diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8191349934469201,
+  "global_step": 2500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0869565217391305e-07,
+      "loss": 3.5098,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.173913043478261e-07,
+      "loss": 3.373,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2608695652173915e-07,
+      "loss": 3.4766,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.347826086956522e-07,
+      "loss": 3.6875,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.434782608695653e-07,
+      "loss": 3.8203,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.521739130434783e-07,
+      "loss": 3.752,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.608695652173914e-07,
+      "loss": 3.2876,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 3.2578,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.782608695652175e-07,
+      "loss": 3.4336,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0869565217391306e-06,
+      "loss": 3.1465,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1956521739130436e-06,
+      "loss": 2.9688,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3043478260869566e-06,
+      "loss": 3.1875,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4130434782608697e-06,
+      "loss": 2.5996,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.521739130434783e-06,
+      "loss": 2.6514,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6304347826086957e-06,
+      "loss": 2.5254,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 2.1348,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8478260869565218e-06,
+      "loss": 2.8545,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.956521739130435e-06,
+      "loss": 2.5352,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.065217391304348e-06,
+      "loss": 2.8105,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 2.8623,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.282608695652174e-06,
+      "loss": 2.3516,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.391304347826087e-06,
+      "loss": 2.3809,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 2.4727,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 2.4541,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7173913043478263e-06,
+      "loss": 2.3037,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8260869565217393e-06,
+      "loss": 2.7344,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9347826086956528e-06,
+      "loss": 2.4229,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.043478260869566e-06,
+      "loss": 2.3594,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.152173913043479e-06,
+      "loss": 2.2148,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2608695652173914e-06,
+      "loss": 2.2988,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3695652173913045e-06,
+      "loss": 2.4619,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 2.1543,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5869565217391305e-06,
+      "loss": 2.2715,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6956521739130436e-06,
+      "loss": 2.2617,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.804347826086957e-06,
+      "loss": 2.2588,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.91304347826087e-06,
+      "loss": 2.375,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.021739130434783e-06,
+      "loss": 1.9697,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.130434782608696e-06,
+      "loss": 2.3027,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.239130434782609e-06,
+      "loss": 2.0322,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 2.1172,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.456521739130435e-06,
+      "loss": 2.1104,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.565217391304348e-06,
+      "loss": 2.0566,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.673913043478261e-06,
+      "loss": 2.2617,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 2.1748,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.891304347826087e-06,
+      "loss": 1.9893,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 2.2793,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.108695652173914e-06,
+      "loss": 2.1406,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 2.2217,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3260869565217395e-06,
+      "loss": 2.0264,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4347826086956525e-06,
+      "loss": 1.9844,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.543478260869566e-06,
+      "loss": 1.9258,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 2.2261,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.760869565217392e-06,
+      "loss": 2.1025,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8695652173913055e-06,
+      "loss": 1.624,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.978260869565218e-06,
+      "loss": 2.1729,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 2.0381,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.195652173913044e-06,
+      "loss": 2.5352,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.304347826086958e-06,
+      "loss": 2.125,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.41304347826087e-06,
+      "loss": 2.125,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 2.2002,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.630434782608696e-06,
+      "loss": 1.9238,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.739130434782609e-06,
+      "loss": 1.9717,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.847826086956523e-06,
+      "loss": 2.4297,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 1.7666,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.065217391304349e-06,
+      "loss": 1.9883,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.173913043478261e-06,
+      "loss": 2.1426,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.282608695652175e-06,
+      "loss": 2.1621,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.391304347826087e-06,
+      "loss": 1.9844,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.2812,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.608695652173914e-06,
+      "loss": 2.2227,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.717391304347827e-06,
+      "loss": 2.0244,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 2.0933,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.934782608695653e-06,
+      "loss": 2.2021,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.043478260869566e-06,
+      "loss": 1.8457,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.15217391304348e-06,
+      "loss": 2.0918,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.260869565217392e-06,
+      "loss": 2.0879,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.369565217391305e-06,
+      "loss": 2.0791,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.478260869565218e-06,
+      "loss": 2.1357,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.586956521739131e-06,
+      "loss": 2.0518,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 2.001,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.804347826086957e-06,
+      "loss": 2.0811,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.91304347826087e-06,
+      "loss": 2.2227,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.021739130434784e-06,
+      "loss": 2.2148,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 2.165,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.23913043478261e-06,
+      "loss": 1.9678,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.347826086956523e-06,
+      "loss": 1.9092,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.456521739130436e-06,
+      "loss": 2.3535,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 2.0039,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.673913043478262e-06,
+      "loss": 2.0381,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.782608695652175e-06,
+      "loss": 2.0078,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.891304347826088e-06,
+      "loss": 2.1963,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.8994,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0108695652173914e-05,
+      "loss": 2.0742,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0217391304347829e-05,
+      "loss": 2.2666,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.032608695652174e-05,
+      "loss": 2.1152,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 2.4111,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0543478260869566e-05,
+      "loss": 1.999,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0652173913043479e-05,
+      "loss": 1.8872,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.076086956521739e-05,
+      "loss": 2.0254,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 2.0547,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0978260869565218e-05,
+      "loss": 2.25,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1086956521739131e-05,
+      "loss": 1.8916,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1195652173913046e-05,
+      "loss": 2.3809,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 2.3926,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.141304347826087e-05,
+      "loss": 2.1318,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1521739130434783e-05,
+      "loss": 2.0449,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1630434782608698e-05,
+      "loss": 1.9824,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1739130434782611e-05,
+      "loss": 2.1475,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1847826086956522e-05,
+      "loss": 2.165,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1956521739130435e-05,
+      "loss": 2.0576,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2065217391304348e-05,
+      "loss": 2.1367,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 2.1885,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2282608695652175e-05,
+      "loss": 2.0986,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2391304347826088e-05,
+      "loss": 1.9697,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.25e-05,
+      "loss": 1.9858,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2608695652173915e-05,
+      "loss": 2.0576,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2717391304347828e-05,
+      "loss": 2.1787,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.282608695652174e-05,
+      "loss": 1.9268,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2934782608695653e-05,
+      "loss": 2.2393,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 2.0537,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.315217391304348e-05,
+      "loss": 2.1299,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3260869565217392e-05,
+      "loss": 2.209,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3369565217391305e-05,
+      "loss": 2.0225,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3478260869565218e-05,
+      "loss": 2.168,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3586956521739133e-05,
+      "loss": 2.0947,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3695652173913046e-05,
+      "loss": 2.4922,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3804347826086957e-05,
+      "loss": 2.0234,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 1.917,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4021739130434783e-05,
+      "loss": 1.9824,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4130434782608698e-05,
+      "loss": 1.9883,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.423913043478261e-05,
+      "loss": 1.8809,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4347826086956522e-05,
+      "loss": 2.1211,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4456521739130435e-05,
+      "loss": 2.2988,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.456521739130435e-05,
+      "loss": 1.9434,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4673913043478263e-05,
+      "loss": 1.6143,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 1.7671,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4891304347826087e-05,
+      "loss": 1.771,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.9727,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5108695652173915e-05,
+      "loss": 1.9775,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 1.8193,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.532608695652174e-05,
+      "loss": 2.1982,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5434782608695654e-05,
