": 0.1804,
+      "step": 13585
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.611044974905467e-06,
+      "loss": 0.181,
+      "step": 13586
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.603485212565297e-06,
+      "loss": 0.1792,
+      "step": 13587
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.595929627423536e-06,
+      "loss": 0.1779,
+      "step": 13588
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.58837821981807e-06,
+      "loss": 0.1753,
+      "step": 13589
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.580830990086684e-06,
+      "loss": 0.1701,
+      "step": 13590
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.5732879385668746e-06,
+      "loss": 0.186,
+      "step": 13591
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.5657490655960716e-06,
+      "loss": 0.1819,
+      "step": 13592
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.558214371511395e-06,
+      "loss": 0.1866,
+      "step": 13593
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.5506838566498866e-06,
+      "loss": 0.1817,
+      "step": 13594
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.543157521348309e-06,
+      "loss": 0.171,
+      "step": 13595
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.535635365943315e-06,
+      "loss": 0.1799,
+      "step": 13596
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.528117390771327e-06,
+      "loss": 0.1797,
+      "step": 13597
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.520603596168562e-06,
+      "loss": 0.1775,
+      "step": 13598
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.513093982471119e-06,
+      "loss": 0.1805,
+      "step": 13599
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.50558855001484e-06,
+      "loss": 0.1746,
+      "step": 13600
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.498087299135436e-06,
+      "loss": 0.1802,
+      "step": 13601
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.4905902301683585e-06,
+      "loss": 0.1814,
+      "step": 13602
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.483097343448963e-06,
+      "loss": 0.1796,
+      "step": 13603
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.475608639312347e-06,
+      "loss": 0.1856,
+      "step": 13604
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.468124118093466e-06,
+      "loss": 0.1835,
+      "step": 13605
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.460643780127051e-06,
+      "loss": 0.1673,
+      "step": 13606
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.453167625747647e-06,
+      "loss": 0.1834,
+      "step": 13607
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.445695655289674e-06,
+      "loss": 0.1731,
+      "step": 13608
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.438227869087287e-06,
+      "loss": 0.1824,
+      "step": 13609
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.430764267474487e-06,
+      "loss": 0.1811,
+      "step": 13610
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.423304850785095e-06,
+      "loss": 0.1864,
+      "step": 13611
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.415849619352721e-06,
+      "loss": 0.175,
+      "step": 13612
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.408398573510843e-06,
+      "loss": 0.183,
+      "step": 13613
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.400951713592651e-06,
+      "loss": 0.1798,
+      "step": 13614
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.393509039931278e-06,
+      "loss": 0.1736,
+      "step": 13615
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.386070552859536e-06,
+      "loss": 0.1911,
+      "step": 13616
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.3786362527101685e-06,
+      "loss": 0.1815,
+      "step": 13617
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.371206139815633e-06,
+      "loss": 0.1794,
+      "step": 13618
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.363780214508274e-06,
+      "loss": 0.1825,
+      "step": 13619
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.356358477120206e-06,
+      "loss": 0.1905,
+      "step": 13620
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.348940927983383e-06,
+      "loss": 0.182,
+      "step": 13621
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.34152756742954e-06,
+      "loss": 0.1761,
+      "step": 13622
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.334118395790256e-06,
+      "loss": 0.1742,
+      "step": 13623
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.326713413396889e-06,
+      "loss": 0.1915,
+      "step": 13624
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.319312620580664e-06,
+      "loss": 0.1788,
+      "step": 13625
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.31191601767257e-06,
+      "loss": 0.1837,
+      "step": 13626
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.304523605003421e-06,
+      "loss": 0.1868,
+      "step": 13627
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.297135382903818e-06,
+      "loss": 0.1916,
+      "step": 13628
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.289751351704254e-06,
+      "loss": 0.1663,
+      "step": 13629
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2823715117349416e-06,
+      "loss": 0.1814,
+      "step": 13630
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.274995863325972e-06,
+      "loss": 0.1764,
+      "step": 13631
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.267624406807193e-06,
+      "loss": 0.185,
+      "step": 13632
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2602571425083415e-06,
+      "loss": 0.1819,
+      "step": 13633
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.252894070758885e-06,
+      "loss": 0.1822,
+      "step": 13634
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.245535191888163e-06,
+      "loss": 0.176,
+      "step": 13635
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.238180506225289e-06,
+      "loss": 0.1941,
+      "step": 13636
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.230830014099209e-06,
+      "loss": 0.1846,
+      "step": 13637
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.223483715838674e-06,
+      "loss": 0.1679,
+      "step": 13638
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.216141611772264e-06,
+      "loss": 0.1772,
+      "step": 13639
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.208803702228339e-06,
+      "loss": 0.174,
+      "step": 13640
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.201469987535113e-06,
+      "loss": 0.1797,
+      "step": 13641
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.1941404680205486e-06,
+      "loss": 0.1828,
+      "step": 13642
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.1868151440125146e-06,
+      "loss": 0.1699,
+      "step": 13643
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.179494015838594e-06,
+      "loss": 0.1769,
+      "step": 13644
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.1721770838262586e-06,
+      "loss": 0.184,
+      "step": 13645
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.1648643483027566e-06,
+      "loss": 0.1759,
+      "step": 13646
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.157555809595139e-06,
+      "loss": 0.1746,
+      "step": 13647
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.150251468030277e-06,
+      "loss": 0.1767,
+      "step": 13648
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.142951323934887e-06,
+      "loss": 0.1712,
+      "step": 13649
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.1356553776354405e-06,
+      "loss": 0.1817,
+      "step": 13650
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.1283636294582894e-06,
+      "loss": 0.1969,
+      "step": 13651
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.1210760797295155e-06,
+      "loss": 0.1762,
+      "step": 13652
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.113792728775103e-06,
+      "loss": 0.1787,
+      "step": 13653
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.106513576920758e-06,
+      "loss": 0.1722,
+      "step": 13654
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0992386244920875e-06,
+      "loss": 0.1788,
+      "step": 13655
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.09196787181443e-06,
+      "loss": 0.1871,
+      "step": 13656
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.084701319213004e-06,
+      "loss": 0.1829,
+      "step": 13657
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.077438967012783e-06,
+      "loss": 0.1792,
+      "step": 13658
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0701808155385955e-06,
+      "loss": 0.1763,
+      "step": 13659
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.06292686511506e-06,
+      "loss": 0.178,
+      "step": 13660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.055677116066616e-06,
+      "loss": 0.1775,
+      "step": 13661
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.048431568717483e-06,
+      "loss": 0.1742,
+      "step": 13662
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0411902233917795e-06,
+      "loss": 0.1773,
+      "step": 13663
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0339530804133345e-06,
+      "loss": 0.1901,
+      "step": 13664
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.026720140105824e-06,
+      "loss": 0.1796,
+      "step": 13665
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.019491402792788e-06,
+      "loss": 0.1712,
+      "step": 13666
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.012266868797501e-06,
+      "loss": 0.1714,
+      "step": 13667
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.005046538443071e-06,
+      "loss": 0.1701,
+      "step": 13668
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.997830412052474e-06,
+      "loss": 0.1754,
+      "step": 13669
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.990618489948408e-06,
+      "loss": 0.175,
+      "step": 13670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.983410772453479e-06,
+      "loss": 0.1767,
+      "step": 13671
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.9762072598900095e-06,
+      "loss": 0.184,
+      "step": 13672
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.969007952580208e-06,
+      "loss": 0.1821,
+      "step": 13673
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.961812850846038e-06,
+      "loss": 0.1847,
+      "step": 13674
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.954621955009343e-06,
+      "loss": 0.1754,
+      "step": 13675
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.947435265391721e-06,
+      "loss": 0.181,
+      "step": 13676
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.940252782314582e-06,
+      "loss": 0.1914,
+      "step": 13677
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.93307450609919e-06,
+      "loss": 0.1824,
+      "step": 13678
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.9259004370665786e-06,
+      "loss": 0.1821,
+      "step": 13679
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.918730575537634e-06,
+      "loss": 0.1825,
+      "step": 13680
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.911564921833001e-06,
+      "loss": 0.1775,
+      "step": 13681
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.90440347627319e-06,
+      "loss": 0.1808,
+      "step": 13682
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.897246239178478e-06,
+      "loss": 0.1792,
+      "step": 13683
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.890093210869008e-06,
+      "loss": 0.1714,
+      "step": 13684
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.882944391664669e-06,
+      "loss": 0.1837,
+      "step": 13685
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.875799781885227e-06,
+      "loss": 0.185,
+      "step": 13686
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.868659381850205e-06,
+      "loss": 0.1794,
+      "step": 13687
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.86152319187897e-06,
+      "loss": 0.1776,
+      "step": 13688
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.854391212290666e-06,
+      "loss": 0.1877,
+      "step": 13689
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.847263443404316e-06,
+      "loss": 0.1876,
+      "step": 13690
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.840139885538665e-06,
+      "loss": 0.1912,
+      "step": 13691
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.833020539012368e-06,
+      "loss": 0.1766,
+      "step": 13692
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.825905404143794e-06,
+      "loss": 0.1882,
+      "step": 13693
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.81879448125121e-06,
+      "loss": 0.1759,
+      "step": 13694
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.81168777065263e-06,
+      "loss": 0.1765,
+      "step": 13695
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.804585272665908e-06,
+      "loss": 0.1782,
+      "step": 13696
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.797486987608714e-06,
+      "loss": 0.1818,
+      "step": 13697
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.790392915798526e-06,
+      "loss": 0.1775,
+      "step": 13698
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.783303057552602e-06,
+      "loss": 0.1805,
+      "step": 13699
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.776217413188079e-06,
+      "loss": 0.1763,
+      "step": 13700
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.769135983021823e-06,
+      "loss": 0.1844,
+      "step": 13701
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7620587673705925e-06,
+      "loss": 0.1749,
+      "step": 13702
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.75498576655088e-06,
+      "loss": 0.1761,
+      "step": 13703
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.747916980879075e-06,
+      "loss": 0.1919,
+      "step": 13704
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.740852410671293e-06,
+      "loss": 0.1804,
+      "step": 13705
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.733792056243536e-06,
+      "loss": 0.1825,
+      "step": 13706
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.726735917911541e-06,
+      "loss": 0.1639,
+      "step": 13707
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.719683995990921e-06,
+      "loss": 0.1673,
+      "step": 13708
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.7126362907970575e-06,
+      "loss": 0.1764,
+      "step": 13709
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.7055928026451875e-06,
+      "loss": 0.1765,
+      "step": 13710
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.698553531850303e-06,
+      "loss": 0.182,
+      "step": 13711
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.691518478727276e-06,
+      "loss": 0.188,
+      "step": 13712
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.684487643590719e-06,
+      "loss": 0.19,
+      "step": 13713
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.677461026755116e-06,
+      "loss": 0.1769,
+      "step": 13714
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.670438628534713e-06,
+      "loss": 0.1788,
+      "step": 13715
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.663420449243606e-06,
+      "loss": 0.189,
+      "step": 13716
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6564064891956645e-06,
+      "loss": 0.1812,
+      "step": 13717
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.649396748704627e-06,
+      "loss": 0.1749,
+      "step": 13718
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.642391228083965e-06,
+      "loss": 0.1818,
+      "step": 13719
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.635389927647039e-06,
+      "loss": 0.1837,
+      "step": 13720
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.628392847706964e-06,
+      "loss": 0.1768,
+      "step": 13721
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.621399988576692e-06,
+      "loss": 0.1754,
+      "step": 13722
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.614411350568982e-06,
+      "loss": 0.1786,
+      "step": 13723
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.607426933996407e-06,
+      "loss": 0.1738,
+      "step": 13724
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.600446739171339e-06,
+      "loss": 0.177,
+      "step": 13725
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.593470766406006e-06,
+      "loss": 0.1705,
+      "step": 13726
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.586499016012348e-06,
+      "loss": 0.1835,
+      "step": 13727
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.579531488302236e-06,
+      "loss": 0.1826,
+      "step": 13728
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5725681835872455e-06,
+      "loss": 0.1813,
+      "step": 13729
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.56560910217887e-06,
+      "loss": 0.1789,
+      "step": 13730
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5586542443883064e-06,
+      "loss": 0.1799,
+      "step": 13731
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5517036105266486e-06,
+      "loss": 0.1724,
+      "step": 13732
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5447572009047375e-06,
+      "loss": 0.1841,
+      "step": 13733
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.537815015833292e-06,
+      "loss": 0.1912,
+      "step": 13734
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.530877055622763e-06,
+      "loss": 0.1718,
+      "step": 13735
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.52394332058348e-06,
+      "loss": 0.1856,
+      "step": 13736
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.517013811025551e-06,
+      "loss": 0.1749,
+      "step": 13737
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.510088527258905e-06,
+      "loss": 0.1923,
+      "step": 13738
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.5031674695932624e-06,
+      "loss": 0.1842,
+      "step": 13739
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.496250638338196e-06,
+      "loss": 0.187,
+      "step": 13740
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4893380338030485e-06,
+      "loss": 0.1787,
+      "step": 13741
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4824296562969725e-06,
+      "loss": 0.1812,
+      "step": 13742
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4755255061289865e-06,
+      "loss": 0.1726,
+      "step": 13743
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.468625583607834e-06,
+      "loss": 0.1844,
+      "step": 13744
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.461729889042166e-06,
+      "loss": 0.1829,
+      "step": 13745
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.454838422740383e-06,
+      "loss": 0.1762,
+      "step": 13746
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.447951185010669e-06,
+      "loss": 0.1771,
+      "step": 13747
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4410681761611125e-06,
+      "loss": 0.1822,
+      "step": 13748
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.434189396499523e-06,
+      "loss": 0.1771,
+      "step": 13749
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4273148463335755e-06,
+      "loss": 0.1851,
+      "step": 13750
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.420444525970714e-06,
+      "loss": 0.1805,
+      "step": 13751
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4135784357182585e-06,
+      "loss": 0.1759,
+      "step": 13752
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.406716575883264e-06,
+      "loss": 0.1781,
+      "step": 13753
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.399858946772618e-06,
+      "loss": 0.1765,
+      "step": 13754
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3930055486930755e-06,
+      "loss": 0.1736,
+      "step": 13755
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.386156381951113e-06,
+      "loss": 0.1753,
+      "step": 13756
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3793114468530966e-06,
+      "loss": 0.1729,
+      "step": 13757
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.372470743705149e-06,
+      "loss": 0.1818,
+      "step": 13758
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.365634272813236e-06,
+      "loss": 0.1759,
+      "step": 13759
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.358802034483102e-06,
+      "loss": 0.1886,
+      "step": 13760
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.351974029020346e-06,
+      "loss": 0.1779,
+      "step": 13761
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.345150256730336e-06,
+      "loss": 0.177,
+      "step": 13762
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3383307179182825e-06,
+      "loss": 0.1749,
+      "step": 13763
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.331515412889165e-06,
+      "loss": 0.178,
+      "step": 13764
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.32470434194784e-06,
+      "loss": 0.179,
+      "step": 13765
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.317897505398917e-06,
+      "loss": 0.1781,
+      "step": 13766
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3110949035468314e-06,
+      "loss": 0.1732,
+      "step": 13767
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.304296536695818e-06,
+      "loss": 0.176,
+      "step": 13768
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.297502405149968e-06,
+      "loss": 0.1776,
+      "step": 13769
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.290712509213114e-06,
+      "loss": 0.1775,
+      "step": 13770
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2839268491889805e-06,
+      "loss": 0.1703,
+      "step": 13771
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.277145425381025e-06,
+      "loss": 0.1732,
+      "step": 13772
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.270368238092582e-06,
+      "loss": 0.1777,
+      "step": 13773
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.263595287626722e-06,
+      "loss": 0.1774,
+      "step": 13774
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2568265742864e-06,
+      "loss": 0.1808,
+      "step": 13775
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.250062098374331e-06,
+      "loss": 0.1793,
+      "step": 13776
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2433018601930835e-06,
+      "loss": 0.1872,
+      "step": 13777
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.236545860044984e-06,
+      "loss": 0.1791,
+      "step": 13778
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2297940982322216e-06,
+      "loss": 0.1795,
+      "step": 13779
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.223046575056745e-06,
+      "loss": 0.1823,
+      "step": 13780
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.216303290820379e-06,
+      "loss": 0.1758,
+      "step": 13781
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.209564245824672e-06,
+      "loss": 0.1794,
+      "step": 13782
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.202829440371071e-06,
+      "loss": 0.1718,
+      "step": 13783
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.196098874760758e-06,
+      "loss": 0.1789,
+      "step": 13784
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.189372549294791e-06,
+      "loss": 0.1839,
+      "step": 13785
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.182650464274008e-06,
+      "loss": 0.1879,
+      "step": 13786
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.175932619999035e-06,
+      "loss": 0.1855,
+      "step": 13787
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.169219016770332e-06,
+      "loss": 0.1894,
+      "step": 13788
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.162509654888182e-06,
+      "loss": 0.1747,
+      "step": 13789
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.155804534652653e-06,
+      "loss": 0.1754,
+      "step": 13790
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.149103656363641e-06,
+      "loss": 0.1767,
+      "step": 13791
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.142407020320839e-06,
+      "loss": 0.174,
+      "step": 13792
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1357146268237736e-06,
+      "loss": 0.1691,
+      "step": 13793
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.129026476171739e-06,
+      "loss": 0.18,
+      "step": 13794
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.122342568663896e-06,
+      "loss": 0.1759,
+      "step": 13795
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.11566290459915e-06,
+      "loss": 0.1765,
+      "step": 13796
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.108987484276284e-06,
+      "loss": 0.1589,
+      "step": 13797
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1023163079938375e-06,
+      "loss": 0.1765,
+      "step": 13798
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.095649376050193e-06,
+      "loss": 0.1788,
+      "step": 13799
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0889866887435246e-06,
+      "loss": 0.1821,
+      "step": 13800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.082328246371837e-06,
+      "loss": 0.1767,
+      "step": 13801
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.075674049232914e-06,