+      "loss": 2.0381,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5543478260869567e-05,
+      "loss": 1.8589,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 2.2002,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.576086956521739e-05,
+      "loss": 1.8608,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5869565217391306e-05,
+      "loss": 2.3086,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.597826086956522e-05,
+      "loss": 2.2061,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6086956521739132e-05,
+      "loss": 1.9883,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6195652173913045e-05,
+      "loss": 2.4795,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.630434782608696e-05,
+      "loss": 2.0879,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.641304347826087e-05,
+      "loss": 1.8916,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 1.7031,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6630434782608698e-05,
+      "loss": 2.1543,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.673913043478261e-05,
+      "loss": 2.1133,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6847826086956524e-05,
+      "loss": 2.1768,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6956521739130437e-05,
+      "loss": 2.1631,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.706521739130435e-05,
+      "loss": 1.8042,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7173913043478263e-05,
+      "loss": 1.9121,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7282608695652176e-05,
+      "loss": 1.9863,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 1.5146,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 2.2178,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7608695652173915e-05,
+      "loss": 2.2754,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7717391304347828e-05,
+      "loss": 2.0859,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.782608695652174e-05,
+      "loss": 2.0459,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7934782608695654e-05,
+      "loss": 1.7715,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8043478260869567e-05,
+      "loss": 2.0771,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.815217391304348e-05,
+      "loss": 2.1445,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 2.2153,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8369565217391306e-05,
+      "loss": 2.2529,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.847826086956522e-05,
+      "loss": 2.0293,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8586956521739132e-05,
+      "loss": 2.0557,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8695652173913045e-05,
+      "loss": 2.2266,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8804347826086958e-05,
+      "loss": 1.8301,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.891304347826087e-05,
+      "loss": 2.0547,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9021739130434784e-05,
+      "loss": 2.0254,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 2.0508,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.923913043478261e-05,
+      "loss": 2.1895,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9347826086956523e-05,
+      "loss": 2.123,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9456521739130436e-05,
+      "loss": 1.9971,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.956521739130435e-05,
+      "loss": 1.9971,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9673913043478263e-05,
+      "loss": 2.1318,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9782608695652176e-05,
+      "loss": 2.0312,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989130434782609e-05,
+      "loss": 2.1934,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 1.8193,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999999859192326e-05,
+      "loss": 2.1797,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999994367693416e-05,
+      "loss": 2.1416,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999998732731167e-05,
+      "loss": 2.3936,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999977470780006e-05,
+      "loss": 2.1406,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999964798101195e-05,
+      "loss": 1.8633,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999994930927881e-05,
+      "loss": 1.9023,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99999310043172e-05,
+      "loss": 2.0059,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999909883221535e-05,
+      "loss": 2.2256,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999988594599776e-05,
+      "loss": 1.9834,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999859192652612e-05,
+      "loss": 2.0918,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999829623193626e-05,
+      "loss": 2.1992,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999979723762913e-05,
+      "loss": 1.9922,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999762035968245e-05,
+      "loss": 2.1221,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999972401822088e-05,
+      "loss": 2.415,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999683184397752e-05,
+      "loss": 2.3311,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999963953451035e-05,
+      "loss": 2.3125,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999593068570968e-05,
+      "loss": 1.6787,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999543786592696e-05,
+      "loss": 2.0,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999949168858941e-05,
+      "loss": 1.8535,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999436774575783e-05,
+      "loss": 2.2383,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999379044567275e-05,
+      "loss": 2.0884,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999931849858015e-05,
+      "loss": 1.9854,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999255136631453e-05,
+      "loss": 1.9521,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999918895873903e-05,
+      "loss": 1.9385,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999911996492152e-05,
+      "loss": 2.3701,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999048155198348e-05,
+      "loss": 1.998,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999897352958974e-05,
+      "loss": 2.1562,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998896088116714e-05,
+      "loss": 2.0127,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999881583080108e-05,
+      "loss": 2.1523,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998732757665428e-05,
+      "loss": 2.1416,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999864686873316e-05,
+      "loss": 2.0068,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998558164028466e-05,
+      "loss": 2.0479,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998466643576323e-05,
+      "loss": 2.0947,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999837230740251e-05,
+      "loss": 2.3496,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998275155533585e-05,
+      "loss": 2.3486,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998175187996916e-05,
+      "loss": 1.9414,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998072404820648e-05,
+      "loss": 1.8174,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999796680603373e-05,
+      "loss": 1.9917,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997858391665906e-05,
+      "loss": 2.0361,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997747161747696e-05,
+      "loss": 2.1035,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999763311631043e-05,
+      "loss": 1.9443,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997516255386224e-05,
+      "loss": 2.2559,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997396579007987e-05,
+      "loss": 1.9512,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997274087209423e-05,
+      "loss": 1.9863,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997148780025028e-05,
+      "loss": 2.4258,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999702065749009e-05,
+      "loss": 2.1436,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996889719640688e-05,
+      "loss": 2.0918,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996755966513702e-05,
+      "loss": 1.998,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996619398146793e-05,
+      "loss": 2.084,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996480014578422e-05,
+      "loss": 2.2363,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999633781584784e-05,
+      "loss": 1.9902,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99961928019951e-05,
+      "loss": 2.2803,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996044973061028e-05,
+      "loss": 2.0156,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995894329087265e-05,
+      "loss": 2.1641,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995740870116233e-05,
+      "loss": 2.2646,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995584596191145e-05,
+      "loss": 2.2617,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999542550735601e-05,
+      "loss": 2.0605,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995263603655638e-05,
+      "loss": 1.7417,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995098885135612e-05,
+      "loss": 1.9326,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994931351842327e-05,
+      "loss": 2.251,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994761003822958e-05,
+      "loss": 1.9111,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999458784112548e-05,
+      "loss": 2.2578,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999441186379866e-05,
+      "loss": 1.9678,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994233071892056e-05,
+      "loss": 2.1523,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994051465456014e-05,
+      "loss": 2.0283,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999386704454168e-05,
+      "loss": 2.5684,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999367980920099e-05,
+      "loss": 1.9521,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993489759486672e-05,
+      "loss": 2.3232,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993296895452247e-05,
+      "loss": 2.1562,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999310121715203e-05,
+      "loss": 2.1182,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992902724641127e-05,
+      "loss": 2.1094,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999270141797543e-05,
+      "loss": 2.0459,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999249729721164e-05,
+      "loss": 1.9785,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992290362407232e-05,
+      "loss": 2.0215,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992080613620486e-05,
+      "loss": 2.1875,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999186805091047e-05,
+      "loss": 2.1221,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991652674337047e-05,
+      "loss": 2.2158,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991434483960866e-05,
+      "loss": 2.3223,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991213479843377e-05,
+      "loss": 1.8828,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999098966204682e-05,
+      "loss": 2.3281,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990763030634214e-05,
+      "loss": 2.3145,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990533585669393e-05,
+      "loss": 2.0264,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999030132721697e-05,
+      "loss": 2.2891,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990066255342348e-05,
+      "loss": 1.9785,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989828370111737e-05,
+      "loss": 2.1436,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989587671592116e-05,
+      "loss": 1.9121,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998934415985128e-05,
+      "loss": 1.8262,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989097834957798e-05,
+      "loss": 2.1318,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988848696981046e-05,
+      "loss": 2.0254,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998859674599118e-05,
+      "loss": 2.0742,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988341982059156e-05,
+      "loss": 1.8857,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988084405256712e-05,
+      "loss": 2.1006,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99878240156564e-05,
+      "loss": 2.1611,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987560813331535e-05,
+      "loss": 2.4092,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998729479835625e-05,
+      "loss": 2.0596,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998702597080545e-05,
+      "loss": 2.082,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986754330754847e-05,
+      "loss": 2.0537,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986479878280936e-05,
+      "loss": 2.4346,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998620261346101e-05,
+      "loss": 2.2295,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998592253637315e-05,
+      "loss": 1.8906,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985639647096228e-05,
+      "loss": 2.1475,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985353945709912e-05,
+      "loss": 2.1396,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985065432294656e-05,
+      "loss": 1.7705,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984774106931714e-05,
+      "loss": 1.8828,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998447996970313e-05,
+      "loss": 2.2529,
+      "step": 289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998418302069173e-05,
+      "loss": 2.1436,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983883259981144e-05,
+      "loss": 1.8105,
+      "step": 291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983580687655792e-05,
+      "loss": 2.2979,
+      "step": 292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983275303800876e-05,
+      "loss": 2.0557,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99829671085024e-05,
+      "loss": 2.3799,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998265610184716e-05,
+      "loss": 2.1445,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998234228392274e-05,
+      "loss": 2.2471,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982025654817507e-05,
+      "loss": 2.1172,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981706214620637e-05,
+      "loss": 2.1582,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981383963422086e-05,
+      "loss": 1.8242,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981058901312607e-05,
+      "loss": 2.1982,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980731028383743e-05,
+      "loss": 1.9766,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980400344727826e-05,
+      "loss": 2.0781,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998006685043798e-05,
+      "loss": 2.4814,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979730545608128e-05,
+      "loss": 2.0078,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997939143033297e-05,
+      "loss": 2.3799,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979049504708014e-05,
+      "loss": 1.9502,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997870476882955e-05,
+      "loss": 2.1377,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978357222794657e-05,
+      "loss": 2.3281,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978006866701212e-05,
+      "loss": 1.938,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977653700647878e-05,
+      "loss": 2.0537,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977297724734117e-05,
+      "loss": 2.1143,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976938939060172e-05,
+      "loss": 2.127,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997657734372709e-05,
+      "loss": 2.1367,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976212938836692e-05,
+      "loss": 2.0322,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975845724491606e-05,
+      "loss": 2.1465,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975475700795247e-05,
+      "loss": 2.0234,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997510286785181e-05,
+      "loss": 2.127,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974727225766303e-05,
+      "loss": 1.9902,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974348774644503e-05,
+      "loss": 1.8799,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9973967514592995e-05,
+      "loss": 1.8262,
+      "step": 320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997358344571914e-05,
+      "loss": 2.2158,
+      "step": 321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973196568131103e-05,
+      "loss": 2.0361,
+      "step": 322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972806881937833e-05,
+      "loss": 2.3535,
+      "step": 323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972414387249074e-05,
+      "loss": 2.4307,
+      "step": 324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972019084175352e-05,
+      "loss": 2.1211,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971620972827997e-05,
+      "loss": 2.0605,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997122005331912e-05,
+      "loss": 2.1113,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970816325761627e-05,
+      "loss": 2.2988,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970409790269216e-05,
+      "loss": 2.0752,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970000446956373e-05,
+      "loss": 2.1416,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996958829593837e-05,
+      "loss": 2.1689,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9969173337331283e-05,
+      "loss": 1.8145,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968755571251964e-05,
+      "loss": 2.0918,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968334997818062e-05,
+      "loss": 2.2637,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967911617148025e-05,
+      "loss": 1.917,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967485429361076e-05,
+      "loss": 2.0781,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996705643457724e-05,
+      "loss": 2.4746,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966624632917325e-05,
+      "loss": 2.1064,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996619002450294e-05,
+      "loss": 2.1172,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965752609456465e-05,
+      "loss": 1.7559,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965312387901096e-05,
+      "loss": 1.9541,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964869359960797e-05,
+      "loss": 2.1582,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964423525760338e-05,
+      "loss": 2.1045,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963974885425267e-05,
+      "loss": 2.1348,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996352343908193e-05,
+      "loss": 2.002,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963069186857464e-05,
+      "loss": 2.3345,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996261212887979e-05,
+      "loss": 2.3975,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962152265277623e-05,
+      "loss": 1.709,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996168959618047e-05,
+      "loss": 2.1895,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961224121718625e-05,
+      "loss": 2.3027,
+      "step": 350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960755842023168e-05,
+      "loss": 2.0186,
+      "step": 351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996028475722598e-05,
+      "loss": 2.0762,
+      "step": 352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959810867459725e-05,
+      "loss": 2.0391,
+      "step": 353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959334172857852e-05,
+      "loss": 2.0,
+      "step": 354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995885467355461e-05,
+      "loss": 1.957,
+      "step": 355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958372369685035e-05,
+      "loss": 2.0068,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995788726138495e-05,
+      "loss": 2.1836,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957399348790966e-05,
+      "loss": 2.0068,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956908632040492e-05,
+      "loss": 2.1396,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995641511127171e-05,
+      "loss": 2.0557,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955918786623617e-05,
+      "loss": 2.1875,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995541965823598e-05,
+      "loss": 2.0107,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995491772624936e-05,
+      "loss": 1.918,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954412990805107e-05,
+      "loss": 1.9424,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995390545204537e-05,
+      "loss": 1.9033,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953395110113068e-05,
+      "loss": 1.9902,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995288196515193e-05,
+      "loss": 1.96,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952366017306465e-05,
+      "loss": 2.4434,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995184726672197e-05,
+      "loss": 1.8838,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995132571354453e-05,
+      "loss": 2.0312,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995080135792103e-05,
+      "loss": 2.0488,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950274199999132e-05,
+      "loss": 2.54,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994974423992729e-05,
+      "loss": 2.1133,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949211477854747e-05,
+      "loss": 2.1504,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948675913931548e-05,
+      "loss": 2.1445,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948137548308504e-05,
+      "loss": 2.1504,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947596381137234e-05,
+      "loss": 1.9238,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947052412570134e-05,
+      "loss": 1.9854,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946505642760398e-05,
+      "loss": 2.0205,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9945956071862006e-05,
+      "loss": 2.0195,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994540370002972e-05,
+      "loss": 2.1445,
+      "step": 381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944848527419102e-05,
+      "loss": 1.7832,
+      "step": 382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944290554186492e-05,
+      "loss": 2.3232,
+      "step": 383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994372978048903e-05,
+      "loss": 2.4082,
+      "step": 384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994316620648463e-05,
+      "loss": 1.9551,
+      "step": 385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9942599832332008e-05,
+      "loss": 2.0264,
+      "step": 386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9942030658190667e-05,
+      "loss": 2.0215,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941458684220893e-05,
+      "loss": 2.1953,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940883910583757e-05,
+      "loss": 1.9316,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994030633744113e-05,
+      "loss": 2.0361,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9939725964955665e-05,
+      "loss": 2.2285,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99391427932908e-05,
+      "loss": 2.0996,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993855682261077e-05,
+      "loss": 2.1709,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993796805308059e-05,
+      "loss": 1.9951,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993737648486607e-05,
+      "loss": 2.0156,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99367821181338e-05,
+      "loss": 2.3574,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993618495305117e-05,
+      "loss": 2.2754,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993558498978634e-05,
+      "loss": 2.1084,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934982228508278e-05,
+      "loss": 2.2793,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934376669386727e-05,
+      "loss": 2.4463,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993376831259222e-05,
+      "loss": 2.0781,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993315715829609e-05,
+      "loss": 2.0918,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932543206670432e-05,
+      "loss": 1.8262,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931926457888155e-05,
+      "loss": 1.8423,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993130691212294e-05,
+      "loss": 2.2236,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930684569549265e-05,
+      "loss": 2.2275,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930059430342386e-05,