+      "loss": 0.1846,
+      "step": 13802
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.069024097624387e-06,
+      "loss": 0.174,
+      "step": 13803
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0623783918436474e-06,
+      "loss": 0.166,
+      "step": 13804
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.055736932187971e-06,
+      "loss": 0.178,
+      "step": 13805
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.049099718954364e-06,
+      "loss": 0.1799,
+      "step": 13806
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.04246675243969e-06,
+      "loss": 0.1855,
+      "step": 13807
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.03583803294061e-06,
+      "loss": 0.1906,
+      "step": 13808
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0292135607535986e-06,
+      "loss": 0.1663,
+      "step": 13809
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.02259333617493e-06,
+      "loss": 0.1787,
+      "step": 13810
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.015977359500712e-06,
+      "loss": 0.174,
+      "step": 13811
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.00936563102683e-06,
+      "loss": 0.1814,
+      "step": 13812
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.002758151049014e-06,
+      "loss": 0.1808,
+      "step": 13813
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.996154919862761e-06,
+      "loss": 0.1793,
+      "step": 13814
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.989555937763435e-06,
+      "loss": 0.1739,
+      "step": 13815
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.982961205046155e-06,
+      "loss": 0.1863,
+      "step": 13816
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.976370722005885e-06,
+      "loss": 0.1772,
+      "step": 13817
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.969784488937379e-06,
+      "loss": 0.1729,
+      "step": 13818
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.963202506135223e-06,
+      "loss": 0.1871,
+      "step": 13819
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.956624773893792e-06,
+      "loss": 0.1734,
+      "step": 13820
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.950051292507252e-06,
+      "loss": 0.1775,
+      "step": 13821
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.943482062269644e-06,
+      "loss": 0.1756,
+      "step": 13822
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.936917083474757e-06,
+      "loss": 0.1806,
+      "step": 13823
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.930356356416222e-06,
+      "loss": 0.1817,
+      "step": 13824
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.923799881387458e-06,
+      "loss": 0.1626,
+      "step": 13825
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.917247658681723e-06,
+      "loss": 0.1733,
+      "step": 13826
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.910699688592068e-06,
+      "loss": 0.1817,
+      "step": 13827
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9041559714113286e-06,
+      "loss": 0.1774,
+      "step": 13828
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.897616507432201e-06,
+      "loss": 0.1768,
+      "step": 13829
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8910812969471535e-06,
+      "loss": 0.1772,
+      "step": 13830
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8845503402484835e-06,
+      "loss": 0.1832,
+      "step": 13831
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.878023637628292e-06,
+      "loss": 0.1686,
+      "step": 13832
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.871501189378469e-06,
+      "loss": 0.1642,
+      "step": 13833
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.864982995790756e-06,
+      "loss": 0.1869,
+      "step": 13834
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.858469057156667e-06,
+      "loss": 0.176,
+      "step": 13835
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.851959373767556e-06,
+      "loss": 0.1829,
+      "step": 13836
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.845453945914546e-06,
+      "loss": 0.1791,
+      "step": 13837
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.838952773888639e-06,
+      "loss": 0.1735,
+      "step": 13838
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.832455857980544e-06,
+      "loss": 0.1795,
+      "step": 13839
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8259631984808985e-06,
+      "loss": 0.1754,
+      "step": 13840
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.819474795680046e-06,
+      "loss": 0.1931,
+      "step": 13841
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.812990649868198e-06,
+      "loss": 0.1796,
+      "step": 13842
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.806510761335359e-06,
+      "loss": 0.1782,
+      "step": 13843
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.800035130371349e-06,
+      "loss": 0.1797,
+      "step": 13844
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.793563757265784e-06,
+      "loss": 0.1838,
+      "step": 13845
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.787096642308131e-06,
+      "loss": 0.1807,
+      "step": 13846
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.78063378578758e-06,
+      "loss": 0.1793,
+      "step": 13847
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.774175187993235e-06,
+      "loss": 0.1821,
+      "step": 13848
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.767720849213919e-06,
+      "loss": 0.1736,
+      "step": 13849
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7612707697383355e-06,
+      "loss": 0.1859,
+      "step": 13850
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.754824949854941e-06,
+      "loss": 0.1825,
+      "step": 13851
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7483833898520625e-06,
+      "loss": 0.1789,
+      "step": 13852
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.741946090017768e-06,
+      "loss": 0.1752,
+      "step": 13853
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.735513050639984e-06,
+      "loss": 0.1775,
+      "step": 13854
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.729084272006423e-06,
+      "loss": 0.1781,
+      "step": 13855
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7226597544046344e-06,
+      "loss": 0.1829,
+      "step": 13856
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.716239498121922e-06,
+      "loss": 0.1789,
+      "step": 13857
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.709823503445476e-06,
+      "loss": 0.1723,
+      "step": 13858
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.703411770662225e-06,
+      "loss": 0.1843,
+      "step": 13859
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.697004300058949e-06,
+      "loss": 0.1815,
+      "step": 13860
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.69060109192222e-06,
+      "loss": 0.1879,
+      "step": 13861
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.68420214653843e-06,
+      "loss": 0.1758,
+      "step": 13862
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.677807464193762e-06,
+      "loss": 0.1623,
+      "step": 13863
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.671417045174231e-06,
+      "loss": 0.1759,
+      "step": 13864
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.665030889765653e-06,
+      "loss": 0.1847,
+      "step": 13865
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6586489982536676e-06,
+      "loss": 0.1769,
+      "step": 13866
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.652271370923655e-06,
+      "loss": 0.1799,
+      "step": 13867
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.645898008060911e-06,
+      "loss": 0.1732,
+      "step": 13868
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.639528909950441e-06,
+      "loss": 0.1799,
+      "step": 13869
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.633164076877161e-06,
+      "loss": 0.1762,
+      "step": 13870
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.626803509125688e-06,
+      "loss": 0.177,
+      "step": 13871
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.620447206980538e-06,
+      "loss": 0.1805,
+      "step": 13872
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.614095170725963e-06,
+      "loss": 0.1717,
+      "step": 13873
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.607747400646101e-06,
+      "loss": 0.1871,
+      "step": 13874
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6014038970248364e-06,
+      "loss": 0.1762,
+      "step": 13875
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.595064660145887e-06,
+      "loss": 0.1775,
+      "step": 13876
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.588729690292781e-06,
+      "loss": 0.172,
+      "step": 13877
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5823989877488485e-06,
+      "loss": 0.1744,
+      "step": 13878
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.576072552797228e-06,
+      "loss": 0.1854,
+      "step": 13879
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.569750385720894e-06,
+      "loss": 0.1806,
+      "step": 13880
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.563432486802588e-06,
+      "loss": 0.1786,
+      "step": 13881
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.557118856324894e-06,
+      "loss": 0.1767,
+      "step": 13882
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.550809494570163e-06,
+      "loss": 0.1701,
+      "step": 13883
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.544504401820626e-06,
+      "loss": 0.1717,
+      "step": 13884
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.538203578358258e-06,
+      "loss": 0.178,
+      "step": 13885
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.531907024464865e-06,
+      "loss": 0.1648,
+      "step": 13886
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.525614740422046e-06,
+      "loss": 0.1742,
+      "step": 13887
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.519326726511275e-06,
+      "loss": 0.1767,
+      "step": 13888
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.513042983013727e-06,
+      "loss": 0.1872,
+      "step": 13889
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5067635102104985e-06,
+      "loss": 0.1764,
+      "step": 13890
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5004883083824e-06,
+      "loss": 0.1712,
+      "step": 13891
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4942173778101285e-06,
+      "loss": 0.1794,
+      "step": 13892
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.487950718774126e-06,
+      "loss": 0.1638,
+      "step": 13893
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.481688331554679e-06,
+      "loss": 0.1751,
+      "step": 13894
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.475430216431875e-06,
+      "loss": 0.181,
+      "step": 13895
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.469176373685624e-06,
+      "loss": 0.1731,
+      "step": 13896
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.462926803595624e-06,
+      "loss": 0.1809,
+      "step": 13897
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.456681506441373e-06,
+      "loss": 0.1783,
+      "step": 13898
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.450440482502227e-06,
+      "loss": 0.1813,
+      "step": 13899
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.444203732057284e-06,
+      "loss": 0.1827,
+      "step": 13900
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.437971255385509e-06,
+      "loss": 0.1755,
+      "step": 13901
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.431743052765636e-06,
+      "loss": 0.1775,
+      "step": 13902
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.425519124476241e-06,
+      "loss": 0.1756,
+      "step": 13903
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.419299470795668e-06,
+      "loss": 0.1794,
+      "step": 13904
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.413084092002129e-06,
+      "loss": 0.1797,
+      "step": 13905
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.406872988373589e-06,
+      "loss": 0.171,
+      "step": 13906
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.400666160187827e-06,
+      "loss": 0.1813,
+      "step": 13907
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3944636077224656e-06,
+      "loss": 0.1779,
+      "step": 13908
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.388265331254926e-06,
+      "loss": 0.183,
+      "step": 13909
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.382071331062399e-06,
+      "loss": 0.1733,
+      "step": 13910
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.375881607421939e-06,
+      "loss": 0.1809,
+      "step": 13911
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3696961606103596e-06,
+      "loss": 0.1866,
+      "step": 13912
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.363514990904338e-06,
+      "loss": 0.1781,
+      "step": 13913
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3573380985803104e-06,
+      "loss": 0.1702,
+      "step": 13914
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.351165483914554e-06,
+      "loss": 0.1815,
+      "step": 13915
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3449971471831165e-06,
+      "loss": 0.1715,
+      "step": 13916
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.338833088661909e-06,
+      "loss": 0.1851,
+      "step": 13917
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3326733086265894e-06,
+      "loss": 0.1851,
+      "step": 13918
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3265178073526926e-06,
+      "loss": 0.1676,
+      "step": 13919
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3203665851154984e-06,
+      "loss": 0.1829,
+      "step": 13920
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.314219642190143e-06,
+      "loss": 0.1886,
+      "step": 13921
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.308076978851528e-06,
+      "loss": 0.1745,
+      "step": 13922
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.30193859537441e-06,
+      "loss": 0.1723,
+      "step": 13923
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.295804492033317e-06,
+      "loss": 0.1832,
+      "step": 13924
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2896746691026054e-06,
+      "loss": 0.18,
+      "step": 13925
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2835491268564344e-06,
+      "loss": 0.1707,
+      "step": 13926
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.277427865568773e-06,
+      "loss": 0.1872,
+      "step": 13927
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2713108855133685e-06,
+      "loss": 0.1871,
+      "step": 13928
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.265198186963848e-06,
+      "loss": 0.1651,
+      "step": 13929
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2590897701935804e-06,
+      "loss": 0.1779,
+      "step": 13930
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.252985635475781e-06,
+      "loss": 0.1839,
+      "step": 13931
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.246885783083432e-06,
+      "loss": 0.1747,
+      "step": 13932
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2407902132893914e-06,
+      "loss": 0.1771,
+      "step": 13933
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.234698926366254e-06,
+      "loss": 0.178,
+      "step": 13934
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.228611922586478e-06,
+      "loss": 0.1712,
+      "step": 13935
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2225292022222806e-06,
+      "loss": 0.1708,
+      "step": 13936
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.216450765545743e-06,
+      "loss": 0.1745,
+      "step": 13937
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.210376612828704e-06,
+      "loss": 0.1865,
+      "step": 13938
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2043067443428565e-06,
+      "loss": 0.1757,
+      "step": 13939
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.198241160359639e-06,
+      "loss": 0.178,
+      "step": 13940
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1921798611503805e-06,
+      "loss": 0.1855,
+      "step": 13941
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.186122846986151e-06,
+      "loss": 0.1753,
+      "step": 13942
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.180070118137857e-06,
+      "loss": 0.1697,
+      "step": 13943
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1740216748762054e-06,
+      "loss": 0.1828,
+      "step": 13944
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.167977517471733e-06,
+      "loss": 0.1833,
+      "step": 13945
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.161937646194758e-06,
+      "loss": 0.1765,
+      "step": 13946
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.155902061315398e-06,
+      "loss": 0.1779,
+      "step": 13947
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.149870763103614e-06,
+      "loss": 0.1666,
+      "step": 13948
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.143843751829157e-06,
+      "loss": 0.1714,
+      "step": 13949
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.137821027761579e-06,
+      "loss": 0.1744,
+      "step": 13950
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.131802591170275e-06,
+      "loss": 0.175,
+      "step": 13951
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.125788442324385e-06,
+      "loss": 0.1876,
+      "step": 13952
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.119778581492928e-06,
+      "loss": 0.1739,
+      "step": 13953
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.113773008944666e-06,
+      "loss": 0.1777,
+      "step": 13954
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.107771724948228e-06,
+      "loss": 0.18,
+      "step": 13955
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1017747297720125e-06,
+      "loss": 0.1738,
+      "step": 13956
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.095782023684236e-06,
+      "loss": 0.1855,
+      "step": 13957
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.089793606952919e-06,
+      "loss": 0.1811,
+      "step": 13958
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.083809479845923e-06,
+      "loss": 0.164,
+      "step": 13959
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.077829642630859e-06,
+      "loss": 0.1764,
+      "step": 13960
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0718540955751895e-06,
+      "loss": 0.1709,
+      "step": 13961
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.065882838946189e-06,
+      "loss": 0.1753,
+      "step": 13962
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.059915873010889e-06,
+      "loss": 0.1751,
+      "step": 13963
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0539531980361974e-06,
+      "loss": 0.1807,
+      "step": 13964
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.04799481428878e-06,
+      "loss": 0.178,
+      "step": 13965
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.042040722035146e-06,
+      "loss": 0.1718,
+      "step": 13966
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.036090921541569e-06,
+      "loss": 0.1662,
+      "step": 13967
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.030145413074171e-06,
+      "loss": 0.1812,
+      "step": 13968
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.024204196898873e-06,
+      "loss": 0.185,
+      "step": 13969
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.018267273281373e-06,
+      "loss": 0.1796,
+      "step": 13970
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.012334642487237e-06,
+      "loss": 0.1817,
+      "step": 13971
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.006406304781785e-06,
+      "loss": 0.1791,
+      "step": 13972
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.00048226043016e-06,
+      "loss": 0.1757,
+      "step": 13973
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.994562509697342e-06,
+      "loss": 0.1769,
+      "step": 13974
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.98864705284806e-06,
+      "loss": 0.1742,
+      "step": 13975
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.982735890146905e-06,
+      "loss": 0.1861,
+      "step": 13976
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.976829021858253e-06,
+      "loss": 0.1747,
+      "step": 13977
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.970926448246303e-06,
+      "loss": 0.1788,
+      "step": 13978
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9650281695750136e-06,
+      "loss": 0.1785,
+      "step": 13979
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.959134186108237e-06,
+      "loss": 0.1836,
+      "step": 13980
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9532444981095425e-06,
+      "loss": 0.1748,
+      "step": 13981
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.947359105842385e-06,
+      "loss": 0.18,
+      "step": 13982
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9414780095699545e-06,
+      "loss": 0.171,
+      "step": 13983
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.935601209555317e-06,
+      "loss": 0.1752,
+      "step": 13984
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.929728706061298e-06,
+      "loss": 0.1715,
+      "step": 13985
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.923860499350551e-06,
+      "loss": 0.1697,
+      "step": 13986
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.917996589685535e-06,
+      "loss": 0.1737,
+      "step": 13987
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9121369773285175e-06,
+      "loss": 0.1756,
+      "step": 13988
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9062816625415555e-06,
+      "loss": 0.1767,
+      "step": 13989
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.900430645586561e-06,
+      "loss": 0.174,
+      "step": 13990
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.894583926725193e-06,
+      "loss": 0.1776,
+      "step": 13991
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.888741506218973e-06,
+      "loss": 0.1698,
+      "step": 13992
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.882903384329184e-06,
+      "loss": 0.1793,
+      "step": 13993
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8770695613169594e-06,
+      "loss": 0.1668,
+      "step": 13994
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.871240037443202e-06,
+      "loss": 0.1736,
+      "step": 13995
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8654148129686485e-06,
+      "loss": 0.175,
+      "step": 13996
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.859593888153834e-06,
+      "loss": 0.1774,
+      "step": 13997
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.853777263259106e-06,
+      "loss": 0.1708,
+      "step": 13998
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8479649385446014e-06,
+      "loss": 0.1783,
+      "step": 13999
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8421569142702895e-06,
+      "loss": 0.167,
+      "step": 14000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.836353190695929e-06,
+      "loss": 0.1657,
+      "step": 14001
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.830553768081113e-06,
+      "loss": 0.1826,
+      "step": 14002
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8247586466852e-06,
+      "loss": 0.1809,
+      "step": 14003
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.818967826767405e-06,
+      "loss": 0.1776,
+      "step": 14004
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8131813085866886e-06,
+      "loss": 0.1677,
+      "step": 14005
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8073990924018975e-06,
+      "loss": 0.167,
+      "step": 14006
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.801621178471604e-06,
+      "loss": 0.1706,
+      "step": 14007
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.795847567054256e-06,
+      "loss": 0.1767,
+      "step": 14008
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7900782584080586e-06,
+      "loss": 0.1698,
+      "step": 14009
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7843132527910717e-06,
+      "loss": 0.1821,
+      "step": 14010
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7785525504611118e-06,
+      "loss": 0.1802,
+      "step": 14011
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7727961516758504e-06,
+      "loss": 0.1662,
+      "step": 14012
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.767044056692726e-06,
+      "loss": 0.169,
+      "step": 14013
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.761296265769021e-06,
+      "loss": 0.1792,
+      "step": 14014
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7555527791617863e-06,
+      "loss": 0.1717,
+      "step": 14015
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7498135971279267e-06,
+      "loss": 0.1708,
+      "step": 14016
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7440787199241158e-06,
+      "loss": 0.1819,
+      "step": 14017
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7383481478068583e-06,
+      "loss": 0.1779,
+      "step": 14018
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7326218810324275e-06,