+      "loss": 1.9883,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9929431494678355e-05,
+      "loss": 2.1543,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928800762734007e-05,
+      "loss": 1.9902,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928167234686968e-05,
+      "loss": 1.9814,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9927530910715646e-05,
+      "loss": 2.1338,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9926891790999244e-05,
+      "loss": 1.8916,
+      "step": 412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992624987571774e-05,
+      "loss": 1.9346,
+      "step": 413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9925605165051917e-05,
+      "loss": 1.7397,
+      "step": 414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992495765918333e-05,
+      "loss": 1.9658,
+      "step": 415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924307358294324e-05,
+      "loss": 2.2119,
+      "step": 416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992365426256804e-05,
+      "loss": 2.1729,
+      "step": 417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922998372188398e-05,
+      "loss": 2.1602,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922339687340102e-05,
+      "loss": 1.8521,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921678208208653e-05,
+      "loss": 1.9678,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921013934980334e-05,
+      "loss": 1.875,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992034686784221e-05,
+      "loss": 2.0332,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9919677006982142e-05,
+      "loss": 2.3252,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9919004352588768e-05,
+      "loss": 2.1191,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918328904851523e-05,
+      "loss": 1.6631,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917650663960624e-05,
+      "loss": 2.2373,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991696963010707e-05,
+      "loss": 2.3613,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991628580348265e-05,
+      "loss": 1.7754,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915599184279943e-05,
+      "loss": 1.9902,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914909772692307e-05,
+      "loss": 1.9072,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99142175689139e-05,
+      "loss": 2.1543,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9913522573139648e-05,
+      "loss": 2.1875,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991282478556528e-05,
+      "loss": 1.9609,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912124206387297e-05,
+      "loss": 1.729,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9911420835802993e-05,
+      "loss": 1.9326,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910714674010453e-05,
+      "loss": 1.9736,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910005721208542e-05,
+      "loss": 2.1514,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990929397759691e-05,
+      "loss": 2.2549,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908579443375995e-05,
+      "loss": 1.8564,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990786211874702e-05,
+      "loss": 2.1533,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9907142003911997e-05,
+      "loss": 2.1694,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906419099073725e-05,
+      "loss": 2.123,
+      "step": 442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990569340443577e-05,
+      "loss": 2.0635,
+      "step": 443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990496492020252e-05,
+      "loss": 1.8574,
+      "step": 444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990423364657911e-05,
+      "loss": 2.4912,
+      "step": 445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903499583771486e-05,
+      "loss": 2.0186,
+      "step": 446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902762731986372e-05,
+      "loss": 2.1719,
+      "step": 447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902023091431274e-05,
+      "loss": 2.1787,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901280662314483e-05,
+      "loss": 1.9282,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9900535444845084e-05,
+      "loss": 2.3555,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989978743923294e-05,
+      "loss": 2.2861,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98990366456887e-05,
+      "loss": 2.0625,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9898283064423804e-05,
+      "loss": 2.2979,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897526695650458e-05,
+      "loss": 2.0713,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896767539581687e-05,
+      "loss": 2.1299,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896005596431264e-05,
+      "loss": 1.9414,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9895240866413774e-05,
+      "loss": 2.1035,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9894473349744573e-05,
+      "loss": 1.9775,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9893703046639806e-05,
+      "loss": 1.8926,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98929299573164e-05,
+      "loss": 2.1025,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989215408199207e-05,
+      "loss": 2.0215,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989137542088532e-05,
+      "loss": 2.1445,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890593974215425e-05,
+      "loss": 2.1826,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889809742202454e-05,
+      "loss": 2.1475,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889022725067263e-05,
+      "loss": 2.332,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888232923031486e-05,
+      "loss": 1.9766,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9887440336317537e-05,
+      "loss": 2.1475,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9886644965148633e-05,
+      "loss": 2.249,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885846809748754e-05,
+      "loss": 2.4648,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885045870342674e-05,
+      "loss": 1.9258,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988424214715595e-05,
+      "loss": 1.9648,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9883435640414922e-05,
+      "loss": 2.1201,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988262635034672e-05,
+      "loss": 1.8389,
+      "step": 473
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9881814277179248e-05,
+      "loss": 1.8052,
+      "step": 474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.98809994211412e-05,
+      "loss": 2.2969,
+      "step": 475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9880181782462048e-05,
+      "loss": 2.0615,
+      "step": 476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9879361361372056e-05,
+      "loss": 2.2705,
+      "step": 477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878538158102264e-05,
+      "loss": 1.9805,
+      "step": 478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9877712172884504e-05,
+      "loss": 2.1211,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 2.373,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9876051857536288e-05,
+      "loss": 1.7515,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.98752175278734e-05,
+      "loss": 2.0391,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987438041719769e-05,
+      "loss": 1.8301,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9873540525744888e-05,
+      "loss": 1.6855,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9872697853751523e-05,
+      "loss": 1.916,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871852401454908e-05,
+      "loss": 2.1025,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987100416909313e-05,
+      "loss": 1.9531,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987015315690507e-05,
+      "loss": 2.0186,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9869299365130384e-05,
+      "loss": 1.8203,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986844279400951e-05,
+      "loss": 1.9414,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9867583443783674e-05,
+      "loss": 2.0391,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9866721314694883e-05,
+      "loss": 1.9512,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9865856406985923e-05,
+      "loss": 1.9062,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9864988720900367e-05,
+      "loss": 2.0205,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9864118256682575e-05,
+      "loss": 2.2539,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986324501457767e-05,
+      "loss": 2.4736,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986236899483158e-05,
+      "loss": 2.292,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9861490197691004e-05,
+      "loss": 2.373,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986060862340342e-05,
+      "loss": 2.0664,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.98597242722171e-05,
+      "loss": 2.1621,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9858837144381087e-05,
+      "loss": 1.9868,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9857947240145208e-05,
+      "loss": 2.1455,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9857054559760077e-05,
+      "loss": 2.0488,
+      "step": 503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9856159103477085e-05,
+      "loss": 2.084,
+      "step": 504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985526087154841e-05,
+      "loss": 1.9619,
+      "step": 505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9854359864227e-05,
+      "loss": 1.9248,
+      "step": 506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98534560817666e-05,
+      "loss": 2.0586,
+      "step": 507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9852549524421724e-05,
+      "loss": 2.1777,
+      "step": 508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9851640192447675e-05,
+      "loss": 2.0781,
+      "step": 509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9850728086100536e-05,
+      "loss": 1.9473,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9849813205637167e-05,
+      "loss": 2.0293,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848895551315214e-05,
+      "loss": 2.0645,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9847975123393102e-05,
+      "loss": 2.0547,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984705192213004e-05,
+      "loss": 2.083,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984612594778601e-05,
+      "loss": 1.9067,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845197200621786e-05,
+      "loss": 2.002,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9844265680898917e-05,
+      "loss": 2.0967,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843331388879733e-05,
+      "loss": 2.2676,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9842394324827342e-05,
+      "loss": 2.1279,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984145448900564e-05,
+      "loss": 2.0303,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9840511881679293e-05,
+      "loss": 2.3057,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9839566503113755e-05,
+      "loss": 1.9033,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983861835357527e-05,
+      "loss": 1.9082,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983766743333084e-05,
+      "loss": 2.2827,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9836713742648257e-05,
+      "loss": 2.0122,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9835757281796104e-05,
+      "loss": 1.8955,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983479805104373e-05,
+      "loss": 1.9219,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983383605066127e-05,
+      "loss": 1.9775,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9832871280919638e-05,
+      "loss": 1.8472,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983190374209053e-05,
+      "loss": 2.0146,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983093343444641e-05,
+      "loss": 1.8867,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9829960358260546e-05,
+      "loss": 2.3369,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9828984513806963e-05,
+      "loss": 2.3809,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9828005901360476e-05,
+      "loss": 1.6929,
+      "step": 534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9827024521196676e-05,