+      "loss": 0.1815,
+      "step": 14019
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7268999198569742e-06,
+      "loss": 0.1633,
+      "step": 14020
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7211822645363934e-06,
+      "loss": 0.1798,
+      "step": 14021
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7154689153264146e-06,
+      "loss": 0.1849,
+      "step": 14022
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7097598724825544e-06,
+      "loss": 0.1793,
+      "step": 14023
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.704055136260176e-06,
+      "loss": 0.1752,
+      "step": 14024
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6983547069143977e-06,
+      "loss": 0.172,
+      "step": 14025
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.692658584700226e-06,
+      "loss": 0.1811,
+      "step": 14026
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6869667698723576e-06,
+      "loss": 0.1732,
+      "step": 14027
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6812792626853998e-06,
+      "loss": 0.1802,
+      "step": 14028
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.675596063393716e-06,
+      "loss": 0.1719,
+      "step": 14029
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6699171722515025e-06,
+      "loss": 0.1945,
+      "step": 14030
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.664242589512723e-06,
+      "loss": 0.1795,
+      "step": 14031
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.658572315431208e-06,
+      "loss": 0.1712,
+      "step": 14032
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6529063502605433e-06,
+      "loss": 0.1753,
+      "step": 14033
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6472446942541596e-06,
+      "loss": 0.1794,
+      "step": 14034
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.641587347665243e-06,
+      "loss": 0.1879,
+      "step": 14035
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.635934310746858e-06,
+      "loss": 0.1685,
+      "step": 14036
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6302855837518134e-06,
+      "loss": 0.1801,
+      "step": 14037
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.624641166932763e-06,
+      "loss": 0.1857,
+      "step": 14038
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6190010605421485e-06,
+      "loss": 0.1713,
+      "step": 14039
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.613365264832247e-06,
+      "loss": 0.172,
+      "step": 14040
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6077337800551003e-06,
+      "loss": 0.182,
+      "step": 14041
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6021066064625743e-06,
+      "loss": 0.177,
+      "step": 14042
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.596483744306367e-06,
+      "loss": 0.1857,
+      "step": 14043
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5908651938379333e-06,
+      "loss": 0.1751,
+      "step": 14044
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5852509553086057e-06,
+      "loss": 0.1753,
+      "step": 14045
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5796410289694603e-06,
+      "loss": 0.1708,
+      "step": 14046
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.574035415071397e-06,
+      "loss": 0.1781,
+      "step": 14047
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.568434113865138e-06,
+      "loss": 0.1776,
+      "step": 14048
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.562837125601204e-06,
+      "loss": 0.1664,
+      "step": 14049
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5572444505299285e-06,
+      "loss": 0.1691,
+      "step": 14050
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.551656088901434e-06,
+      "loss": 0.1769,
+      "step": 14051
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5460720409656756e-06,
+      "loss": 0.1909,
+      "step": 14052
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.540492306972387e-06,
+      "loss": 0.176,
+      "step": 14053
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5349168871711356e-06,
+      "loss": 0.1755,
+      "step": 14054
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5293457818112773e-06,
+      "loss": 0.176,
+      "step": 14055
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5237789911419904e-06,
+      "loss": 0.1752,
+      "step": 14056
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.518216515412243e-06,
+      "loss": 0.1677,
+      "step": 14057
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5126583548708237e-06,
+      "loss": 0.1793,
+      "step": 14058
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5071045097663234e-06,
+      "loss": 0.1765,
+      "step": 14059
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5015549803471436e-06,
+      "loss": 0.1706,
+      "step": 14060
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4960097668614856e-06,
+      "loss": 0.1814,
+      "step": 14061
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4904688695573506e-06,
+      "loss": 0.1854,
+      "step": 14062
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4849322886825854e-06,
+      "loss": 0.1823,
+      "step": 14063
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.47940002448478e-06,
+      "loss": 0.1833,
+      "step": 14064
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4738720772114043e-06,
+      "loss": 0.1692,
+      "step": 14065
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.468348447109682e-06,
+      "loss": 0.1681,
+      "step": 14066
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4628291344266485e-06,
+      "loss": 0.1815,
+      "step": 14067
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4573141394091623e-06,
+      "loss": 0.1783,
+      "step": 14068
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.451803462303893e-06,
+      "loss": 0.1896,
+      "step": 14069
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.446297103357299e-06,
+      "loss": 0.1783,
+      "step": 14070
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4407950628156714e-06,
+      "loss": 0.1757,
+      "step": 14071
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4352973409250587e-06,
+      "loss": 0.1816,
+      "step": 14072
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4298039379313864e-06,
+      "loss": 0.1759,
+      "step": 14073
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.424314854080324e-06,
+      "loss": 0.1774,
+      "step": 14074
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.418830089617386e-06,
+      "loss": 0.1753,
+      "step": 14075
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4133496447878665e-06,
+      "loss": 0.1802,
+      "step": 14076
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4078735198368904e-06,
+      "loss": 0.1829,
+      "step": 14077
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4024017150093845e-06,
+      "loss": 0.1793,
+      "step": 14078
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.396934230550075e-06,
+      "loss": 0.1807,
+      "step": 14079
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3914710667034777e-06,
+      "loss": 0.1947,
+      "step": 14080
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3860122237139745e-06,
+      "loss": 0.17,
+      "step": 14081
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.380557701825671e-06,
+      "loss": 0.1778,
+      "step": 14082
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3751075012825593e-06,
+      "loss": 0.1814,
+      "step": 14083
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3696616223283684e-06,
+      "loss": 0.1869,
+      "step": 14084
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3642200652067025e-06,
+      "loss": 0.1806,
+      "step": 14085
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3587828301609224e-06,
+      "loss": 0.184,
+      "step": 14086
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.353349917434212e-06,
+      "loss": 0.1877,
+      "step": 14087
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.347921327269532e-06,
+      "loss": 0.1736,
+      "step": 14088
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3424970599097216e-06,
+      "loss": 0.1716,
+      "step": 14089
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3370771155973648e-06,
+      "loss": 0.1836,
+      "step": 14090
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.331661494574867e-06,
+      "loss": 0.1799,
+      "step": 14091
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.326250197084446e-06,
+      "loss": 0.1865,
+      "step": 14092
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.320843223368142e-06,
+      "loss": 0.1778,
+      "step": 14093
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3154405736677497e-06,
+      "loss": 0.1786,
+      "step": 14094
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3100422482249317e-06,
+      "loss": 0.1819,
+      "step": 14095
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.304648247281128e-06,
+      "loss": 0.1746,
+      "step": 14096
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.299258571077579e-06,
+      "loss": 0.1873,
+      "step": 14097
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2938732198553367e-06,
+      "loss": 0.1834,
+      "step": 14098
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.288492193855286e-06,
+      "loss": 0.1739,
+      "step": 14099
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2831154933180674e-06,
+      "loss": 0.1901,
+      "step": 14100
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.277743118484178e-06,
+      "loss": 0.1633,
+      "step": 14101
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2723750695938805e-06,
+      "loss": 0.1713,
+      "step": 14102
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.267011346887283e-06,
+      "loss": 0.1718,
+      "step": 14103
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2616519506042496e-06,
+      "loss": 0.1776,
+      "step": 14104
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.256296880984522e-06,
+      "loss": 0.1717,
+      "step": 14105
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2509461382675967e-06,
+      "loss": 0.1732,
+      "step": 14106
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2455997226927602e-06,
+      "loss": 0.1844,
+      "step": 14107
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2402576344991552e-06,
+      "loss": 0.1605,
+      "step": 14108
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2349198739257127e-06,
+      "loss": 0.177,
+      "step": 14109
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.229586441211152e-06,
+      "loss": 0.1717,
+      "step": 14110
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2242573365940276e-06,
+      "loss": 0.1759,
+      "step": 14111
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2189325603126595e-06,
+      "loss": 0.1578,
+      "step": 14112
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2136121126052465e-06,
+      "loss": 0.1757,
+      "step": 14113
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.208295993709698e-06,
+      "loss": 0.177,
+      "step": 14114
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2029842038638237e-06,
+      "loss": 0.1794,
+      "step": 14115
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.197676743305156e-06,
+      "loss": 0.1736,
+      "step": 14116
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1923736122711045e-06,
+      "loss": 0.1769,
+      "step": 14117
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.187074810998847e-06,
+      "loss": 0.1769,
+      "step": 14118
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1817803397253487e-06,
+      "loss": 0.1849,
+      "step": 14119
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.176490198687454e-06,
+      "loss": 0.1767,
+      "step": 14120
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.171204388121718e-06,
+      "loss": 0.1775,
+      "step": 14121
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.165922908264585e-06,
+      "loss": 0.1788,
+      "step": 14122
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1606457593522432e-06,
+      "loss": 0.1726,
+      "step": 14123
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.155372941620749e-06,
+      "loss": 0.1823,
+      "step": 14124
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1501044553059024e-06,
+      "loss": 0.1733,
+      "step": 14125
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1448403006433702e-06,
+      "loss": 0.1811,
+      "step": 14126
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1395804778685643e-06,
+      "loss": 0.182,
+      "step": 14127
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1343249872167302e-06,
+      "loss": 0.1764,
+      "step": 14128
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.129073828922946e-06,
+      "loss": 0.1759,
+      "step": 14129
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.123827003222057e-06,
+      "loss": 0.1803,
+      "step": 14130
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.118584510348721e-06,
+      "loss": 0.1913,
+      "step": 14131
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1133463505374383e-06,
+      "loss": 0.1857,
+      "step": 14132
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1081125240224663e-06,
+      "loss": 0.1832,
+      "step": 14133
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1028830310378953e-06,
+      "loss": 0.1732,
+      "step": 14134
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.097657871817605e-06,
+      "loss": 0.1813,
+      "step": 14135
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0924370465953313e-06,
+      "loss": 0.1839,
+      "step": 14136
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0872205556045198e-06,
+      "loss": 0.1736,
+      "step": 14137
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.08200839907854e-06,
+      "loss": 0.168,
+      "step": 14138
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0768005772504607e-06,
+      "loss": 0.1775,
+      "step": 14139
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0715970903532286e-06,
+      "loss": 0.1696,
+      "step": 14140
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0663979386195584e-06,
+      "loss": 0.1738,
+      "step": 14141
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0612031222820083e-06,
+      "loss": 0.1863,
+      "step": 14142
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0560126415728806e-06,
+      "loss": 0.1759,
+      "step": 14143
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0508264967243682e-06,
+      "loss": 0.1739,
+      "step": 14144
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0456446879683743e-06,
+      "loss": 0.1692,
+      "step": 14145
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0404672155367132e-06,
+      "loss": 0.1739,
+      "step": 14146
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0352940796608888e-06,
+      "loss": 0.1759,
+      "step": 14147
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.030125280572316e-06,
+      "loss": 0.1743,
+      "step": 14148
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.024960818502143e-06,
+      "loss": 0.1728,
+      "step": 14149
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.019800693681385e-06,
+      "loss": 0.1783,
+      "step": 14150
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.014644906340791e-06,
+      "loss": 0.1649,
+      "step": 14151
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0094934567109987e-06,
+      "loss": 0.1661,
+      "step": 14152
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0043463450223685e-06,
+      "loss": 0.1663,
+      "step": 14153
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.999203571505138e-06,
+      "loss": 0.1865,
+      "step": 14154
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.99406513638929e-06,
+      "loss": 0.1761,
+      "step": 14155
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9889310399046743e-06,
+      "loss": 0.1778,
+      "step": 14156
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.983801282280896e-06,
+      "loss": 0.1828,
+      "step": 14157
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9786758637474043e-06,
+      "loss": 0.1764,
+      "step": 14158
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9735547845334055e-06,
+      "loss": 0.1831,
+      "step": 14159
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9684380448679718e-06,
+      "loss": 0.1877,
+      "step": 14160
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9633256449799306e-06,
+      "loss": 0.1728,
+      "step": 14161
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9582175850979556e-06,
+      "loss": 0.1682,
+      "step": 14162
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.953113865450474e-06,
+      "loss": 0.1746,
+      "step": 14163
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9480144862657932e-06,
+      "loss": 0.1786,
+      "step": 14164
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9429194477719635e-06,
+      "loss": 0.1777,
+      "step": 14165
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9378287501968695e-06,
+      "loss": 0.1731,
+      "step": 14166
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.932742393768184e-06,
+      "loss": 0.1697,
+      "step": 14167
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9276603787134038e-06,
+      "loss": 0.1746,
+      "step": 14168
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.922582705259813e-06,
+      "loss": 0.1758,
+      "step": 14169
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9175093736345416e-06,
+      "loss": 0.1738,
+      "step": 14170
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9124403840644522e-06,
+      "loss": 0.168,
+      "step": 14171
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9073757367763078e-06,
+      "loss": 0.1797,
+      "step": 14172
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.902315431996583e-06,
+      "loss": 0.1774,
+      "step": 14173
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8972594699516297e-06,
+      "loss": 0.1837,
+      "step": 14174
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8922078508675675e-06,
+      "loss": 0.1676,
+      "step": 14175
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.887160574970338e-06,
+      "loss": 0.177,
+      "step": 14176
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.88211764248566e-06,
+      "loss": 0.19,
+      "step": 14177
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.87707905363912e-06,
+      "loss": 0.1792,
+      "step": 14178
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.872044808656038e-06,
+      "loss": 0.1794,
+      "step": 14179
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8670149077616006e-06,
+      "loss": 0.1851,
+      "step": 14180
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8619893511807385e-06,
+      "loss": 0.1809,
+      "step": 14181
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.85696813913825e-06,
+      "loss": 0.1807,
+      "step": 14182
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8519512718586995e-06,
+      "loss": 0.1753,
+      "step": 14183
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8469387495664745e-06,
+      "loss": 0.1752,
+      "step": 14184
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8419305724857516e-06,
+      "loss": 0.1696,
+      "step": 14185
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.83692674084054e-06,
+      "loss": 0.1793,
+      "step": 14186
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.831927254854616e-06,
+      "loss": 0.1786,
+      "step": 14187
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8269321147516014e-06,
+      "loss": 0.1724,
+      "step": 14188
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.821941320754884e-06,
+      "loss": 0.1863,
+      "step": 14189
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8169548730877183e-06,
+      "loss": 0.176,
+      "step": 14190
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8119727719730816e-06,
+      "loss": 0.1831,
+      "step": 14191
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.80699501763384e-06,
+      "loss": 0.1768,
+      "step": 14192
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8020216102925934e-06,
+      "loss": 0.1797,
+      "step": 14193
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.797052550171797e-06,
+      "loss": 0.1719,
+      "step": 14194
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7920878374936844e-06,
+      "loss": 0.1767,
+      "step": 14195
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.787127472480322e-06,
+      "loss": 0.176,
+      "step": 14196
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.782171455353555e-06,
+      "loss": 0.1765,
+      "step": 14197
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.777219786335017e-06,
+      "loss": 0.1708,
+      "step": 14198
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7722724656462197e-06,
+      "loss": 0.1746,
+      "step": 14199
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.767329493508397e-06,
+      "loss": 0.1788,
+      "step": 14200
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7623908701426502e-06,
+      "loss": 0.1749,
+      "step": 14201
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.757456595769836e-06,
+      "loss": 0.1781,
+      "step": 14202
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.752526670610678e-06,
+      "loss": 0.181,
+      "step": 14203
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7476010948856324e-06,
+      "loss": 0.1825,
+      "step": 14204
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7426798688150234e-06,
+      "loss": 0.1799,
+      "step": 14205
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7377629926189417e-06,
+      "loss": 0.1806,
+      "step": 14206
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.732850466517289e-06,
+      "loss": 0.1713,
+      "step": 14207
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7279422907298015e-06,
+      "loss": 0.164,
+      "step": 14208
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.723038465475991e-06,
+      "loss": 0.1892,
+      "step": 14209
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.718138990975172e-06,
+      "loss": 0.1805,
+      "step": 14210
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.713243867446491e-06,
+      "loss": 0.1846,
+      "step": 14211
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.708353095108873e-06,
+      "loss": 0.1682,
+      "step": 14212
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7034666741810656e-06,
+      "loss": 0.181,
+      "step": 14213
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6985846048816155e-06,
+      "loss": 0.1795,
+      "step": 14214
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.693706887428882e-06,
+      "loss": 0.1668,
+      "step": 14215
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.688833522041001e-06,
+      "loss": 0.1734,
+      "step": 14216
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6839645089359655e-06,
+      "loss": 0.1792,
+      "step": 14217
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6790998483315234e-06,
+      "loss": 0.1748,
+      "step": 14218
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.674239540445256e-06,
+      "loss": 0.1784,
+      "step": 14219
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6693835854945337e-06,
+      "loss": 0.1757,
+      "step": 14220
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.664531983696561e-06,
+      "loss": 0.1754,
+      "step": 14221
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6596847352683086e-06,
+      "loss": 0.1786,
+      "step": 14222
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6548418404265807e-06,
+      "loss": 0.1826,
+      "step": 14223
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.650003299387971e-06,
+      "loss": 0.1697,
+      "step": 14224
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.645169112368895e-06,
+      "loss": 0.1709,
+      "step": 14225
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6403392795855685e-06,
+      "loss": 0.1601,
+      "step": 14226
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6355138012539858e-06,
+      "loss": 0.1771,
+      "step": 14227
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6306926775899852e-06,
+      "loss": 0.1618,
+      "step": 14228
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6258759088091833e-06,
+      "loss": 0.1693,
+      "step": 14229
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6210634951270185e-06,
+      "loss": 0.1803,
+      "step": 14230
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.616255436758741e-06,
+      "loss": 0.1711,
+      "step": 14231
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.611451733919357e-06,
+      "loss": 0.1796,
+      "step": 14232
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.60665238682376e-06,
+      "loss": 0.1727,
+      "step": 14233
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6018573956865576e-06,
+      "loss": 0.1705,
+      "step": 14234
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5970667607222445e-06,
+      "loss": 0.1821,
+      "step": 14235