+      "loss": 2.0684,
+      "step": 535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826040373591934e-05,
+      "loss": 2.0703,
+      "step": 536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9825053458823405e-05,
+      "loss": 2.1992,
+      "step": 537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9824063777169013e-05,
+      "loss": 2.1816,
+      "step": 538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982307132890747e-05,
+      "loss": 2.0771,
+      "step": 539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9822076114318273e-05,
+      "loss": 2.0244,
+      "step": 540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9821078133681676e-05,
+      "loss": 2.2285,
+      "step": 541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9820077387278734e-05,
+      "loss": 1.9893,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9819073875391272e-05,
+      "loss": 1.9961,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9818067598301894e-05,
+      "loss": 1.9893,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981705855629398e-05,
+      "loss": 1.832,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9816046749651694e-05,
+      "loss": 1.9863,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9815032178659977e-05,
+      "loss": 2.2061,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9814014843604546e-05,
+      "loss": 2.1846,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9812994744771898e-05,
+      "loss": 1.876,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9811971882449313e-05,
+      "loss": 2.0898,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981094625692484e-05,
+      "loss": 2.2061,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980991786848731e-05,
+      "loss": 1.9492,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9808886717426336e-05,
+      "loss": 2.0879,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 2.1602,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9806816128596385e-05,
+      "loss": 2.1777,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9805776691410515e-05,
+      "loss": 1.7246,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980473449276742e-05,
+      "loss": 1.9941,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9803689532960603e-05,
+      "loss": 1.8828,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980264181228433e-05,
+      "loss": 1.8438,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980159133103366e-05,
+      "loss": 1.9067,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9800538089504432e-05,
+      "loss": 2.3369,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979948208799325e-05,
+      "loss": 1.731,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9798423326797494e-05,
+      "loss": 1.9922,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9797361806215335e-05,
+      "loss": 2.2773,
+      "step": 564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979629752654571e-05,
+      "loss": 2.1699,
+      "step": 565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9795230488088335e-05,
+      "loss": 2.1484,
+      "step": 566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9794160691143708e-05,
+      "loss": 1.875,
+      "step": 567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97930881360131e-05,
+      "loss": 2.1367,
+      "step": 568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979201282299856e-05,
+      "loss": 1.9736,
+      "step": 569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979093475240291e-05,
+      "loss": 2.4971,
+      "step": 570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9789853924529753e-05,
+      "loss": 2.415,
+      "step": 571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9788770339683462e-05,
+      "loss": 1.8398,
+      "step": 572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97876839981692e-05,
+      "loss": 2.0107,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978659490029289e-05,
+      "loss": 1.7637,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9785503046361237e-05,
+      "loss": 2.1387,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9784408436681732e-05,
+      "loss": 1.8838,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978331107156263e-05,
+      "loss": 2.0381,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9782210951312965e-05,
+      "loss": 1.9951,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9781108076242548e-05,
+      "loss": 1.9873,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9780002446661967e-05,
+      "loss": 2.0732,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9778894062882582e-05,
+      "loss": 2.0234,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977778292521653e-05,
+      "loss": 2.1484,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9776669033976726e-05,
+      "loss": 2.1348,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9775552389476865e-05,
+      "loss": 2.0947,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97744329920314e-05,
+      "loss": 1.792,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9773310841955578e-05,
+      "loss": 2.0996,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977218593956541e-05,
+      "loss": 2.4395,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977105828517769e-05,
+      "loss": 1.9922,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9769927879109982e-05,
+      "loss": 2.1982,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9768794721680622e-05,
+      "loss": 2.1465,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9767658813208725e-05,
+      "loss": 2.2109,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9766520154014184e-05,
+      "loss": 1.9375,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976537874441766e-05,
+      "loss": 2.0869,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9764234584740592e-05,
+      "loss": 2.374,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976308767530519e-05,
+      "loss": 2.1416,
+      "step": 595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.976193801643445e-05,
+      "loss": 1.8467,
+      "step": 596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9760785608452126e-05,
+      "loss": 1.6655,
+      "step": 597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759630451682754e-05,
+      "loss": 1.9771,
+      "step": 598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9758472546451645e-05,
+      "loss": 2.2812,
+      "step": 599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9757311893084887e-05,
+      "loss": 2.0771,
+      "step": 600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975614849190933e-05,
+      "loss": 2.042,
+      "step": 601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9754982343252615e-05,
+      "loss": 1.7131,
+      "step": 602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9753813447443138e-05,
+      "loss": 2.0947,
+      "step": 603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9752641804810083e-05,
+      "loss": 1.8706,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9751467415683403e-05,
+      "loss": 2.1572,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9750290280393824e-05,
+      "loss": 1.9111,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9749110399272843e-05,
+      "loss": 2.0508,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974792777265273e-05,
+      "loss": 1.9482,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974674240086654e-05,
+      "loss": 1.9531,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745554284248086e-05,
+      "loss": 2.0059,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9744363423131957e-05,
+      "loss": 1.8516,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9743169817853526e-05,
+      "loss": 2.2158,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9741973468748922e-05,
+      "loss": 2.0547,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974077437615506e-05,
+      "loss": 2.2998,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9739572540409622e-05,
+      "loss": 1.7061,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9738367961851062e-05,
+      "loss": 2.1211,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973716064081861e-05,
+      "loss": 2.0029,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9735950577652264e-05,
+      "loss": 2.2051,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97347377726928e-05,
+      "loss": 2.1973,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9733522226281762e-05,
+      "loss": 1.9551,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973230393876146e-05,
+      "loss": 2.1543,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973108291047499e-05,
+      "loss": 1.9326,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9729859141766207e-05,
+      "loss": 1.8213,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9728632632979746e-05,
+      "loss": 2.1396,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9727403384461013e-05,
+      "loss": 1.9912,
+      "step": 625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972617139655618e-05,
+      "loss": 2.1006,
+      "step": 626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972493666961219e-05,
+      "loss": 2.001,
+      "step": 627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9723699203976768e-05,
+      "loss": 2.1611,
+      "step": 628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9722458999998398e-05,
+      "loss": 2.0957,
+      "step": 629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972121605802635e-05,
+      "loss": 2.3457,
+      "step": 630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971997037841064e-05,
+      "loss": 2.3486,
+      "step": 631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971872196150208e-05,
+      "loss": 2.0908,
+      "step": 632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9717470807652244e-05,
+      "loss": 2.1289,
+      "step": 633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971621691721348e-05,
+      "loss": 2.002,
+      "step": 634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9714960290538888e-05,
+      "loss": 2.2031,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9713700927982372e-05,
+      "loss": 2.0,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9712438829898574e-05,
+      "loss": 2.167,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971117399664292e-05,
+      "loss": 1.7568,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9709906428571616e-05,
+      "loss": 1.8906,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970863612604162e-05,
+      "loss": 2.1895,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970736308941068e-05,
+      "loss": 2.2031,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970608731903729e-05,
+      "loss": 2.1748,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970480881528073e-05,
+      "loss": 1.7251,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9703527578501052e-05,
+      "loss": 2.1309,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9702243609059068e-05,
+      "loss": 2.3496,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970095690731636e-05,
+      "loss": 2.1191,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9699667473635288e-05,
+      "loss": 1.8608,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9698375308378973e-05,
+      "loss": 2.0469,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9697080411911313e-05,
+      "loss": 2.2461,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9695782784596966e-05,
+      "loss": 1.8154,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9694482426801366e-05,
+      "loss": 2.1357,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9693179338890712e-05,
+      "loss": 1.9385,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9691873521231976e-05,
+      "loss": 2.1777,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9690564974192893e-05,
+      "loss": 2.2051,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9689253698141976e-05,
+      "loss": 2.5039,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9687939693448495e-05,
+      "loss": 2.0088,
+      "step": 656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9686622960482496e-05,
+      "loss": 2.0537,
+      "step": 657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968530349961479e-05,
+      "loss": 2.043,
+      "step": 658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968398131121696e-05,
+      "loss": 2.1611,
+      "step": 659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968265639566135e-05,
+      "loss": 1.9473,
+      "step": 660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9681328753321082e-05,
+      "loss": 2.0557,
+      "step": 661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9679998384570033e-05,
+      "loss": 1.7744,
+      "step": 662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9678665289782862e-05,
+      "loss": 1.8857,
+      "step": 663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967732946933499e-05,
+      "loss": 2.0107,
+      "step": 664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96759909236026e-05,
+      "loss": 2.1504,
+      "step": 665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9674649652962646e-05,
+      "loss": 1.8193,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9673305657792855e-05,
+      "loss": 2.043,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9671958938471715e-05,
+      "loss": 2.0156,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9670609495378484e-05,
+      "loss": 1.9922,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966925732889318e-05,
+      "loss": 1.979,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9667902439396604e-05,
+      "loss": 1.999,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9666544827270302e-05,
+      "loss": 2.0625,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9665184492896606e-05,
+      "loss": 1.9463,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9663821436658607e-05,
+      "loss": 1.7441,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9662455658940158e-05,
+      "loss": 2.1377,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9661087160125888e-05,
+      "loss": 2.3438,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9659715940601182e-05,
+      "loss": 2.167,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9658342000752202e-05,
+      "loss": 2.2051,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9656965340965872e-05,
+      "loss": 2.043,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965558596162987e-05,
+      "loss": 2.1162,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965420386313266e-05,
+      "loss": 2.042,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9652819045863457e-05,
+      "loss": 1.9648,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9651431510212252e-05,
+      "loss": 1.9302,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9650041256569792e-05,
+      "loss": 2.0205,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9648648285327593e-05,
+      "loss": 2.2002,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964725259687794e-05,
+      "loss": 1.9883,
+      "step": 686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964585419161388e-05,
+      "loss": 2.0537,
+      "step": 687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9644453069929227e-05,
+      "loss": 2.1182,
+      "step": 688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643049232218554e-05,
+      "loss": 2.0659,
+      "step": 689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9641642678877205e-05,
+      "loss": 1.9502,
+      "step": 690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9640233410301288e-05,
+      "loss": 2.3516,
+      "step": 691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9638821426887675e-05,
+      "loss": 2.2393,
+      "step": 692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9637406729034e-05,
+      "loss": 2.0137,
+      "step": 693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9635989317138666e-05,
+      "loss": 1.8047,
+      "step": 694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9634569191600837e-05,
+      "loss": 2.0786,
+      "step": 695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963314635282044e-05,
+      "loss": 2.1709,
+      "step": 696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9631720801198173e-05,
+      "loss": 2.0889,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9630292537135487e-05,
+      "loss": 2.0522,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962886156103461e-05,
+      "loss": 2.0137,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9627427873298522e-05,
+      "loss": 1.9824,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9625991474330973e-05,
+      "loss": 2.0469,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9624552364536472e-05,
+      "loss": 1.9395,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.96231105443203e-05,
+      "loss": 2.3867,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9621666014088495e-05,
+      "loss": 2.29,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9620218774247855e-05,
+      "loss": 2.1172,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9618768825205947e-05,
+      "loss": 2.1553,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9617316167371098e-05,
+      "loss": 2.4961,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9615860801152398e-05,
+      "loss": 1.9385,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9614402726959703e-05,
+      "loss": 2.0322,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961294194520363e-05,
+      "loss": 2.0,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9611478456295558e-05,
+      "loss": 2.2275,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961001226064762e-05,
+      "loss": 2.2578,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9608543358672728e-05,
+      "loss": 1.8008,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960707175078454e-05,
+      "loss": 2.1416,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960559743739749e-05,
+      "loss": 1.9404,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9604120418926764e-05,
+      "loss": 1.959,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9602640695788317e-05,
+      "loss": 1.9072,
+      "step": 717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9601158268398855e-05,
+      "loss": 2.1934,
+      "step": 718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9599673137175855e-05,
+      "loss": 2.1147,
+      "step": 719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9598185302537557e-05,
+      "loss": 2.2949,
+      "step": 720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9596694764902954e-05,
+      "loss": 2.0869,
+      "step": 721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9595201524691802e-05,
+      "loss": 2.1836,
+      "step": 722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9593705582324623e-05,
+      "loss": 2.1035,
+      "step": 723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9592206938222703e-05,
+      "loss": 1.9004,
+      "step": 724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9590705592808072e-05,
+      "loss": 2.1348,
+      "step": 725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958920154650354e-05,
+      "loss": 2.0581,
+      "step": 726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9587694799732668e-05,
+      "loss": 1.9438,
+      "step": 727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958618535291978e-05,
+      "loss": 2.127,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9584673206489955e-05,
+      "loss": 1.812,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958315836086904e-05,
+      "loss": 2.2246,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958164081648364e-05,
+      "loss": 1.9961,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958012057376111e-05,
+      "loss": 2.2578,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9578597633129587e-05,
+      "loss": 2.1133,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9577071995017945e-05,
+      "loss": 2.1143,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957554365985583e-05,
+      "loss": 2.0981,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957401262807365e-05,
+      "loss": 2.0332,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9572478900102556e-05,
+      "loss": 2.1592,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9570942476374478e-05,
+      "loss": 2.0615,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956940335732209e-05,
+      "loss": 2.1689,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9567861543378838e-05,
+      "loss": 2.0596,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9566317034978916e-05,
+      "loss": 1.9917,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9564769832557284e-05,
+      "loss": 2.3301,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956321993654966e-05,
+      "loss": 2.0312,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956166734739251e-05,
+      "loss": 2.1807,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9560112065523076e-05,
+      "loss": 2.25,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9558554091379344e-05,
+      "loss": 1.8838,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9556993425400065e-05,
+      "loss": 1.9551,
+      "step": 747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9555430068024746e-05,
+      "loss": 2.2812,
+      "step": 748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9553864019693652e-05,
+      "loss": 1.8906,
+      "step": 749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955229528084781e-05,
+      "loss": 1.9902,
+      "step": 750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9550723851929e-05,
+      "loss": 2.2539,
+      "step": 751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9549149733379757e-05,
+      "loss": 2.1631,
+      "step": 752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9547572925643377e-05,
+      "loss": 1.9521,
+      "step": 753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9545993429163913e-05,
+      "loss": 2.1367,
+      "step": 754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954441124438618e-05,
+      "loss": 2.1543,
+      "step": 755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9542826371755743e-05,
+      "loss": 2.042,
+      "step": 756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9541238811718926e-05,
+      "loss": 1.7124,
+      "step": 757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953964856472281e-05,
+      "loss": 2.2031,
+      "step": 758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9538055631215233e-05,
+      "loss": 2.1211,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9536460011644788e-05,
+      "loss": 2.0557,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953486170646083e-05,
+      "loss": 1.7871,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9533260716113463e-05,
+      "loss": 1.9629,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9531657041053556e-05,
+      "loss": 1.9805,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953005068173272e-05,
+      "loss": 2.127,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9528441638603336e-05,
+      "loss": 2.2148,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9526829912118536e-05,
+      "loss": 1.6875,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9525215502732203e-05,
+      "loss": 1.79,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9523598410898985e-05,
+      "loss": 1.8413,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952197863707427e-05,
+      "loss": 2.1582,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952035618171422e-05,
+      "loss": 2.208,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951873104527574e-05,
+      "loss": 1.7573,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9517103228216495e-05,
+      "loss": 2.0537,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95154727309949e-05,
+      "loss": 2.0977,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9513839554070132e-05,
+      "loss": 2.2217,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951220369790212e-05,
+      "loss": 2.0469,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 2.0811,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950892394967983e-05,
+      "loss": 1.9912,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950728005854918e-05,
+      "loss": 2.4414,
+      "step": 778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9505633490022545e-05,
+      "loss": 2.1211,
+      "step": 779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9503984244563617e-05,
+      "loss": 1.9746,
+      "step": 780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9502332322636844e-05,
+      "loss": 1.6533,
+      "step": 781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950067772470744e-05,
+      "loss": 2.0176,
+      "step": 782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949902045124136e-05,
+      "loss": 2.1152,
+      "step": 783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949736050270532e-05,