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5922804821450598e-06,
+      "loss": 0.1826,
+      "step": 14236
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.587498560169099e-06,
+      "loss": 0.1772,
+      "step": 14237
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5827209950082034e-06,
+      "loss": 0.1792,
+      "step": 14238
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.577947786876067e-06,
+      "loss": 0.1694,
+      "step": 14239
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5731789359861646e-06,
+      "loss": 0.1785,
+      "step": 14240
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5684144425518144e-06,
+      "loss": 0.1752,
+      "step": 14241
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5636543067860675e-06,
+      "loss": 0.1882,
+      "step": 14242
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.558898528901865e-06,
+      "loss": 0.1709,
+      "step": 14243
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5541471091118705e-06,
+      "loss": 0.1765,
+      "step": 14244
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.549400047628625e-06,
+      "loss": 0.1648,
+      "step": 14245
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.544657344664436e-06,
+      "loss": 0.1704,
+      "step": 14246
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.539919000431412e-06,
+      "loss": 0.1753,
+      "step": 14247
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5351850151414836e-06,
+      "loss": 0.1848,
+      "step": 14248
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5304553890063807e-06,
+      "loss": 0.1801,
+      "step": 14249
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.525730122237624e-06,
+      "loss": 0.1831,
+      "step": 14250
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5210092150465882e-06,
+      "loss": 0.1758,
+      "step": 14251
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.516292667644371e-06,
+      "loss": 0.1799,
+      "step": 14252
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5115804802419708e-06,
+      "loss": 0.1743,
+      "step": 14253
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5068726530500964e-06,
+      "loss": 0.173,
+      "step": 14254
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.502169186279346e-06,
+      "loss": 0.1928,
+      "step": 14255
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.497470080140063e-06,
+      "loss": 0.1793,
+      "step": 14256
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.492775334842423e-06,
+      "loss": 0.1826,
+      "step": 14257
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4880849505963923e-06,
+      "loss": 0.1769,
+      "step": 14258
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4833989276117575e-06,
+      "loss": 0.1706,
+      "step": 14259
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4787172660981074e-06,
+      "loss": 0.1799,
+      "step": 14260
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4740399662648294e-06,
+      "loss": 0.1893,
+      "step": 14261
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.469367028321112e-06,
+      "loss": 0.1745,
+      "step": 14262
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.464698452475944e-06,
+      "loss": 0.1772,
+      "step": 14263
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.460034238938158e-06,
+      "loss": 0.1757,
+      "step": 14264
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4553743879163314e-06,
+      "loss": 0.1864,
+      "step": 14265
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.450718899618909e-06,
+      "loss": 0.1804,
+      "step": 14266
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4460677742540904e-06,
+      "loss": 0.1663,
+      "step": 14267
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4414210120298986e-06,
+      "loss": 0.1802,
+      "step": 14268
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.436778613154167e-06,
+      "loss": 0.1643,
+      "step": 14269
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4321405778345186e-06,
+      "loss": 0.1662,
+      "step": 14270
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4275069062784205e-06,
+      "loss": 0.1776,
+      "step": 14271
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.422877598693085e-06,
+      "loss": 0.1806,
+      "step": 14272
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4182526552855688e-06,
+      "loss": 0.1687,
+      "step": 14273
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4136320762627394e-06,
+      "loss": 0.1737,
+      "step": 14274
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.409015861831221e-06,
+      "loss": 0.1733,
+      "step": 14275
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.404404012197514e-06,
+      "loss": 0.1685,
+      "step": 14276
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.399796527567855e-06,
+      "loss": 0.1781,
+      "step": 14277
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.395193408148333e-06,
+      "loss": 0.1763,
+      "step": 14278
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3905946541448176e-06,
+      "loss": 0.1741,
+      "step": 14279
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3860002657629888e-06,
+      "loss": 0.1808,
+      "step": 14280
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3814102432083373e-06,
+      "loss": 0.1782,
+      "step": 14281
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.376824586686155e-06,
+      "loss": 0.1738,
+      "step": 14282
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.372243296401533e-06,
+      "loss": 0.186,
+      "step": 14283
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.367666372559374e-06,
+      "loss": 0.1803,
+      "step": 14284
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3630938153643813e-06,
+      "loss": 0.1711,
+      "step": 14285
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3585256250210797e-06,
+      "loss": 0.1795,
+      "step": 14286
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3539618017337394e-06,
+      "loss": 0.1835,
+      "step": 14287
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3494023457065308e-06,
+      "loss": 0.1877,
+      "step": 14288
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3448472571433343e-06,
+      "loss": 0.1768,
+      "step": 14289
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.340296536247921e-06,
+      "loss": 0.1775,
+      "step": 14290
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.335750183223784e-06,
+      "loss": 0.1889,
+      "step": 14291
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3312081982742927e-06,
+      "loss": 0.1709,
+      "step": 14292
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3266705816025636e-06,
+      "loss": 0.1694,
+      "step": 14293
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.322137333411567e-06,
+      "loss": 0.1883,
+      "step": 14294
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.31760845390403e-06,
+      "loss": 0.1672,
+      "step": 14295
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3130839432825234e-06,
+      "loss": 0.1751,
+      "step": 14296
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3085638017494083e-06,
+      "loss": 0.1863,
+      "step": 14297
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3040480295068556e-06,
+      "loss": 0.1826,
+      "step": 14298
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2995366267568043e-06,
+      "loss": 0.177,
+      "step": 14299
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2950295937010703e-06,
+      "loss": 0.177,
+      "step": 14300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2905269305412036e-06,
+      "loss": 0.18,
+      "step": 14301
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.28602863747861e-06,
+      "loss": 0.176,
+      "step": 14302
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2815347147144507e-06,
+      "loss": 0.176,
+      "step": 14303
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2770451624497424e-06,
+      "loss": 0.1717,
+      "step": 14304
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2725599808852694e-06,
+      "loss": 0.1686,
+      "step": 14305
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2680791702216486e-06,
+      "loss": 0.177,
+      "step": 14306
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.263602730659253e-06,
+      "loss": 0.172,
+      "step": 14307
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2591306623983343e-06,
+      "loss": 0.1721,
+      "step": 14308
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2546629656388653e-06,
+      "loss": 0.1835,
+      "step": 14309
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.250199640580708e-06,
+      "loss": 0.1802,
+      "step": 14310
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2457406874234477e-06,
+      "loss": 0.1817,
+      "step": 14311
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.241286106366547e-06,
+      "loss": 0.1784,
+      "step": 14312
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2368358976092128e-06,
+      "loss": 0.1743,
+      "step": 14313
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.232390061350498e-06,
+      "loss": 0.1764,
+      "step": 14314
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2279485977892313e-06,
+      "loss": 0.1801,
+      "step": 14315
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2235115071240762e-06,
+      "loss": 0.172,
+      "step": 14316
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2190787895534747e-06,
+      "loss": 0.1741,
+      "step": 14317
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.214650445275679e-06,
+      "loss": 0.174,
+      "step": 14318
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2102264744887525e-06,
+      "loss": 0.1718,
+      "step": 14319
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2058068773905592e-06,
+      "loss": 0.1786,
+      "step": 14320
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.201391654178764e-06,
+      "loss": 0.1664,
+      "step": 14321
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.196980805050852e-06,
+      "loss": 0.1716,
+      "step": 14322
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.192574330204078e-06,
+      "loss": 0.1699,
+      "step": 14323
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1881722298355387e-06,
+      "loss": 0.1777,
+      "step": 14324
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.183774504142111e-06,
+      "loss": 0.1771,
+      "step": 14325
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1793811533205143e-06,
+      "loss": 0.177,
+      "step": 14326
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.174992177567203e-06,
+      "loss": 0.1711,
+      "step": 14327
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1706075770784982e-06,
+      "loss": 0.1724,
+      "step": 14328
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.166227352050487e-06,
+      "loss": 0.1848,
+      "step": 14329
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.161851502679102e-06,
+      "loss": 0.1711,
+      "step": 14330
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1574800291600193e-06,
+      "loss": 0.1794,
+      "step": 14331
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1531129316887943e-06,
+      "loss": 0.1754,
+      "step": 14332
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.148750210460715e-06,
+      "loss": 0.1768,
+      "step": 14333
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1443918656709363e-06,
+      "loss": 0.1732,
+      "step": 14334
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1400378975143578e-06,
+      "loss": 0.1722,
+      "step": 14335
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1356883061857347e-06,
+      "loss": 0.1804,
+      "step": 14336
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1313430918795785e-06,
+      "loss": 0.1744,
+      "step": 14337
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.127002254790267e-06,
+      "loss": 0.1748,
+      "step": 14338
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.122665795111911e-06,
+      "loss": 0.1741,
+      "step": 14339
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1183337130384785e-06,
+      "loss": 0.1715,
+      "step": 14340
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.114006008763725e-06,
+      "loss": 0.1742,
+      "step": 14341
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1096826824811956e-06,
+      "loss": 0.1689,
+      "step": 14342
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1053637343842804e-06,
+      "loss": 0.1697,
+      "step": 14343
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1010491646661024e-06,
+      "loss": 0.1849,
+      "step": 14344
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.096738973519685e-06,
+      "loss": 0.1741,
+      "step": 14345
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0924331611377635e-06,
+      "loss": 0.1643,
+      "step": 14346
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0881317277129163e-06,
+      "loss": 0.1722,
+      "step": 14347
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.083834673437557e-06,
+      "loss": 0.1694,
+      "step": 14348
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.079541998503853e-06,
+      "loss": 0.1745,
+      "step": 14349
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.075253703103797e-06,
+      "loss": 0.1801,
+      "step": 14350
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.070969787429189e-06,
+      "loss": 0.1744,
+      "step": 14351
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0666902516716213e-06,
+      "loss": 0.1782,
+      "step": 14352
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0624150960225184e-06,
+      "loss": 0.1737,
+      "step": 14353
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.058144320673061e-06,
+      "loss": 0.1804,
+      "step": 14354
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0538779258142737e-06,
+      "loss": 0.1623,
+      "step": 14355
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0496159116369705e-06,
+      "loss": 0.1819,
+      "step": 14356
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0453582783317883e-06,
+      "loss": 0.1802,
+      "step": 14357
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0411050260891295e-06,
+      "loss": 0.1722,
+      "step": 14358
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0368561550992314e-06,
+      "loss": 0.1738,
+      "step": 14359
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.03261166555212e-06,
+      "loss": 0.1757,
+      "step": 14360
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0283715576376426e-06,
+      "loss": 0.1677,
+      "step": 14361
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0241358315454374e-06,
+      "loss": 0.1874,
+      "step": 14362
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.019904487464952e-06,
+      "loss": 0.1872,
+      "step": 14363
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0156775255854134e-06,
+      "loss": 0.1708,
+      "step": 14364
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.011454946095903e-06,
+      "loss": 0.1804,
+      "step": 14365
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0072367491852705e-06,
+      "loss": 0.1738,
+      "step": 14366
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.003022935042165e-06,
+      "loss": 0.1656,
+      "step": 14367
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.998813503855046e-06,
+      "loss": 0.1838,
+      "step": 14368
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.994608455812208e-06,
+      "loss": 0.1817,
+      "step": 14369
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9904077911017003e-06,
+      "loss": 0.1764,
+      "step": 14370
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9862115099114175e-06,
+      "loss": 0.1748,
+      "step": 14371
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9820196124290093e-06,
+      "loss": 0.1737,
+      "step": 14372
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.977832098842003e-06,
+      "loss": 0.1876,
+      "step": 14373
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.973648969337649e-06,
+      "loss": 0.1616,
+      "step": 14374
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9694702241030762e-06,
+      "loss": 0.1769,
+      "step": 14375
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.965295863325134e-06,
+      "loss": 0.1669,
+      "step": 14376
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9611258871905737e-06,
+      "loss": 0.1728,
+      "step": 14377
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.956960295885857e-06,
+      "loss": 0.1745,
+      "step": 14378
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9527990895973125e-06,
+      "loss": 0.1713,
+      "step": 14379
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9486422685110475e-06,
+      "loss": 0.1706,
+      "step": 14380
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.94448983281299e-06,
+      "loss": 0.1767,
+      "step": 14381
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.940341782688826e-06,
+      "loss": 0.1819,
+      "step": 14382
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9361981183241175e-06,
+      "loss": 0.1854,
+      "step": 14383
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9320588399041716e-06,
+      "loss": 0.1769,
+      "step": 14384
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9279239476141187e-06,
+      "loss": 0.1844,
+      "step": 14385
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.92379344163891e-06,
+      "loss": 0.1829,
+      "step": 14386
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9196673221632655e-06,
+      "loss": 0.1718,
+      "step": 14387
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9155455893717257e-06,
+      "loss": 0.1758,
+      "step": 14388
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.911428243448643e-06,
+      "loss": 0.1753,
+      "step": 14389
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9073152845781706e-06,
+      "loss": 0.1773,
+      "step": 14390
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9032067129442722e-06,
+      "loss": 0.1772,
+      "step": 14391
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8991025287306786e-06,
+      "loss": 0.1776,
+      "step": 14392
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8950027321209762e-06,
+      "loss": 0.1746,
+      "step": 14393
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8909073232985185e-06,
+      "loss": 0.1802,
+      "step": 14394
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8868163024464814e-06,
+      "loss": 0.178,
+      "step": 14395
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.882729669747818e-06,
+      "loss": 0.1755,
+      "step": 14396
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.878647425385327e-06,
+      "loss": 0.1862,
+      "step": 14397
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8745695695415845e-06,
+      "loss": 0.1681,
+      "step": 14398
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8704961023989665e-06,
+      "loss": 0.1759,
+      "step": 14399
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.866427024139661e-06,
+      "loss": 0.172,
+      "step": 14400
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8623623349456664e-06,
+      "loss": 0.1826,
+      "step": 14401
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8583020349987713e-06,
+      "loss": 0.1823,
+      "step": 14402
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8542461244805853e-06,
+      "loss": 0.1886,
+      "step": 14403
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8501946035724972e-06,
+      "loss": 0.1885,
+      "step": 14404
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.846147472455717e-06,
+      "loss": 0.1808,
+      "step": 14405
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.842104731311267e-06,
+      "loss": 0.1749,
+      "step": 14406
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.838066380319936e-06,
+      "loss": 0.1704,
+      "step": 14407
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.834032419662357e-06,
+      "loss": 0.1802,
+      "step": 14408
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8300028495189524e-06,
+      "loss": 0.1765,
+      "step": 14409
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8259776700699226e-06,
+      "loss": 0.1724,
+      "step": 14410
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8219568814953348e-06,
+      "loss": 0.1755,
+      "step": 14411
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8179404839749891e-06,
+      "loss": 0.1671,
+      "step": 14412
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8139284776885423e-06,
+      "loss": 0.1741,
+      "step": 14413
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8099208628154063e-06,
+      "loss": 0.169,
+      "step": 14414
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8059176395348487e-06,
+      "loss": 0.1721,
+      "step": 14415
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8019188080259043e-06,
+      "loss": 0.178,
+      "step": 14416
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7979243684674297e-06,
+      "loss": 0.1761,
+      "step": 14417
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.793934321038071e-06,
+      "loss": 0.1741,
+      "step": 14418
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7899486659162856e-06,
+      "loss": 0.1759,
+      "step": 14419
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7859674032803308e-06,
+      "loss": 0.1645,
+      "step": 14420
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7819905333082642e-06,
+      "loss": 0.1873,
+      "step": 14421
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7780180561779769e-06,
+      "loss": 0.1753,
+      "step": 14422
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7740499720671156e-06,
+      "loss": 0.185,
+      "step": 14423
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7700862811531826e-06,
+      "loss": 0.1782,
+      "step": 14424
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7661269836134141e-06,
+      "loss": 0.1785,
+      "step": 14425
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.762172079624924e-06,
+      "loss": 0.1642,
+      "step": 14426
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7582215693645932e-06,
+      "loss": 0.1778,
+      "step": 14427
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.754275453009091e-06,
+      "loss": 0.1724,
+      "step": 14428
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7503337307349321e-06,
+      "loss": 0.1675,
+      "step": 14429
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7463964027184089e-06,
+      "loss": 0.1804,
+      "step": 14430
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7424634691355913e-06,
+      "loss": 0.1772,
+      "step": 14431
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7385349301624165e-06,
+      "loss": 0.1753,
+      "step": 14432
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7346107859745774e-06,
+      "loss": 0.1788,
+      "step": 14433
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7306910367475783e-06,
+      "loss": 0.174,
+      "step": 14434
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7267756826567339e-06,
+      "loss": 0.1722,
+      "step": 14435
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7228647238771711e-06,
+      "loss": 0.1796,
+      "step": 14436
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7189581605837834e-06,
+      "loss": 0.1691,
+      "step": 14437
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.71505599295132e-06,
+      "loss": 0.1757,
+      "step": 14438
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7111582211542966e-06,
+      "loss": 0.1857,
+      "step": 14439
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7072648453670403e-06,
+      "loss": 0.1722,
+      "step": 14440
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.70337586576369e-06,
+      "loss": 0.1645,
+      "step": 14441
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.699491282518173e-06,
+      "loss": 0.1729,
+      "step": 14442
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6956110958042392e-06,
+      "loss": 0.1766,
+      "step": 14443
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6917353057954278e-06,
+      "loss": 0.1713,
+      "step": 14444
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6878639126650775e-06,
+      "loss": 0.1663,
+      "step": 14445
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6839969165863612e-06,
+      "loss": 0.1745,
+      "step": 14446
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.680134317732196e-06,
+      "loss": 0.1821,
+      "step": 14447
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.676276116275377e-06,
+      "loss": 0.1761,
+      "step": 14448
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.672422312388422e-06,
+      "loss": 0.1768,
+      "step": 14449
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.668572906243726e-06,
+      "loss": 0.1862,
+      "step": 14450
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6647278980134518e-06,
+      "loss": 0.1822,
+      "step": 14451
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6608872878695615e-06,
+      "loss": 0.1784,