+      "loss": 2.2275,
+      "step": 784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9495697879566787e-05,
+      "loss": 1.9502,
+      "step": 785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9494032582293983e-05,
+      "loss": 2.0879,
+      "step": 786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9492364611355877e-05,
+      "loss": 1.7852,
+      "step": 787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94906939672222e-05,
+      "loss": 1.9316,
+      "step": 788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9489020650363427e-05,
+      "loss": 1.7031,
+      "step": 789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948734466125079e-05,
+      "loss": 2.2148,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485666000356275e-05,
+      "loss": 1.9204,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9483984668152618e-05,
+      "loss": 1.7861,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948230066511331e-05,
+      "loss": 2.2705,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948061399171259e-05,
+      "loss": 2.2236,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947892464842545e-05,
+      "loss": 2.375,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9477232635727637e-05,
+      "loss": 1.6421,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9475537954095646e-05,
+      "loss": 2.1562,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947384060400673e-05,
+      "loss": 2.2256,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9472140585938882e-05,
+      "loss": 2.1641,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9470437900370856e-05,
+      "loss": 1.876,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9468732547782156e-05,
+      "loss": 2.0527,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9467024528653036e-05,
+      "loss": 2.0762,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9465313843464496e-05,
+      "loss": 1.9375,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9463600492698297e-05,
+      "loss": 1.8408,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946188447683694e-05,
+      "loss": 1.9531,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9460165796363685e-05,
+      "loss": 1.6777,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9458444451762536e-05,
+      "loss": 2.0684,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945672044351825e-05,
+      "loss": 2.2451,
+      "step": 808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9454993772116336e-05,
+      "loss": 1.8696,
+      "step": 809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945326443804305e-05,
+      "loss": 2.2168,
+      "step": 810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9451532441785398e-05,
+      "loss": 2.1436,
+      "step": 811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944979778383114e-05,
+      "loss": 2.3096,
+      "step": 812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944806046466878e-05,
+      "loss": 2.1484,
+      "step": 813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9446320484787576e-05,
+      "loss": 1.6318,
+      "step": 814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944457784467753e-05,
+      "loss": 2.0811,
+      "step": 815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9442832544829397e-05,
+      "loss": 2.0459,
+      "step": 816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9441084585734683e-05,
+      "loss": 1.6318,
+      "step": 817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9439333967885635e-05,
+      "loss": 2.1455,
+      "step": 818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943758069177526e-05,
+      "loss": 2.0781,
+      "step": 819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9435824757897302e-05,
+      "loss": 2.085,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943406616674626e-05,
+      "loss": 1.9863,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9432304918817385e-05,
+      "loss": 1.9692,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943054101460666e-05,
+      "loss": 2.2549,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9428774454610845e-05,
+      "loss": 2.0132,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9427005239327414e-05,
+      "loss": 1.9375,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9425233369254613e-05,
+      "loss": 2.0342,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942345884489143e-05,
+      "loss": 1.6084,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9421681666737592e-05,
+      "loss": 1.8535,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9419901835293585e-05,
+      "loss": 1.9326,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9418119351060633e-05,
+      "loss": 2.0537,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941633421454071e-05,
+      "loss": 1.9316,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941454642623654e-05,
+      "loss": 2.1768,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94127559866516e-05,
+      "loss": 2.0537,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9410962896290092e-05,
+      "loss": 1.6201,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940916715565698e-05,
+      "loss": 2.3633,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940736876525798e-05,
+      "loss": 2.1611,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9405567725599538e-05,
+      "loss": 2.0986,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940376403718886e-05,
+      "loss": 2.1133,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940195770053389e-05,
+      "loss": 1.8516,
+      "step": 839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9400148716143316e-05,
+      "loss": 1.9473,
+      "step": 840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9398337084526585e-05,
+      "loss": 2.2842,
+      "step": 841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9396522806193874e-05,
+      "loss": 2.1885,
+      "step": 842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939470588165611e-05,
+      "loss": 1.9766,
+      "step": 843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9392886311424975e-05,
+      "loss": 1.9824,
+      "step": 844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9391064096012878e-05,
+      "loss": 1.8223,
+      "step": 845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938923923593299e-05,
+      "loss": 2.0654,
+      "step": 846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9387411731699217e-05,
+      "loss": 2.207,
+      "step": 847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9385581583826215e-05,
+      "loss": 2.1279,
+      "step": 848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9383748792829374e-05,
+      "loss": 2.1113,
+      "step": 849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9381913359224844e-05,
+      "loss": 1.8418,
+      "step": 850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9380075283529506e-05,
+      "loss": 1.7627,
+      "step": 851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378234566260997e-05,
+      "loss": 1.9434,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9376391207937683e-05,
+      "loss": 2.2607,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9374545209078687e-05,
+      "loss": 2.1289,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937269657020387e-05,
+      "loss": 2.0381,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937084529183384e-05,
+      "loss": 1.8828,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9368991374489935e-05,
+      "loss": 2.085,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936713481869426e-05,
+      "loss": 1.9229,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936527562496964e-05,
+      "loss": 1.6848,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9363413793839658e-05,
+      "loss": 1.8535,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936154932582863e-05,
+      "loss": 1.8828,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935968222146162e-05,
+      "loss": 2.0381,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9357812481264435e-05,
+      "loss": 2.0957,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9355940105763622e-05,
+      "loss": 1.959,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935406509548647e-05,
+      "loss": 1.9619,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9352187450961014e-05,
+      "loss": 2.1895,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935030717271602e-05,
+      "loss": 2.2588,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9348424261281012e-05,
+      "loss": 2.0654,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.934653871718624e-05,
+      "loss": 2.2227,
+      "step": 869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9344650540962702e-05,
+      "loss": 1.9521,
+      "step": 870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9342759733142146e-05,
+      "loss": 1.9805,
+      "step": 871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9340866294257044e-05,
+      "loss": 1.9805,
+      "step": 872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9338970224840617e-05,
+      "loss": 2.4375,
+      "step": 873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933707152542683e-05,
+      "loss": 2.1377,
+      "step": 874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9335170196550393e-05,
+      "loss": 1.9785,
+      "step": 875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9333266238746736e-05,
+      "loss": 2.0898,
+      "step": 876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9331359652552052e-05,
+      "loss": 1.79,
+      "step": 877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932945043850326e-05,
+      "loss": 2.0732,
+      "step": 878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932753859713803e-05,
+      "loss": 2.0068,
+      "step": 879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932562412899476e-05,
+      "loss": 1.7852,
+      "step": 880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9323707034612594e-05,
+      "loss": 2.0029,
+      "step": 881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9321787314531417e-05,
+      "loss": 1.6528,
+      "step": 882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9319864969291853e-05,
+      "loss": 1.6885,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9317939999435262e-05,
+      "loss": 2.4424,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9316012405503748e-05,
+      "loss": 2.0771,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9314082188040145e-05,
+      "loss": 2.0205,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9312149347588035e-05,
+      "loss": 2.4531,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931021388469174e-05,
+      "loss": 2.2793,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.930827579989631e-05,
+      "loss": 2.2627,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.930633509374754e-05,
+      "loss": 2.042,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9304391766791964e-05,
+      "loss": 1.8452,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9302445819576855e-05,
+      "loss": 2.0234,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9300497252650217e-05,
+      "loss": 2.084,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9298546066560802e-05,
+      "loss": 2.3301,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929659226185809e-05,
+      "loss": 1.9307,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.92946358390923e-05,
+      "loss": 1.9746,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9292676798814397e-05,
+      "loss": 2.0586,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9290715141576068e-05,
+      "loss": 2.1436,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.928875086792976e-05,
+      "loss": 1.9229,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9286783978428625e-05,
+      "loss": 2.0742,
+      "step": 900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9284814473626584e-05,
+      "loss": 1.8896,
+      "step": 901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9282842354078274e-05,
+      "loss": 2.166,
+      "step": 902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9280867620339073e-05,
+      "loss": 1.9736,
+      "step": 903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9278890272965097e-05,
+      "loss": 2.2393,
+      "step": 904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9276910312513198e-05,
+      "loss": 1.8081,
+      "step": 905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927492773954096e-05,
+      "loss": 1.8955,