+      "step": 14452
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6570510759838176e-06,
+      "loss": 0.1806,
+      "step": 14453
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6532192625278164e-06,
+      "loss": 0.1882,
+      "step": 14454
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6493918476729319e-06,
+      "loss": 0.173,
+      "step": 14455
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6455688315903384e-06,
+      "loss": 0.1758,
+      "step": 14456
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6417502144510323e-06,
+      "loss": 0.1829,
+      "step": 14457
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.637935996425799e-06,
+      "loss": 0.1701,
+      "step": 14458
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6341261776852246e-06,
+      "loss": 0.1752,
+      "step": 14459
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6303207583997172e-06,
+      "loss": 0.172,
+      "step": 14460
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6265197387394515e-06,
+      "loss": 0.1726,
+      "step": 14461
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6227231188744696e-06,
+      "loss": 0.1678,
+      "step": 14462
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6189308989745355e-06,
+      "loss": 0.1726,
+      "step": 14463
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6151430792092914e-06,
+      "loss": 0.1735,
+      "step": 14464
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6113596597481239e-06,
+      "loss": 0.176,
+      "step": 14465
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6075806407602868e-06,
+      "loss": 0.1811,
+      "step": 14466
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6038060224147444e-06,
+      "loss": 0.1787,
+      "step": 14467
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.600035804880351e-06,
+      "loss": 0.1801,
+      "step": 14468
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.596269988325716e-06,
+      "loss": 0.1765,
+      "step": 14469
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5925085729192823e-06,
+      "loss": 0.1657,
+      "step": 14470
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.58875155882926e-06,
+      "loss": 0.1738,
+      "step": 14471
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5849989462237148e-06,
+      "loss": 0.1744,
+      "step": 14472
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5812507352704565e-06,
+      "loss": 0.1719,
+      "step": 14473
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.57750692613714e-06,
+      "loss": 0.1809,
+      "step": 14474
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5737675189911982e-06,
+      "loss": 0.1754,
+      "step": 14475
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.570032513999886e-06,
+      "loss": 0.1745,
+      "step": 14476
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.566301911330248e-06,
+      "loss": 0.1719,
+      "step": 14477
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5625757111491502e-06,
+      "loss": 0.1735,
+      "step": 14478
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5588539136232149e-06,
+      "loss": 0.1789,
+      "step": 14479
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5551365189189427e-06,
+      "loss": 0.1804,
+      "step": 14480
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5514235272025669e-06,
+      "loss": 0.1827,
+      "step": 14481
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5477149386401657e-06,
+      "loss": 0.1867,
+      "step": 14482
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5440107533976067e-06,
+      "loss": 0.1837,
+      "step": 14483
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5403109716405572e-06,
+      "loss": 0.1758,
+      "step": 14484
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5366155935344962e-06,
+      "loss": 0.1844,
+      "step": 14485
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5329246192446911e-06,
+      "loss": 0.1849,
+      "step": 14486
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5292380489362213e-06,
+      "loss": 0.1788,
+      "step": 14487
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.525555882773988e-06,
+      "loss": 0.179,
+      "step": 14488
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5218781209226597e-06,
+      "loss": 0.1731,
+      "step": 14489
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.518204763546738e-06,
+      "loss": 0.173,
+      "step": 14490
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5145358108105135e-06,
+      "loss": 0.174,
+      "step": 14491
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5108712628780775e-06,
+      "loss": 0.182,
+      "step": 14492
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.507211119913321e-06,
+      "loss": 0.1766,
+      "step": 14493
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5035553820799574e-06,
+      "loss": 0.17,
+      "step": 14494
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4999040495415008e-06,
+      "loss": 0.1827,
+      "step": 14495
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.496257122461231e-06,
+      "loss": 0.1775,
+      "step": 14496
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4926146010022735e-06,
+      "loss": 0.1719,
+      "step": 14497
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4889764853275423e-06,
+      "loss": 0.1733,
+      "step": 14498
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4853427755997517e-06,
+      "loss": 0.1791,
+      "step": 14499
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4817134719814274e-06,
+      "loss": 0.1796,
+      "step": 14500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4780885746348837e-06,
+      "loss": 0.1685,
+      "step": 14501
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.474468083722247e-06,
+      "loss": 0.1852,
+      "step": 14502
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4708519994054425e-06,
+      "loss": 0.1754,
+      "step": 14503
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4672403218462083e-06,
+      "loss": 0.1705,
+      "step": 14504
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4636330512060704e-06,
+      "loss": 0.1723,
+      "step": 14505
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4600301876463773e-06,
+      "loss": 0.1772,
+      "step": 14506
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.456431731328245e-06,
+      "loss": 0.1806,
+      "step": 14507
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4528376824126556e-06,
+      "loss": 0.1661,
+      "step": 14508
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4492480410603137e-06,
+      "loss": 0.1808,
+      "step": 14509
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4456628074317802e-06,
+      "loss": 0.178,
+      "step": 14510
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4420819816874264e-06,
+      "loss": 0.1833,
+      "step": 14511
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4385055639873802e-06,
+      "loss": 0.1762,
+      "step": 14512
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4349335544916132e-06,
+      "loss": 0.177,
+      "step": 14513
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4313659533598755e-06,
+      "loss": 0.1671,
+      "step": 14514
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4278027607517396e-06,
+      "loss": 0.1753,
+      "step": 14515
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4242439768265558e-06,
+      "loss": 0.1821,
+      "step": 14516
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4206896017435078e-06,
+      "loss": 0.1787,
+      "step": 14517
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4171396356615573e-06,
+      "loss": 0.1748,
+      "step": 14518
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4135940787394886e-06,
+      "loss": 0.1778,
+      "step": 14519
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4100529311358636e-06,
+      "loss": 0.1928,
+      "step": 14520
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4065161930090776e-06,
+      "loss": 0.1742,
+      "step": 14521
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4029838645172933e-06,
+      "loss": 0.1687,
+      "step": 14522
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3994559458185064e-06,
+      "loss": 0.1759,
+      "step": 14523
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3959324370705128e-06,
+      "loss": 0.1679,
+      "step": 14524
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3924133384308868e-06,
+      "loss": 0.1878,
+      "step": 14525
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3888986500570467e-06,
+      "loss": 0.1736,
+      "step": 14526
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3853883721061555e-06,
+      "loss": 0.1812,
+      "step": 14527
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3818825047352323e-06,
+      "loss": 0.1723,
+      "step": 14528
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.378381048101074e-06,
+      "loss": 0.1789,
+      "step": 14529
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3748840023602772e-06,
+      "loss": 0.1774,
+      "step": 14530
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3713913676692613e-06,
+      "loss": 0.1812,
+      "step": 14531
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3679031441842238e-06,
+      "loss": 0.178,
+      "step": 14532
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3644193320611952e-06,
+      "loss": 0.1738,
+      "step": 14533
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3609399314559734e-06,
+      "loss": 0.1726,
+      "step": 14534
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.357464942524178e-06,
+      "loss": 0.1807,
+      "step": 14535
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3539943654212296e-06,
+      "loss": 0.1699,
+      "step": 14536
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.35052820030237e-06,
+      "loss": 0.1781,
+      "step": 14537
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.347066447322598e-06,
+      "loss": 0.171,
+      "step": 14538
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3436091066367563e-06,
+      "loss": 0.1775,
+      "step": 14539
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3401561783994654e-06,
+      "loss": 0.1849,
+      "step": 14540
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3367076627651798e-06,
+      "loss": 0.1863,
+      "step": 14541
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3332635598881094e-06,
+      "loss": 0.1856,
+      "step": 14542
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3298238699223198e-06,
+      "loss": 0.1823,
+      "step": 14543
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3263885930216325e-06,
+      "loss": 0.1695,
+      "step": 14544
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.322957729339702e-06,
+      "loss": 0.1898,
+      "step": 14545
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3195312790299618e-06,
+      "loss": 0.1753,
+      "step": 14546
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3161092422456777e-06,
+      "loss": 0.173,
+      "step": 14547
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3126916191398942e-06,
+      "loss": 0.1727,
+      "step": 14548
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3092784098654665e-06,
+      "loss": 0.1736,
+      "step": 14549
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3058696145750504e-06,
+      "loss": 0.1796,
+      "step": 14550
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.302465233421113e-06,
+      "loss": 0.1726,
+      "step": 14551
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2990652665558988e-06,
+      "loss": 0.1773,
+      "step": 14552
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2956697141314978e-06,
+      "loss": 0.1704,
+      "step": 14553
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2922785762997547e-06,
+      "loss": 0.1711,
+      "step": 14554
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2888918532123595e-06,
+      "loss": 0.1864,
+      "step": 14555
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2855095450207688e-06,
+      "loss": 0.1783,
+      "step": 14556
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2821316518762616e-06,
+      "loss": 0.1788,
+      "step": 14557
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.278758173929917e-06,
+      "loss": 0.179,
+      "step": 14558
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2753891113326255e-06,
+      "loss": 0.1799,
+      "step": 14559
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2720244642350553e-06,
+      "loss": 0.1749,
+      "step": 14560
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2686642327876973e-06,
+      "loss": 0.1678,
+      "step": 14561
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2653084171408424e-06,
+      "loss": 0.1614,
+      "step": 14562
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2619570174445595e-06,
+      "loss": 0.1712,
+      "step": 14563
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2586100338487727e-06,
+      "loss": 0.1676,
+      "step": 14564
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2552674665031628e-06,
+      "loss": 0.1688,
+      "step": 14565
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.251929315557232e-06,
+      "loss": 0.1766,
+      "step": 14566
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2485955811602724e-06,
+      "loss": 0.177,
+      "step": 14567
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2452662634613866e-06,
+      "loss": 0.1846,
+      "step": 14568
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.241941362609489e-06,
+      "loss": 0.1845,
+      "step": 14569
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2386208787532827e-06,
+      "loss": 0.1727,
+      "step": 14570
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2353048120412824e-06,
+      "loss": 0.17,
+      "step": 14571
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2319931626217806e-06,
+      "loss": 0.1718,
+      "step": 14572
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2286859306429255e-06,
+      "loss": 0.173,
+      "step": 14573
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2253831162526097e-06,
+      "loss": 0.1801,
+      "step": 14574
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2220847195985485e-06,
+      "loss": 0.1719,
+      "step": 14575
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2187907408282906e-06,
+      "loss": 0.1825,
+      "step": 14576
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2155011800891291e-06,
+      "loss": 0.1805,
+      "step": 14577
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2122160375282132e-06,
+      "loss": 0.1855,
+      "step": 14578
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2089353132924697e-06,
+      "loss": 0.1838,
+      "step": 14579
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2056590075286256e-06,
+      "loss": 0.174,
+      "step": 14580
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2023871203832083e-06,
+      "loss": 0.1826,
+      "step": 14581
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1991196520025782e-06,
+      "loss": 0.185,
+      "step": 14582
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.195856602532841e-06,
+      "loss": 0.1852,
+      "step": 14583
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1925979721199687e-06,
+      "loss": 0.1777,
+      "step": 14584
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.189343760909678e-06,
+      "loss": 0.1755,
+      "step": 14585
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1860939690475414e-06,
+      "loss": 0.1789,
+      "step": 14586
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1828485966788872e-06,
+      "loss": 0.1814,
+      "step": 14587
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1796076439488768e-06,
+      "loss": 0.1725,
+      "step": 14588
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.17637111100245e-06,
+      "loss": 0.1703,
+      "step": 14589
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.17313899798438e-06,
+      "loss": 0.1736,
+      "step": 14590
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1699113050392062e-06,
+      "loss": 0.1755,
+      "step": 14591
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1666880323113028e-06,
+      "loss": 0.1733,
+      "step": 14592
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1634691799448317e-06,
+      "loss": 0.1762,
+      "step": 14593
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1602547480837556e-06,
+      "loss": 0.1817,
+      "step": 14594
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1570447368718262e-06,
+      "loss": 0.1749,
+      "step": 14595
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1538391464526288e-06,
+      "loss": 0.1699,
+      "step": 14596
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1506379769695264e-06,
+      "loss": 0.1716,
+      "step": 14597
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1474412285657154e-06,
+      "loss": 0.1731,
+      "step": 14598
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1442489013841373e-06,
+      "loss": 0.1848,
+      "step": 14599
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1410609955675888e-06,
+      "loss": 0.1845,
+      "step": 14600
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.137877511258656e-06,
+      "loss": 0.1804,
+      "step": 14601
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1346984485997025e-06,
+      "loss": 0.1764,
+      "step": 14602
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.131523807732926e-06,
+      "loss": 0.1768,
+      "step": 14603
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1283535888003238e-06,
+      "loss": 0.1699,
+      "step": 14604
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1251877919436605e-06,
+      "loss": 0.1604,
+      "step": 14605
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1220264173045558e-06,
+      "loss": 0.1807,
+      "step": 14606
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1188694650243747e-06,
+      "loss": 0.1698,
+      "step": 14607
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1157169352443375e-06,
+      "loss": 0.1655,
+      "step": 14608
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1125688281054204e-06,
+      "loss": 0.17,
+      "step": 14609
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1094251437484438e-06,
+      "loss": 0.1676,
+      "step": 14610
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1062858823139955e-06,
+      "loss": 0.168,
+      "step": 14611
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1031510439424854e-06,
+      "loss": 0.1827,
+      "step": 14612
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1000206287741232e-06,
+      "loss": 0.1798,
+      "step": 14613
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0968946369489196e-06,
+      "loss": 0.1752,
+      "step": 14614
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.093773068606685e-06,
+      "loss": 0.1689,
+      "step": 14615
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0906559238870295e-06,
+      "loss": 0.1757,
+      "step": 14616
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.087543202929364e-06,
+      "loss": 0.1807,
+      "step": 14617
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.084434905872911e-06,
+      "loss": 0.171,
+      "step": 14618
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0813310328566917e-06,
+      "loss": 0.1726,
+      "step": 14619
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0782315840195401e-06,
+      "loss": 0.1783,
+      "step": 14620
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0751365595000561e-06,
+      "loss": 0.1738,
+      "step": 14621
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.072045959436685e-06,
+      "loss": 0.1828,
+      "step": 14622
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.068959783967638e-06,
+      "loss": 0.1653,
+      "step": 14623
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0658780332309716e-06,
+      "loss": 0.1824,
+      "step": 14624
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0628007073644863e-06,
+      "loss": 0.1781,
+      "step": 14625
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0597278065058503e-06,
+      "loss": 0.1607,
+      "step": 14626
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0566593307924644e-06,
+      "loss": 0.1694,
+      "step": 14627
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0535952803615967e-06,
+      "loss": 0.171,
+      "step": 14628
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.050535655350282e-06,
+      "loss": 0.1699,
+      "step": 14629
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0474804558953554e-06,
+      "loss": 0.1768,
+      "step": 14630
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.044429682133452e-06,
+      "loss": 0.1685,
+      "step": 14631
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0413833342010515e-06,
+      "loss": 0.1691,
+      "step": 14632
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0383414122343672e-06,
+      "loss": 0.1751,
+      "step": 14633
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0353039163694788e-06,
+      "loss": 0.1703,
+      "step": 14634
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0322708467422225e-06,
+      "loss": 0.1765,
+      "step": 14635
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0292422034882676e-06,
+      "loss": 0.1689,
+      "step": 14636
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.02621798674305e-06,
+      "loss": 0.1681,
+      "step": 14637
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0231981966418503e-06,
+      "loss": 0.1835,
+      "step": 14638
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0201828333197273e-06,
+      "loss": 0.1733,
+      "step": 14639
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0171718969115285e-06,
+      "loss": 0.1662,
+      "step": 14640
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0141653875519352e-06,
+      "loss": 0.1787,
+      "step": 14641
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0111633053754066e-06,
+      "loss": 0.1675,
+      "step": 14642
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0081656505162129e-06,
+      "loss": 0.1914,
+      "step": 14643
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0051724231084248e-06,
+      "loss": 0.1784,
+      "step": 14644
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.002183623285924e-06,
+      "loss": 0.1811,
+      "step": 14645
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.99199251182381e-07,
+      "loss": 0.1802,
+      "step": 14646
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.962193069312674e-07,
+      "loss": 0.1763,
+      "step": 14647
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.932437906658764e-07,
+      "loss": 0.1729,
+      "step": 14648
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.902727025192682e-07,
+      "loss": 0.1688,
+      "step": 14649
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.873060426243363e-07,
+      "loss": 0.1728,
+      "step": 14650
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.843438111137748e-07,
+      "loss": 0.178,
+      "step": 14651
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.813860081200554e-07,
+      "loss": 0.1767,
+      "step": 14652
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.784326337754834e-07,
+      "loss": 0.1785,
+      "step": 14653
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.75483688212142e-07,
+      "loss": 0.171,
+      "step": 14654
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.725391715619148e-07,
+      "loss": 0.1775,
+      "step": 14655
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.695990839565073e-07,
+      "loss": 0.1768,
+      "step": 14656
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.666634255274144e-07,
+      "loss": 0.1652,
+      "step": 14657
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.637321964059309e-07,
+      "loss": 0.1706,
+      "step": 14658
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.608053967231633e-07,
+      "loss": 0.1712,
+      "step": 14659
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.578830266100069e-07,
+      "loss": 0.1682,
+      "step": 14660
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.549650861971793e-07,
+      "loss": 0.1821,
+      "step": 14661
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.520515756151649e-07,
+      "loss": 0.1791,
+      "step": 14662
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.491424949942928e-07,
+      "loss": 0.171,
+      "step": 14663
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.462378444646591e-07,
+      "loss": 0.17,
+      "step": 14664
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.433376241561931e-07,
+      "loss": 0.1778,
+      "step": 14665
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.40441834198591e-07,
+      "loss": 0.1792,
+      "step": 14666
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.375504747213714e-07,
+      "loss": 0.177,
+      "step": 14667
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.346635458538422e-07,
+      "loss": 0.1789,
+      "step": 14668
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.317810477251443e-07,
+      "loss": 0.1917,
+      "step": 14669
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.289029804641747e-07,
+      "loss": 0.1778,
+      "step": 14670
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.260293441996859e-07,
+      "loss": 0.1714,
+      "step": 14671
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.231601390601752e-07,
+      "loss": 0.178,
+      "step": 14672
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.202953651739843e-07,
+      "loss": 0.1816,
+      "step": 14673
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.17435022669233e-07,
+      "loss": 0.1863,
+      "step": 14674
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.145791116738633e-07,
+      "loss": 0.1738,
+      "step": 14675
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.117276323155954e-07,
+      "loss": 0.1981,
+      "step": 14676
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.088805847219606e-07,
+      "loss": 0.167,
+      "step": 14677
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.060379690203013e-07,
+      "loss": 0.1817,
+      "step": 14678
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.031997853377605e-07,
+      "loss": 0.175,
+      "step": 14679
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.003660338012698e-07,
+      "loss": 0.1731,
+      "step": 14680
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.975367145375613e-07,
+      "loss": 0.1746,
+      "step": 14681
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.947118276731892e-07,
+      "loss": 0.1691,
+      "step": 14682
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.918913733344969e-07,
+      "loss": 0.1779,
+      "step": 14683
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.89075351647628e-07,
+      "loss": 0.1672,
+      "step": 14684
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.862637627385373e-07,
+      "loss": 0.1824,
+      "step": 14685
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.834566067329575e-07,
+      "loss": 0.1831,
+      "step": 14686
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.80653883756466e-07,
+      "loss": 0.1821,
+      "step": 14687
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.778555939343847e-07,
+      "loss": 0.1759,
+      "step": 14688
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.750617373918912e-07,
+      "loss": 0.1732,
+      "step": 14689
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.722723142539191e-07,
+      "loss": 0.1782,
+      "step": 14690
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.694873246452573e-07,
+      "loss": 0.1809,
+      "step": 14691
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.667067686904395e-07,
+      "loss": 0.1796,
+      "step": 14692
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.639306465138442e-07,
+      "loss": 0.1698,
+      "step": 14693
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.611589582396273e-07,
+      "loss": 0.1854,
+      "step": 14694
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.583917039917567e-07,
+      "loss": 0.1829,
+      "step": 14695
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.556288838939885e-07,
+      "loss": 0.1683,
+      "step": 14696
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.528704980699132e-07,
+      "loss": 0.1779,
+      "step": 14697
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.501165466428651e-07,
+      "loss": 0.1694,
+      "step": 14698
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.47367029736057e-07,
+      "loss": 0.1781,
+      "step": 14699
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.446219474724349e-07,
+      "loss": 0.1818,
+      "step": 14700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.418812999747894e-07,
+      "loss": 0.1792,
+      "step": 14701
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.391450873656781e-07,
+      "loss": 0.1734,
+      "step": 14702
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.364133097674919e-07,
+      "loss": 0.1783,
+      "step": 14703
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.336859673024222e-07,
+      "loss": 0.1698,
+      "step": 14704
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.309630600924268e-07,
+      "loss": 0.1711,
+      "step": 14705
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.28244588259297e-07,
+      "loss": 0.1796,
+      "step": 14706
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.255305519246248e-07,
+      "loss": 0.1802,
+      "step": 14707
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.228209512097906e-07,
+      "loss": 0.1788,
+      "step": 14708
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.201157862359865e-07,
+      "loss": 0.178,
+      "step": 14709
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.174150571241934e-07,
+      "loss": 0.179,
+      "step": 14710
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.147187639952258e-07,
+      "loss": 0.1819,
+      "step": 14711
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.120269069696429e-07,
+      "loss": 0.1767,
+      "step": 14712
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.093394861678594e-07,
+      "loss": 0.1681,
+      "step": 14713
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.06656501710068e-07,
+      "loss": 0.1749,
+      "step": 14714
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.039779537162618e-07,
+      "loss": 0.177,
+      "step": 14715
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.013038423062447e-07,
+      "loss": 0.171,
+      "step": 14716
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.986341675996322e-07,
+      "loss": 0.1812,
+      "step": 14717
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.959689297157957e-07,
+      "loss": 0.1722,
+      "step": 14718
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.933081287739619e-07,
+      "loss": 0.165,
+      "step": 14719
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.906517648931244e-07,
+      "loss": 0.1775,
+      "step": 14720
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.879998381920883e-07,
+      "loss": 0.1747,
+      "step": 14721
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.853523487894809e-07,
+      "loss": 0.1588,
+      "step": 14722
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.827092968036964e-07,
+      "loss": 0.1763,
+      "step": 14723
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.800706823529513e-07,
+      "loss": 0.1769,
+      "step": 14724
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.774365055552513e-07,
+      "loss": 0.1822,
+      "step": 14725
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.748067665284242e-07,
+      "loss": 0.1675,
+      "step": 14726
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.72181465390076e-07,
+      "loss": 0.1703,
+      "step": 14727
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.69560602257624e-07,
+      "loss": 0.1787,
+      "step": 14728
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.669441772482966e-07,
+      "loss": 0.1779,
+      "step": 14729
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.643321904791112e-07,
+      "loss": 0.1779,
+      "step": 14730
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.617246420668855e-07,
+      "loss": 0.1737,
+      "step": 14731
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.591215321282486e-07,
+      "loss": 0.1637,
+      "step": 14732
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.565228607796071e-07,
+      "loss": 0.1717,
+      "step": 14733
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.539286281372238e-07,
+      "loss": 0.1804,
+      "step": 14734
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.513388343170836e-07,
+      "loss": 0.1699,
+      "step": 14735
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.487534794350603e-07,
+      "loss": 0.1788,
+      "step": 14736
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.461725636067507e-07,
+      "loss": 0.1773,
+      "step": 14737
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.435960869476066e-07,
+      "loss": 0.1816,
+      "step": 14738
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.410240495728471e-07,
+      "loss": 0.179,
+      "step": 14739
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.384564515975246e-07,
+      "loss": 0.1814,
+      "step": 14740
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.358932931364582e-07,
+      "loss": 0.176,
+      "step": 14741
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.333345743043007e-07,
+      "loss": 0.1718,
+      "step": 14742
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.30780295215483e-07,
+      "loss": 0.1713,
+      "step": 14743
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.282304559842579e-07,
+      "loss": 0.1824,
+      "step": 14744
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.256850567246565e-07,
+      "loss": 0.1769,
+      "step": 14745
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.231440975505321e-07,
+      "loss": 0.174,
+      "step": 14746
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.206075785755162e-07,
+      "loss": 0.1741,
+      "step": 14747
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.180754999130735e-07,
+      "loss": 0.1669,
+      "step": 14748
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.155478616764244e-07,
+      "loss": 0.1736,
+      "step": 14749
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.130246639786565e-07,
+      "loss": 0.1797,
+      "step": 14750
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.105059069325792e-07,
+      "loss": 0.1735,
+      "step": 14751
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.079915906508916e-07,
+      "loss": 0.177,
+      "step": 14752
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.054817152460036e-07,
+      "loss": 0.1717,
+      "step": 14753
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.029762808302032e-07,
+      "loss": 0.1663,
+      "step": 14754
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.004752875155229e-07,
+      "loss": 0.171,
+      "step": 14755
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.9797873541384e-07,
+      "loss": 0.1644,
+      "step": 14756
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.954866246367985e-07,
+      "loss": 0.1841,
+      "step": 14757
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.929989552958649e-07,
+      "loss": 0.1761,
+      "step": 14758
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.905157275023055e-07,
+      "loss": 0.1736,
+      "step": 14759
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.880369413671872e-07,
+      "loss": 0.1629,
+      "step": 14760
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.855625970013546e-07,
+      "loss": 0.1692,
+      "step": 14761
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.83092694515497e-07,
+      "loss": 0.1883,
+      "step": 14762
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.806272340200703e-07,
+      "loss": 0.16,
+      "step": 14763
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.781662156253421e-07,
+      "loss": 0.1706,
+      "step": 14764
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.757096394413908e-07,
+      "loss": 0.1758,
+      "step": 14765
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.732575055780843e-07,
+      "loss": 0.1785,
+      "step": 14766
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.708098141450792e-07,
+      "loss": 0.1663,
+      "step": 14767
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.683665652518767e-07,
+      "loss": 0.1814,
+      "step": 14768
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.65927759007734e-07,
+      "loss": 0.1712,
+      "step": 14769
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.634933955217304e-07,
+      "loss": 0.1661,
+      "step": 14770
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.610634749027456e-07,
+      "loss": 0.1815,
+      "step": 14771
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.586379972594702e-07,
+      "loss": 0.1838,
+      "step": 14772
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.562169627003623e-07,
+      "loss": 0.1729,
+      "step": 14773
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.538003713337237e-07,
+      "loss": 0.1691,
+      "step": 14774
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.513882232676238e-07,
+      "loss": 0.1757,
+      "step": 14775
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.48980518609954e-07,
+      "loss": 0.1649,
+      "step": 14776
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.465772574683948e-07,
+      "loss": 0.1739,
+      "step": 14777
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.44178439950438e-07,
+      "loss": 0.1706,
+      "step": 14778
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.417840661633645e-07,
+      "loss": 0.1675,
+      "step": 14779
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.393941362142885e-07,
+      "loss": 0.1732,
+      "step": 14780
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.370086502100581e-07,
+      "loss": 0.1726,
+      "step": 14781
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.346276082574098e-07,
+      "loss": 0.1753,
+      "step": 14782
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.322510104628033e-07,
+      "loss": 0.1801,
+      "step": 14783
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.29878856932542e-07,
+      "loss": 0.1754,
+      "step": 14784
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.275111477727303e-07,
+      "loss": 0.1675,
+      "step": 14785
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.25147883089261e-07,
+      "loss": 0.1751,
+      "step": 14786
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.227890629878274e-07,
+      "loss": 0.1837,
+      "step": 14787
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.204346875739342e-07,
+      "loss": 0.1747,
+      "step": 14788
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.180847569528747e-07,
+      "loss": 0.1787,
+      "step": 14789
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.157392712297538e-07,
+      "loss": 0.1699,
+      "step": 14790
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.133982305094765e-07,
+      "loss": 0.1786,
+      "step": 14791
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.110616348967591e-07,
+      "loss": 0.1724,
+      "step": 14792
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.087294844960733e-07,
+      "loss": 0.1726,
+      "step": 14793
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.064017794117582e-07,
+      "loss": 0.1855,
+      "step": 14794
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.040785197479081e-07,
+      "loss": 0.174,
+      "step": 14795
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.01759705608429e-07,
+      "loss": 0.1697,
+      "step": 14796
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.994453370970488e-07,
+      "loss": 0.1746,
+      "step": 14797
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.971354143172403e-07,
+      "loss": 0.1704,
+      "step": 14798
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.948299373723543e-07,
+      "loss": 0.1825,
+      "step": 14799
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.92528906365486e-07,
+      "loss": 0.17,
+      "step": 14800
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.902323213995642e-07,
+      "loss": 0.1671,
+      "step": 14801
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.879401825772735e-07,
+      "loss": 0.1665,
+      "step": 14802
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.856524900011651e-07,
+      "loss": 0.1691,
+      "step": 14803
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.833692437735461e-07,
+      "loss": 0.1861,
+      "step": 14804
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.810904439965237e-07,
+      "loss": 0.1622,
+      "step": 14805
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.788160907720386e-07,
+      "loss": 0.167,
+      "step": 14806
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.765461842017983e-07,
+      "loss": 0.1712,
+      "step": 14807
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.742807243873216e-07,
+      "loss": 0.1702,
+      "step": 14808
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.720197114299386e-07,
+      "loss": 0.1646,
+      "step": 14809
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.697631454307794e-07,
+      "loss": 0.1695,
+      "step": 14810
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.675110264907635e-07,
+      "loss": 0.1666,
+      "step": 14811
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.652633547106101e-07,
+      "loss": 0.1699,
+      "step": 14812
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.630201301908611e-07,
+      "loss": 0.1785,
+      "step": 14813
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.607813530318473e-07,
+      "loss": 0.1623,
+      "step": 14814
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.585470233336887e-07,
+      "loss": 0.1703,
+      "step": 14815
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.563171411963164e-07,
+      "loss": 0.1714,
+      "step": 14816
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.540917067194729e-07,
+      "loss": 0.1712,
+      "step": 14817
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.518707200026785e-07,
+      "loss": 0.1735,
+      "step": 14818
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.49654181145276e-07,
+      "loss": 0.1742,
+      "step": 14819
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.474420902463973e-07,
+      "loss": 0.1712,
+      "step": 14820
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.452344474049853e-07,
+      "loss": 0.1819,
+      "step": 14821
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.430312527197723e-07,
+      "loss": 0.1736,
+      "step": 14822
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.408325062893016e-07,
+      "loss": 0.1829,
+      "step": 14823
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.386382082119056e-07,
+      "loss": 0.1767,
+      "step": 14824
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.364483585857394e-07,
+      "loss": 0.1853,
+      "step": 14825
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.342629575087244e-07,
+      "loss": 0.1778,
+      "step": 14826
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.320820050786268e-07,
+      "loss": 0.1723,
+      "step": 14827
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.299055013929688e-07,
+      "loss": 0.1699,
+      "step": 14828
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.277334465491058e-07,
+      "loss": 0.1808,
+      "step": 14829
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.255658406441822e-07,
+      "loss": 0.1699,
+      "step": 14830
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.234026837751426e-07,
+      "loss": 0.1825,
+      "step": 14831
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.212439760387433e-07,
+      "loss": 0.1753,
+      "step": 14832
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.190897175315291e-07,
+      "loss": 0.175,
+      "step": 14833
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.169399083498561e-07,
+      "loss": 0.1911,
+      "step": 14834
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.147945485898587e-07,
+      "loss": 0.1758,
+      "step": 14835
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.126536383475044e-07,
+      "loss": 0.1753,
+      "step": 14836
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.105171777185391e-07,
+      "loss": 0.171,
+      "step": 14837
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.083851667985195e-07,
+      "loss": 0.18,
+      "step": 14838
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.062576056827916e-07,
+      "loss": 0.1786,
+      "step": 14839
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.041344944665349e-07,
+      "loss": 0.1839,
+      "step": 14840
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.020158332446845e-07,
+      "loss": 0.1787,
+      "step": 14841
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.999016221119979e-07,
+      "loss": 0.1695,
+      "step": 14842
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.977918611630549e-07,
+      "loss": 0.1774,
+      "step": 14843
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.956865504921915e-07,
+      "loss": 0.1805,
+      "step": 14844
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.935856901935876e-07,
+      "loss": 0.1749,
+      "step": 14845
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.914892803611904e-07,
+      "loss": 0.1742,
+      "step": 14846
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.893973210887803e-07,
+      "loss": 0.1757,
+      "step": 14847
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.873098124698939e-07,
+      "loss": 0.1735,
+      "step": 14848
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.852267545979228e-07,
+      "loss": 0.1733,
+      "step": 14849
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.831481475660149e-07,
+      "loss": 0.1751,
+      "step": 14850
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.810739914671514e-07,
+      "loss": 0.1741,
+      "step": 14851
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.790042863940802e-07,
+      "loss": 0.1806,
+      "step": 14852
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.769390324393941e-07,
+      "loss": 0.1655,
+      "step": 14853
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7487822969544127e-07,
+      "loss": 0.1706,
+      "step": 14854
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.728218782544036e-07,
+      "loss": 0.1783,
+      "step": 14855
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.707699782082409e-07,
+      "loss": 0.1798,
+      "step": 14856
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.687225296487463e-07,
+      "loss": 0.1785,
+      "step": 14857
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.666795326674689e-07,
+      "loss": 0.1707,
+      "step": 14858
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.646409873558022e-07,
+      "loss": 0.177,
+      "step": 14859
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.626068938049066e-07,
+      "loss": 0.1783,
+      "step": 14860
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.60577252105765e-07,
+      "loss": 0.1774,
+      "step": 14861
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5855206234916016e-07,
+      "loss": 0.177,
+      "step": 14862
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5653132462565306e-07,
+      "loss": 0.1695,
+      "step": 14863
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5451503902563807e-07,
+      "loss": 0.177,
+      "step": 14864
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.525032056392764e-07,
+      "loss": 0.1674,
+      "step": 14865
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.504958245565738e-07,
+      "loss": 0.1783,
+      "step": 14866
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4849289586729184e-07,
+      "loss": 0.1642,
+      "step": 14867
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4649441966101436e-07,
+      "loss": 0.1707,
+      "step": 14868
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4450039602712543e-07,
+      "loss": 0.1713,
+      "step": 14869
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.425108250548093e-07,
+      "loss": 0.1717,
+      "step": 14870
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.4052570683305037e-07,
+      "loss": 0.1783,
+      "step": 14871
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3854504145063315e-07,
+      "loss": 0.1888,
+      "step": 14872
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3656882899615334e-07,
+      "loss": 0.1692,
+      "step": 14873
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3459706955798486e-07,
+      "loss": 0.176,
+      "step": 14874
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3262976322432367e-07,
+      "loss": 0.1711,
+      "step": 14875
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3066691008315507e-07,
+      "loss": 0.1747,
+      "step": 14876
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2870851022226433e-07,
+      "loss": 0.1734,
+      "step": 14877
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2675456372924805e-07,
+      "loss": 0.1688,
+      "step": 14878
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.248050706914808e-07,
+      "loss": 0.1721,
+      "step": 14879
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2286003119619276e-07,
+      "loss": 0.1717,
+      "step": 14880
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.209194453303367e-07,
+      "loss": 0.1658,
+      "step": 14881
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.189833131807208e-07,
+      "loss": 0.1804,
+      "step": 14882
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.170516348339537e-07,
+      "loss": 0.1738,
+      "step": 14883
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.151244103764107e-07,
+      "loss": 0.1703,
+      "step": 14884
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.132016398942895e-07,
+      "loss": 0.1758,
+      "step": 14885
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.112833234735991e-07,
+      "loss": 0.1814,
+      "step": 14886
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0936946120012643e-07,
+      "loss": 0.183,
+      "step": 14887
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.074600531594697e-07,
+      "loss": 0.1749,
+      "step": 14888
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.055550994370383e-07,
+      "loss": 0.1824,
+      "step": 14889
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.036546001180197e-07,
+      "loss": 0.1753,
+      "step": 14890
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0175855528742366e-07,
+      "loss": 0.1735,
+      "step": 14891
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9986696503003795e-07,
+      "loss": 0.167,
+      "step": 14892
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.979798294304837e-07,
+      "loss": 0.168,
+      "step": 14893
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.960971485731602e-07,
+      "loss": 0.1787,
+      "step": 14894
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9421892254225546e-07,
+      "loss": 0.167,
+      "step": 14895
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.923451514217802e-07,
+      "loss": 0.1763,
+      "step": 14896
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.904758352955562e-07,
+      "loss": 0.1707,
+      "step": 14897
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.886109742471722e-07,
+      "loss": 0.1764,
+      "step": 14898
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.867505683600392e-07,
+      "loss": 0.1831,
+      "step": 14899
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8489461771735736e-07,
+      "loss": 0.1788,
+      "step": 14900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8304312240214913e-07,
+      "loss": 0.1686,
+      "step": 14901
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.8119608249722605e-07,
+      "loss": 0.1743,
+      "step": 14902
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7935349808517764e-07,
+      "loss": 0.1824,
+      "step": 14903
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7751536924842676e-07,
+      "loss": 0.1746,
+      "step": 14904
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.756816960691745e-07,
+      "loss": 0.1683,
+      "step": 14905
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7385247862945503e-07,
+      "loss": 0.1693,
+      "step": 14906
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7202771701105866e-07,
+      "loss": 0.1687,
+      "step": 14907
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.702074112956089e-07,
+      "loss": 0.1785,
+      "step": 14908
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6839156156451837e-07,
+      "loss": 0.1707,
+      "step": 14909
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6658016789899997e-07,
+      "loss": 0.1763,
+      "step": 14910
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6477323038007773e-07,
+      "loss": 0.1784,
+      "step": 14911
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.629707490885537e-07,
+      "loss": 0.1822,
+      "step": 14912
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.611727241050411e-07,
+      "loss": 0.1799,
+      "step": 14913
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.593791555099757e-07,
+      "loss": 0.1733,
+      "step": 14914
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5759004338357105e-07,
+      "loss": 0.1814,
+      "step": 14915
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5580538780584094e-07,
+      "loss": 0.1728,
+      "step": 14916
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.540251888565993e-07,
+      "loss": 0.1668,
+      "step": 14917
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5224944661547134e-07,
+      "loss": 0.1774,
+      "step": 14918
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.504781611618824e-07,
+      "loss": 0.1914,
+      "step": 14919
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4871133257505794e-07,
+      "loss": 0.1766,
+      "step": 14920
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4694896093400153e-07,
+      "loss": 0.1766,
+      "step": 14921
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.451910463175501e-07,
+      "loss": 0.1669,
+      "step": 14922
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.434375888043295e-07,
+      "loss": 0.1766,
+      "step": 14923
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4168858847275497e-07,
+      "loss": 0.1747,
+      "step": 14924
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.399440454010638e-07,
+      "loss": 0.1568,
+      "step": 14925
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.382039596672715e-07,
+      "loss": 0.1742,
+      "step": 14926
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3646833134919344e-07,
+      "loss": 0.1683,
+      "step": 14927
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.347371605244787e-07,
+      "loss": 0.1832,
+      "step": 14928
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3301044727053197e-07,
+      "loss": 0.1821,
+      "step": 14929
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3128819166461376e-07,
+      "loss": 0.1714,
+      "step": 14930
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.2957039378371803e-07,
+      "loss": 0.184,
+      "step": 14931
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.278570537046832e-07,
+      "loss": 0.175,
+      "step": 14932
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.261481715041592e-07,
+      "loss": 0.1741,
+      "step": 14933
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.244437472585515e-07,
+      "loss": 0.1746,
+      "step": 14934
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.227437810440992e-07,
+      "loss": 0.169,
+      "step": 14935
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.210482729368414e-07,
+      "loss": 0.172,
+      "step": 14936
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.193572230126063e-07,
+      "loss": 0.1739,
+      "step": 14937
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1767063134702233e-07,
+      "loss": 0.1794,
+      "step": 14938
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.159884980155292e-07,
+      "loss": 0.1741,
+      "step": 14939
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1431082309335556e-07,
+      "loss": 0.1723,
+      "step": 14940
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.126376066555414e-07,
+      "loss": 0.1775,
+      "step": 14941
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.109688487769158e-07,
+      "loss": 0.1747,
+      "step": 14942
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0930454953213006e-07,
+      "loss": 0.1677,
+      "step": 14943
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.076447089955914e-07,
+      "loss": 0.1834,
+      "step": 14944
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0598932724156257e-07,
+      "loss": 0.1761,
+      "step": 14945
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.043384043440733e-07,
+      "loss": 0.1652,
+      "step": 14946
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0269194037696456e-07,
+      "loss": 0.1672,
+      "step": 14947
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0104993541386627e-07,
+      "loss": 0.1781,
+      "step": 14948
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.994123895282308e-07,
+      "loss": 0.1782,
+      "step": 14949
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9777930279327734e-07,
+      "loss": 0.1733,
+      "step": 14950
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9615067528206973e-07,
+      "loss": 0.1662,
+      "step": 14951
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9452650706743855e-07,
+      "loss": 0.1735,
+      "step": 14952
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9290679822202573e-07,
+      "loss": 0.1769,
+      "step": 14953
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9129154881827323e-07,
+      "loss": 0.1707,
+      "step": 14954
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8968075892842337e-07,
+      "loss": 0.1808,
+      "step": 14955
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.880744286245185e-07,
+      "loss": 0.1744,
+      "step": 14956
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.864725579784122e-07,
+      "loss": 0.167,
+      "step": 14957
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.84875147061725e-07,
+      "loss": 0.1609,
+      "step": 14958
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.832821959459331e-07,
+      "loss": 0.1713,
+      "step": 14959
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.816937047022572e-07,
+      "loss": 0.1741,
+      "step": 14960
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.801096734017627e-07,
+      "loss": 0.1732,
+      "step": 14961
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7853010211527085e-07,
+      "loss": 0.1778,
+      "step": 14962
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7695499091345833e-07,
+      "loss": 0.1731,
+      "step": 14963
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.753843398667466e-07,
+      "loss": 0.181,
+      "step": 14964
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.738181490454017e-07,
+      "loss": 0.1714,
+      "step": 14965
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.722564185194676e-07,
+      "loss": 0.164,
+      "step": 14966
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7069914835878843e-07,
+      "loss": 0.1751,
+      "step": 14967
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.691463386330084e-07,
+      "loss": 0.1732,
+      "step": 14968
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.675979894115943e-07,
+      "loss": 0.1839,
+      "step": 14969
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6605410076379065e-07,
+      "loss": 0.1759,
+      "step": 14970
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.645146727586423e-07,
+      "loss": 0.1821,
+      "step": 14971
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.629797054650052e-07,
+      "loss": 0.1724,
+      "step": 14972
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.614491989515244e-07,
+      "loss": 0.1694,
+      "step": 14973
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.599231532866675e-07,
+      "loss": 0.1745,
+      "step": 14974
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.584015685386798e-07,
+      "loss": 0.1737,
+      "step": 14975
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5688444477560693e-07,
+      "loss": 0.1676,
+      "step": 14976
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5537178206530565e-07,
+      "loss": 0.1736,
+      "step": 14977
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5386358047544413e-07,
+      "loss": 0.1681,
+      "step": 14978
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.523598400734684e-07,
+      "loss": 0.1766,
+      "step": 14979
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.508605609266246e-07,
+      "loss": 0.1851,
+      "step": 14980
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4936574310198137e-07,
+      "loss": 0.1677,
+      "step": 14981
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4787538666638524e-07,
+      "loss": 0.1762,
+      "step": 14982
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4638949168651616e-07,
+      "loss": 0.1609,
+      "step": 14983
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4490805822879883e-07,
+      "loss": 0.1714,
+      "step": 14984
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.434310863595135e-07,
+      "loss": 0.1843,
+      "step": 14985
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.419585761447074e-07,
+      "loss": 0.1841,
+      "step": 14986
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4049052765023893e-07,
+      "loss": 0.174,
+      "step": 14987
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.390269409417778e-07,
+      "loss": 0.1865,
+      "step": 14988
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3756781608477163e-07,
+      "loss": 0.1783,
+      "step": 14989
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3611315314449044e-07,
+      "loss": 0.1763,
+      "step": 14990
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3466295218598223e-07,
+      "loss": 0.1718,
+      "step": 14991
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3321721327411727e-07,
+      "loss": 0.1772,
+      "step": 14992
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.317759364735661e-07,
+      "loss": 0.1651,
+      "step": 14993
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.3033912184876605e-07,
+      "loss": 0.1925,
+      "step": 14994
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2890676946399902e-07,
+      "loss": 0.1612,
+      "step": 14995
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.274788793833138e-07,
+      "loss": 0.1731,
+      "step": 14996
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.260554516705926e-07,
+      "loss": 0.1757,
+      "step": 14997
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2463648638947344e-07,
+      "loss": 0.1692,
+      "step": 14998
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.232219836034499e-07,
+      "loss": 0.1759,
+      "step": 14999
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2181194337574927e-07,
+      "loss": 0.1751,
+      "step": 15000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.204063657694766e-07,
+      "loss": 0.1756,
+      "step": 15001
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1900525084745938e-07,
+      "loss": 0.1811,
+      "step": 15002
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1760859867239193e-07,
+      "loss": 0.184,
+      "step": 15003
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1621640930672427e-07,
+      "loss": 0.1745,
+      "step": 15004
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1482868281272884e-07,
+      "loss": 0.1621,
+      "step": 15005
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1344541925247818e-07,
+      "loss": 0.1773,
+      "step": 15006
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1206661868782284e-07,
+      "loss": 0.1816,
+      "step": 15007
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1069228118044682e-07,
+      "loss": 0.1651,
+      "step": 15008
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0932240679180094e-07,
+      "loss": 0.1685,
+      "step": 15009
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.079569955831695e-07,
+      "loss": 0.1764,
+      "step": 15010
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.065960476156148e-07,
+      "loss": 0.1753,
+      "step": 15011
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0523956294999925e-07,
+      "loss": 0.1757,
+      "step": 15012
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0388754164699652e-07,
+      "loss": 0.1765,
+      "step": 15013
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.025399837670805e-07,
+      "loss": 0.1757,
+      "step": 15014
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0119688937052516e-07,
+      "loss": 0.1697,
+      "step": 15015
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9985825851739358e-07,
+      "loss": 0.1842,
+      "step": 15016
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.985240912675601e-07,
+      "loss": 0.1717,
+      "step": 15017
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.971943876806881e-07,
+      "loss": 0.1707,
+      "step": 15018
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9586914781625222e-07,
+      "loss": 0.1706,
+      "step": 15019
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9454837173353834e-07,
+      "loss": 0.1643,
+      "step": 15020
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9323205949159928e-07,
+      "loss": 0.1808,
+      "step": 15021
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9192021114931014e-07,
+      "loss": 0.1826,
+      "step": 15022
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.906128267653573e-07,
+      "loss": 0.1743,
+      "step": 15023
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.893099063982051e-07,
+      "loss": 0.1705,
+      "step": 15024
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.880114501061292e-07,
+      "loss": 0.1722,
+      "step": 15025
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.867174579471942e-07,
+      "loss": 0.1624,
+      "step": 15026
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8542792997929825e-07,
+      "loss": 0.1742,
+      "step": 15027
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8414286626009526e-07,
+      "loss": 0.1676,
+      "step": 15028
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8286226684706143e-07,
+      "loss": 0.1712,
+      "step": 15029
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.815861317974732e-07,
+      "loss": 0.1717,
+      "step": 15030
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.803144611684071e-07,
+      "loss": 0.177,
+      "step": 15031
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.79047255016751e-07,
+      "loss": 0.1735,
+      "step": 15032
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.777845133991707e-07,
+      "loss": 0.1717,
+      "step": 15033
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7652623637214317e-07,
+      "loss": 0.1747,
+      "step": 15034
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7527242399194566e-07,
+      "loss": 0.1686,
+      "step": 15035
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7402307631465554e-07,
+      "loss": 0.1841,
+      "step": 15036
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.727781933961392e-07,
+      "loss": 0.177,
+      "step": 15037
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7153777529209657e-07,
+      "loss": 0.1745,
+      "step": 15038
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.703018220579944e-07,
+      "loss": 0.1871,
+      "step": 15039
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6907033374912174e-07,
+      "loss": 0.1799,
+      "step": 15040
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.678433104205346e-07,
+      "loss": 0.1842,
+      "step": 15041
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6662075212713347e-07,
+      "loss": 0.178,
+      "step": 15042
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6540265892358574e-07,
+      "loss": 0.187,
+      "step": 15043
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.641890308643701e-07,
+      "loss": 0.1682,
+      "step": 15044
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.629798680037764e-07,
+      "loss": 0.182,
+      "step": 15045
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6177517039588362e-07,
+      "loss": 0.1711,
+      "step": 15046
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.605749380945598e-07,
+      "loss": 0.1753,
+      "step": 15047
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5937917115350643e-07,
+      "loss": 0.1723,
+      "step": 15048
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5818786962618071e-07,
+      "loss": 0.1742,
+      "step": 15049
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5700103356588448e-07,
+      "loss": 0.1868,
+      "step": 15050
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5581866302569747e-07,
+      "loss": 0.164,
+      "step": 15051
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5464075805848854e-07,
+      "loss": 0.1773,
+      "step": 15052
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5346731871693776e-07,
+      "loss": 0.1695,
+      "step": 15053
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5229834505354756e-07,
+      "loss": 0.1713,
+      "step": 15054
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5113383712058726e-07,
+      "loss": 0.1842,
+      "step": 15055
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.499737949701485e-07,
+      "loss": 0.1721,
+      "step": 15056
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4881821865410094e-07,
+      "loss": 0.1684,
+      "step": 15057
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4766710822414764e-07,
+      "loss": 0.1768,
+      "step": 15058
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4652046373174743e-07,
+      "loss": 0.1835,
+      "step": 15059
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4537828522821484e-07,
+      "loss": 0.175,
+      "step": 15060
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.44240572764609e-07,
+      "loss": 0.1778,
+      "step": 15061
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4310732639182257e-07,
+      "loss": 0.1855,
+      "step": 15062
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4197854616053717e-07,
+      "loss": 0.17,
+      "step": 15063
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4085423212124582e-07,
+      "loss": 0.1822,
+      "step": 15064
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3973438432423046e-07,
+      "loss": 0.1749,
+      "step": 15065
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.386190028195844e-07,
+      "loss": 0.1665,
+      "step": 15066
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.375080876571788e-07,
+      "loss": 0.1931,
+      "step": 15067
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3640163888670733e-07,
+      "loss": 0.1752,
+      "step": 15068
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3529965655765253e-07,
+      "loss": 0.173,
+      "step": 15069
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3420214071930838e-07,
+      "loss": 0.1709,
+      "step": 15070
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3310909142074667e-07,
+      "loss": 0.1639,
+      "step": 15071
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.320205087108728e-07,
+      "loss": 0.1709,
+      "step": 15072
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3093639263836998e-07,
+      "loss": 0.1675,
+      "step": 15073
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.298567432517217e-07,
+      "loss": 0.1738,
+      "step": 15074
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2878156059921153e-07,
+      "loss": 0.1849,
+      "step": 15075
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2771084472894545e-07,
+      "loss": 0.1835,
+      "step": 15076
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2664459568878516e-07,
+      "loss": 0.1791,
+      "step": 15077
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2558281352643696e-07,
+      "loss": 0.1746,
+      "step": 15078
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.245254982893851e-07,
+      "loss": 0.1705,
+      "step": 15079
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2347265002491392e-07,
+      "loss": 0.1765,
+      "step": 15080
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2242426878011915e-07,
+      "loss": 0.1704,
+      "step": 15081
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2138035460189656e-07,
+      "loss": 0.18,
+      "step": 15082
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2034090753691996e-07,
+      "loss": 0.1819,
+      "step": 15083
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1930592763168546e-07,
+      "loss": 0.1848,
+      "step": 15084
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.182754149324894e-07,
+      "loss": 0.1808,
+      "step": 15085
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1724936948540599e-07,
+      "loss": 0.1722,
+      "step": 15086
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1622779133634299e-07,
+      "loss": 0.172,
+      "step": 15087
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1521068053098604e-07,
+      "loss": 0.1715,
+      "step": 15088
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1419803711482102e-07,
+      "loss": 0.1778,
+      "step": 15089
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1318986113313391e-07,
+      "loss": 0.1736,
+      "step": 15090
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1218615263103305e-07,
+      "loss": 0.1684,
+      "step": 15091
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1118691165339368e-07,
+      "loss": 0.179,
+      "step": 15092
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1019213824491337e-07,
+      "loss": 0.1796,
+      "step": 15093
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0920183245008986e-07,
+      "loss": 0.1688,
+      "step": 15094
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0821599431320994e-07,
+      "loss": 0.1648,
+      "step": 15095
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0723462387836058e-07,
+      "loss": 0.1761,
+      "step": 15096
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0625772118944e-07,
+      "loss": 0.1701,
+      "step": 15097
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0528528629012435e-07,
+      "loss": 0.1739,
+      "step": 15098
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.043173192239344e-07,
+      "loss": 0.1713,
+      "step": 15099
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0335382003414662e-07,
+      "loss": 0.1853,
+      "step": 15100
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0239478876385988e-07,
+      "loss": 0.1634,
+      "step": 15101
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.014402254559621e-07,
+      "loss": 0.1717,
+      "step": 15102
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0049013015314135e-07,
+      "loss": 0.1605,
+      "step": 15103
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.954450289790806e-08,
+      "loss": 0.1822,
+      "step": 15104
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.860334373253954e-08,
+      "loss": 0.1699,
+      "step": 15105
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.766665269913544e-08,
+      "loss": 0.1782,
+      "step": 15106
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.673442983959558e-08,
+      "loss": 0.1712,
+      "step": 15107
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.580667519559771e-08,
+      "loss": 0.1749,
+      "step": 15108
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.488338880865311e-08,
+      "loss": 0.1784,
+      "step": 15109
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.396457072005094e-08,
+      "loss": 0.1812,
+      "step": 15110
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.305022097088057e-08,
+      "loss": 0.1767,
+      "step": 15111
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.21403396020426e-08,
+      "loss": 0.1714,
+      "step": 15112
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.123492665422672e-08,
+      "loss": 0.1755,
+      "step": 15113
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.033398216793387e-08,
+      "loss": 0.1764,
+      "step": 15114
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.943750618345404e-08,
+      "loss": 0.1714,
+      "step": 15115
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.854549874088847e-08,
+      "loss": 0.1716,
+      "step": 15116
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.765795988012748e-08,
+      "loss": 0.1691,
+      "step": 15117
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.677488964086156e-08,
+      "loss": 0.18,
+      "step": 15118
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.589628806260352e-08,
+      "loss": 0.1662,
+      "step": 15119
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.502215518462197e-08,
+      "loss": 0.1662,
+      "step": 15120
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.415249104604117e-08,
+      "loss": 0.1764,
+      "step": 15121
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.328729568575222e-08,
+      "loss": 0.1688,
+      "step": 15122
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.24265691424353e-08,
+      "loss": 0.1824,
+      "step": 15123
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.157031145459293e-08,
+      "loss": 0.1808,
+      "step": 15124
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.071852266052782e-08,
+      "loss": 0.1688,
+      "step": 15125
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.987120279833171e-08,
+      "loss": 0.1849,
+      "step": 15126
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.902835190590763e-08,
+      "loss": 0.1829,
+      "step": 15127
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.818997002094764e-08,
+      "loss": 0.1752,
+      "step": 15128
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.735605718094396e-08,
+      "loss": 0.1675,
+      "step": 15129
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.652661342320011e-08,
+      "loss": 0.1758,
+      "step": 15130
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.570163878481972e-08,
+      "loss": 0.1662,
+      "step": 15131
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.488113330268443e-08,
+      "loss": 0.192,
+      "step": 15132
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.406509701349817e-08,
+      "loss": 0.1732,
+      "step": 15133
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.325352995376511e-08,
+      "loss": 0.1795,
+      "step": 15134
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.244643215976732e-08,
+      "loss": 0.1777,
+      "step": 15135
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.164380366762036e-08,
+      "loss": 0.1773,
+      "step": 15136
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.084564451321773e-08,
+      "loss": 0.1776,
+      "step": 15137
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.005195473225312e-08,
+      "loss": 0.1756,
+      "step": 15138
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.926273436022035e-08,
+      "loss": 0.1686,
+      "step": 15139
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.847798343242451e-08,
+      "loss": 0.1761,
+      "step": 15140
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.769770198397085e-08,
+      "loss": 0.1865,
+      "step": 15141
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.692189004974259e-08,
+      "loss": 0.1724,
+      "step": 15142
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.615054766444529e-08,
+      "loss": 0.187,
+      "step": 15143
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.538367486257357e-08,
+      "loss": 0.1657,
+      "step": 15144
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.462127167843335e-08,
+      "loss": 0.1707,
+      "step": 15145
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.386333814611955e-08,
+      "loss": 0.1873,
+      "step": 15146
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.310987429953841e-08,
+      "loss": 0.174,
+      "step": 15147
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.236088017237406e-08,
+      "loss": 0.1632,
+      "step": 15148
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.161635579813308e-08,
+      "loss": 0.1806,
+      "step": 15149
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.087630121012212e-08,
+      "loss": 0.1922,
+      "step": 15150
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.014071644142583e-08,
+      "loss": 0.1744,
+      "step": 15151
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.940960152496233e-08,
+      "loss": 0.1728,
+      "step": 15152
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.8682956493405496e-08,
+      "loss": 0.1822,
+      "step": 15153
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.796078137928485e-08,
+      "loss": 0.1834,
+      "step": 15154
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.7243076214874567e-08,
+      "loss": 0.1711,
+      "step": 15155
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.6529841032282314e-08,
+      "loss": 0.1757,
+      "step": 15156
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.5821075863415896e-08,
+      "loss": 0.1852,
+      "step": 15157
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.511678073996107e-08,
+      "loss": 0.1655,
+      "step": 15158
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.441695569343708e-08,
+      "loss": 0.1802,
+      "step": 15159
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.3721600755118895e-08,
+      "loss": 0.1772,
+      "step": 15160
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.303071595612608e-08,
+      "loss": 0.1786,
+      "step": 15161
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.234430132735613e-08,
+      "loss": 0.184,
+      "step": 15162
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.1662356899506715e-08,
+      "loss": 0.1895,
+      "step": 15163
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.0984882703075666e-08,
+      "loss": 0.1716,
+      "step": 15164
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.0311878768360965e-08,
+      "loss": 0.1748,
+      "step": 15165
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.9643345125460764e-08,
+      "loss": 0.1653,
+      "step": 15166
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.897928180429556e-08,
+      "loss": 0.1766,
+      "step": 15167
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.831968883454163e-08,
+      "loss": 0.1765,
+      "step": 15168
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.76645662457198e-08,
+      "loss": 0.1739,
+      "step": 15169
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.701391406710665e-08,
+      "loss": 0.1825,
+      "step": 15170
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.6367732327834424e-08,
+      "loss": 0.1732,
+      "step": 15171
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.572602105676893e-08,
+      "loss": 0.1798,
+      "step": 15172
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.508878028264274e-08,
+      "loss": 0.1847,
+      "step": 15173
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.4456010033933074e-08,
+      "loss": 0.1817,
+      "step": 15174
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.382771033895061e-08,
+      "loss": 0.1763,
+      "step": 15175
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.320388122580621e-08,
+      "loss": 0.1705,
+      "step": 15176
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.2584522722377564e-08,
+      "loss": 0.1796,
+      "step": 15177
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1969634856386944e-08,
+      "loss": 0.1883,
+      "step": 15178
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.135921765532347e-08,
+      "loss": 0.1715,
+      "step": 15179
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.0753271146487524e-08,
+      "loss": 0.1681,
+      "step": 15180
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.015179535699076e-08,
+      "loss": 0.1725,
+      "step": 15181
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.955479031373388e-08,
+      "loss": 0.1803,
+      "step": 15182
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8962256043406644e-08,
+      "loss": 0.1841,
+      "step": 15183
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.837419257251007e-08,
+      "loss": 0.168,
+      "step": 15184
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.779059992735645e-08,
+      "loss": 0.1643,
+      "step": 15185
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.721147813404713e-08,
+      "loss": 0.1809,
+      "step": 15186
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6636827218483606e-08,
+      "loss": 0.1791,
+      "step": 15187
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.606664720635644e-08,
+      "loss": 0.1675,
+      "step": 15188
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5500938123178563e-08,
+      "loss": 0.1744,
+      "step": 15189
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.493969999424085e-08,
+      "loss": 0.1823,
+      "step": 15190
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.438293284464544e-08,
+      "loss": 0.1741,
+      "step": 15191
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.383063669930575e-08,
+      "loss": 0.1781,
+      "step": 15192
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.328281158292423e-08,
+      "loss": 0.1745,
+      "step": 15193
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.273945751999241e-08,
+      "loss": 0.1704,
+      "step": 15194
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.2200574534801967e-08,
+      "loss": 0.1814,
+      "step": 15195
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.166616265147804e-08,
+      "loss": 0.1713,
+      "step": 15196
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.113622189391263e-08,
+      "loss": 0.172,
+      "step": 15197
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.061075228580901e-08,
+      "loss": 0.1694,
+      "step": 15198
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.008975385065948e-08,
+      "loss": 0.1848,
+      "step": 15199
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.957322661177875e-08,
+      "loss": 0.1785,
+      "step": 15200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.9061170592270536e-08,
+      "loss": 0.1714,
+      "step": 15201
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8553585815016547e-08,
+      "loss": 0.176,
+      "step": 15202
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8050472302743046e-08,
+      "loss": 0.1742,
+      "step": 15203
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.755183007793205e-08,
+      "loss": 0.1809,
+      "step": 15204
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.705765916289904e-08,
+      "loss": 0.1768,
+      "step": 15205
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.6567959579737457e-08,
+      "loss": 0.1688,
+      "step": 15206
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.6082731350363098e-08,
+      "loss": 0.1683,
+      "step": 15207
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5601974496458625e-08,
+      "loss": 0.1836,
+      "step": 15208
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5125689039540156e-08,
+      "loss": 0.1761,
+      "step": 15209
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4653875000901773e-08,
+      "loss": 0.1802,
+      "step": 15210
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.418653240164881e-08,
+      "loss": 0.1679,
+      "step": 15211
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.372366126268677e-08,
+      "loss": 0.1733,
+      "step": 15212
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.326526160471021e-08,
+      "loss": 0.1663,
+      "step": 15213
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2811333448224946e-08,
+      "loss": 0.173,
+      "step": 15214
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2361876813536965e-08,
+      "loss": 0.1696,
+      "step": 15215
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1916891720741296e-08,
+      "loss": 0.1732,
+      "step": 15216
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1476378189755343e-08,
+      "loss": 0.1789,
+      "step": 15217
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1040336240252256e-08,
+      "loss": 0.1823,
+      "step": 15218
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.060876589176086e-08,
+      "loss": 0.1781,
+      "step": 15219
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0181667163576834e-08,
+      "loss": 0.173,
+      "step": 15220
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9759040074796008e-08,
+      "loss": 0.1689,
+      "step": 15221
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9340884644325486e-08,
+      "loss": 0.1728,
+      "step": 15222
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.892720089087252e-08,
+      "loss": 0.1752,
+      "step": 15223
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8517988832922328e-08,
+      "loss": 0.1677,
+      "step": 15224
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8113248488793588e-08,
+      "loss": 0.1945,
+      "step": 15225
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7712979876582937e-08,
+      "loss": 0.1794,
+      "step": 15226
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7317183014198268e-08,
+      "loss": 0.1796,
+      "step": 15227
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.692585791932544e-08,
+      "loss": 0.1693,
+      "step": 15228
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6539004609472662e-08,
+      "loss": 0.1712,
+      "step": 15229
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6156623101959423e-08,
+      "loss": 0.1768,
+      "step": 15230
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5778713413872047e-08,
+      "loss": 0.1775,
+      "step": 15231
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5405275562108133e-08,
+      "loss": 0.1702,
+      "step": 15232
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5036309563387642e-08,
+      "loss": 0.1805,
+      "step": 15233
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4671815434197378e-08,
+      "loss": 0.1796,
+      "step": 15234
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.431179319083542e-08,
+      "loss": 0.1734,
+      "step": 15235
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3956242849422208e-08,
+      "loss": 0.1772,
+      "step": 15236
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3605164425856132e-08,
+      "loss": 0.1771,
+      "step": 15237
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3258557935824644e-08,
+      "loss": 0.1712,
+      "step": 15238
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2916423394837563e-08,
+      "loss": 0.1664,
+      "step": 15239
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.257876081820486e-08,
+      "loss": 0.1808,
+      "step": 15240
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2245570221025571e-08,
+      "loss": 0.1601,
+      "step": 15241
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1916851618187785e-08,
+      "loss": 0.1789,
+      "step": 15242
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1592605024413061e-08,
+      "loss": 0.178,
+      "step": 15243
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.127283045418981e-08,
+      "loss": 0.1796,
+      "step": 15244
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0957527921828803e-08,
+      "loss": 0.1834,
+      "step": 15245
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0646697441429876e-08,
+      "loss": 0.1731,
+      "step": 15246
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0340339026881918e-08,
+      "loss": 0.1678,
+      "step": 15247
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0038452691907286e-08,
+      "loss": 0.176,
+      "step": 15248
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.74103845000629e-09,
+      "loss": 0.1816,
+      "step": 15249
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.4480963144572e-09,
+      "loss": 0.187,
+      "step": 15250
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.159626298393953e-09,
+      "loss": 0.1778,
+      "step": 15251
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.875628414695137e-09,
+      "loss": 0.1848,
+      "step": 15252
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.596102676072804e-09,
+      "loss": 0.1714,
+      "step": 15253
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.321049095016964e-09,
+      "loss": 0.1761,
+      "step": 15254
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.050467683851093e-09,
+      "loss": 0.1819,
+      "step": 15255
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.784358454665518e-09,
+      "loss": 0.1795,
+      "step": 15256
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.522721419350732e-09,
+      "loss": 0.1651,
+      "step": 15257
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.265556589630684e-09,
+      "loss": 0.1746,
+      "step": 15258
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.012863976996187e-09,
+      "loss": 0.1763,
+      "step": 15259
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.7646435927382065e-09,
+      "loss": 0.1753,
+      "step": 15260
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.520895447981179e-09,
+      "loss": 0.1705,
+      "step": 15261
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.281619553605289e-09,
+      "loss": 0.1693,
+      "step": 15262
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.0468159203241895e-09,
+      "loss": 0.173,
+      "step": 15263
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.816484558640589e-09,
+      "loss": 0.1673,
+      "step": 15264
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.590625478846256e-09,
+      "loss": 0.1741,
+      "step": 15265
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.369238691055323e-09,
+      "loss": 0.1731,
+      "step": 15266
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.152324205170978e-09,
+      "loss": 0.1776,
+      "step": 15267
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.939882030874365e-09,
+      "loss": 0.1808,
+      "step": 15268
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.731912177680098e-09,
+      "loss": 0.1802,
+      "step": 15269
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.5284146548918435e-09,
+      "loss": 0.1803,
+      "step": 15270
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.329389471613432e-09,
+      "loss": 0.181,
+      "step": 15271
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.134836636737749e-09,
+      "loss": 0.1708,
+      "step": 15272
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.944756158968943e-09,
+      "loss": 0.1828,
+      "step": 15273
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.759148046811323e-09,
+      "loss": 0.1729,
+      "step": 15274
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.5780123085693563e-09,
+      "loss": 0.1801,
+      "step": 15275
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.4013489523365695e-09,
+      "loss": 0.1808,
+      "step": 15276
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.2291579860177503e-09,
+      "loss": 0.1839,
+      "step": 15277
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.0614394173178462e-09,
+      "loss": 0.1801,
+      "step": 15278
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.898193253719761e-09,
+      "loss": 0.1704,
+      "step": 15279
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.739419502550966e-09,
+      "loss": 0.174,
+      "step": 15280
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.585118170905787e-09,
+      "loss": 0.1681,
+      "step": 15281
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.4352892656676064e-09,
+      "loss": 0.1753,
+      "step": 15282
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.2899327935532734e-09,
+      "loss": 0.1703,
+      "step": 15283
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.149048761068695e-09,
+      "loss": 0.1733,
+      "step": 15284
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0126371745088356e-09,
+      "loss": 0.1688,
+      "step": 15285
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.88069803996882e-09,
+      "loss": 0.1619,
+      "step": 15286
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.7532313633439324e-09,
+      "loss": 0.1711,
+      "step": 15287
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6302371503629233e-09,
+      "loss": 0.1737,
+      "step": 15288
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5117154064991922e-09,
+      "loss": 0.1668,
+      "step": 15289
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3976661370707077e-09,
+      "loss": 0.1749,
+      "step": 15290
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2880893471622912e-09,
+      "loss": 0.1796,
+      "step": 15291
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1829850416922305e-09,
+      "loss": 0.1815,
+      "step": 15292
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0823532253456669e-09,
+      "loss": 0.1804,
+      "step": 15293
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.861939026301058e-10,
+      "loss": 0.1868,
+      "step": 15294
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.945070778532127e-10,
+      "loss": 0.1741,
+      "step": 15295
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.072927551117104e-10,
+      "loss": 0.1764,
+      "step": 15296
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.245509383024818e-10,
+      "loss": 0.1719,
+      "step": 15297
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.462816311225695e-10,
+      "loss": 0.1771,
+      "step": 15298
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.724848370802782e-10,
+      "loss": 0.1709,
+      "step": 15299
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.031605594840727e-10,
+      "loss": 0.174,
+      "step": 15300
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.3830880140927065e-10,
+      "loss": 0.1709,
+      "step": 15301
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.7792956577575865e-10,
+      "loss": 0.1764,
+      "step": 15302
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.2202285528137866e-10,
+      "loss": 0.1839,
+      "step": 15303
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.705886724241324e-10,
+      "loss": 0.1752,
+      "step": 15304
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.236270195132839e-10,
+      "loss": 0.1759,
+      "step": 15305
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.8113789862495012e-10,
+      "loss": 0.1691,
+      "step": 15306
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4312131169091913e-10,
+      "loss": 0.1769,
+      "step": 15307
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0957726038762772e-10,
+      "loss": 0.1718,
+      "step": 15308
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.05057462249792e-11,
+      "loss": 0.172,
+      "step": 15309
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.590677050193449e-11,
+      "loss": 0.1679,
+      "step": 15310
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.5780334328716635e-11,
+      "loss": 0.1659,
+      "step": 15311
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0126438582401818e-11,
+      "loss": 0.1914,
+      "step": 15312
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.945083984635005e-12,
+      "loss": 0.1743,
+      "step": 15313
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.236271023914327e-12,
+      "loss": 0.1768,
+      "step": 15314
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.1623,
+      "step": 15315
+    },
+    {
+      "epoch": 5.0,
+      "step": 15315,
+      "total_flos": 1.361034726146048e+16,
+      "train_loss": 0.6721667354334775,
+      "train_runtime": 374267.6092,
+      "train_samples_per_second": 13.745,
+      "train_steps_per_second": 0.041
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 15315,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "total_flos": 1.361034726146048e+16,
+  "train_batch_size": 56,
+  "trial_name": null,
+  "trial_params": null
+}