+      "step": 906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9272942554606712e-05,
+      "loss": 1.7676,
+      "step": 907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9270954758269512e-05,
+      "loss": 1.9512,
+      "step": 908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926896435108915e-05,
+      "loss": 1.958,
+      "step": 909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9266971333626153e-05,
+      "loss": 1.9785,
+      "step": 910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9264975706441793e-05,
+      "loss": 2.0068,
+      "step": 911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9262977470098067e-05,
+      "loss": 1.8057,
+      "step": 912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9260976625157703e-05,
+      "loss": 1.8506,
+      "step": 913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9258973172184176e-05,
+      "loss": 1.9023,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9256967111741686e-05,
+      "loss": 1.874,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9254958444395175e-05,
+      "loss": 1.9463,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9252947170710307e-05,
+      "loss": 2.0459,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9250933291253494e-05,
+      "loss": 1.8428,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924891680659187e-05,
+      "loss": 2.417,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9246897717293316e-05,
+      "loss": 2.1846,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924487602392643e-05,
+      "loss": 2.2354,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9242851727060557e-05,
+      "loss": 2.2344,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9240824827265766e-05,
+      "loss": 1.9053,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 2.0078,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9236763221173402e-05,
+      "loss": 2.0322,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923472851601963e-05,
+      "loss": 1.7002,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923269121022457e-05,
+      "loss": 2.1729,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923065130436195e-05,
+      "loss": 2.1279,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922860879900624e-05,
+      "loss": 2.0234,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9226563694732644e-05,
+      "loss": 2.0439,
+      "step": 930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.922451599211709e-05,
+      "loss": 2.0615,
+      "step": 931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9222465691736248e-05,
+      "loss": 1.9023,
+      "step": 932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9220412794167506e-05,
+      "loss": 1.9512,
+      "step": 933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9218357299988998e-05,
+      "loss": 2.0156,
+      "step": 934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9216299209779583e-05,
+      "loss": 1.999,
+      "step": 935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921423852411885e-05,
+      "loss": 1.8735,
+      "step": 936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9212175243587116e-05,
+      "loss": 1.8672,
+      "step": 937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9210109368765435e-05,
+      "loss": 1.9902,
+      "step": 938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920804090023559e-05,
+      "loss": 2.0576,
+      "step": 939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9205969838580098e-05,
+      "loss": 2.2695,
+      "step": 940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9203896184382186e-05,
+      "loss": 2.1035,
+      "step": 941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9201819938225844e-05,
+      "loss": 1.7969,
+      "step": 942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919974110069577e-05,
+      "loss": 1.9072,
+      "step": 943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9197659672377388e-05,
+      "loss": 2.0146,
+      "step": 944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919557565385687e-05,
+      "loss": 1.9844,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9193489045721105e-05,
+      "loss": 2.4336,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9191399848557713e-05,
+      "loss": 2.1377,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9189308062955044e-05,
+      "loss": 1.7832,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9187213689502177e-05,
+      "loss": 2.4131,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918511672878892e-05,
+      "loss": 2.0879,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918301718140581e-05,
+      "loss": 2.125,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918091504794411e-05,
+      "loss": 1.6138,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9178810328995814e-05,
+      "loss": 1.9336,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9176703025153643e-05,
+      "loss": 2.2158,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9174593137011047e-05,
+      "loss": 2.1338,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9172480665162202e-05,
+      "loss": 2.3486,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9170365610202016e-05,
+      "loss": 2.0752,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9168247972726114e-05,
+      "loss": 1.8933,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9166127753330856e-05,
+      "loss": 1.9473,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9164004952613337e-05,
+      "loss": 1.7515,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916187957117136e-05,
+      "loss": 1.9395,
+      "step": 961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915975160960347e-05,
+      "loss": 2.4028,
+      "step": 962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9157621068508935e-05,
+      "loss": 1.9199,
+      "step": 963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915548794848775e-05,
+      "loss": 1.9551,
+      "step": 964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9153352250140624e-05,
+      "loss": 2.3184,
+      "step": 965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915121397406901e-05,
+      "loss": 2.3008,
+      "step": 966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914907312087508e-05,
+      "loss": 1.9971,
+      "step": 967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9146929691161728e-05,
+      "loss": 2.0303,
+      "step": 968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914478368553258e-05,
+      "loss": 2.2422,
+      "step": 969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9142635104591982e-05,
+      "loss": 1.9404,
+      "step": 970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9140483948945005e-05,
+      "loss": 1.9653,
+      "step": 971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913833021919745e-05,
+      "loss": 1.9111,
+      "step": 972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9136173915955843e-05,
+      "loss": 2.2725,
+      "step": 973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9134015039827433e-05,
+      "loss": 1.9736,
+      "step": 974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9131853591420183e-05,
+      "loss": 2.0449,
+      "step": 975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.91296895713428e-05,
+      "loss": 1.8359,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.91275229802047e-05,
+      "loss": 2.042,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912535381861603e-05,
+      "loss": 2.0889,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9123182087187657e-05,
+      "loss": 1.6865,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9121007786531177e-05,
+      "loss": 1.8418,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9118830917258907e-05,
+      "loss": 2.2256,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9116651479983887e-05,
+      "loss": 2.1602,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9114469475319874e-05,
+      "loss": 1.9531,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911228490388136e-05,
+      "loss": 2.2842,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9110097766283552e-05,
+      "loss": 1.9507,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910790806314238e-05,
+      "loss": 1.9922,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9105715795074503e-05,
+      "loss": 2.0762,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9103520962697287e-05,
+      "loss": 1.8799,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9101323566628842e-05,
+      "loss": 2.1025,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9099123607487985e-05,
+      "loss": 2.2295,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9096921085894258e-05,
+      "loss": 2.0811,
+      "step": 991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9094716002467918e-05,
+      "loss": 1.9219,
+      "step": 992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9092508357829966e-05,
+      "loss": 2.1064,
+      "step": 993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.909029815260209e-05,
+      "loss": 2.2285,
+      "step": 994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908808538740673e-05,
+      "loss": 1.9717,
+      "step": 995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908587006286703e-05,
+      "loss": 2.0635,
+      "step": 996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9083652179606862e-05,
+      "loss": 1.8037,
+      "step": 997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9081431738250815e-05,
+      "loss": 2.1895,
+      "step": 998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9079208739424198e-05,
+      "loss": 1.834,
+      "step": 999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9076983183753045e-05,
+      "loss": 1.8506,
+      "step": 1000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.90747550718641e-05,
+      "loss": 2.1924,
+      "step": 1001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9072524404384842e-05,
+      "loss": 2.082,
+      "step": 1002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907029118194345e-05,
+      "loss": 2.2041,
+      "step": 1003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906805540516885e-05,
+      "loss": 2.3525,
+      "step": 1004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9065817074690657e-05,
+      "loss": 1.6846,
+      "step": 1005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9063576191139225e-05,
+      "loss": 2.1958,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9061332755145618e-05,
+      "loss": 1.7925,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9059086767341628e-05,
+      "loss": 2.1064,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905683822835975e-05,
+      "loss": 1.9453,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905458713883322e-05,
+      "loss": 2.457,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9052333499395966e-05,
+      "loss": 1.9961,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905007731068266e-05,
+      "loss": 1.8975,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9047818573328667e-05,
+      "loss": 2.3486,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904555728797009e-05,
+      "loss": 2.1875,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904329345524374e-05,
+      "loss": 1.7998,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9041027075787148e-05,
+      "loss": 1.9316,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903875815023856e-05,
+      "loss": 1.9297,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9036486679236938e-05,
+      "loss": 1.7188,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903421266342197e-05,
+      "loss": 2.3506,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9031936103434044e-05,
+      "loss": 2.0205,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9029656999914284e-05,
+      "loss": 2.0645,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9027375353504513e-05,
+      "loss": 2.3135,
+      "step": 1022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9025091164847286e-05,
+      "loss": 2.0469,
+      "step": 1023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9022804434585854e-05,
+      "loss": 2.1348,
+      "step": 1024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9020515163364205e-05,
+      "loss": 2.1904,
+      "step": 1025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9018223351827027e-05,
+      "loss": 1.8711,
+      "step": 1026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901592900061973e-05,
+      "loss": 2.1494,
+      "step": 1027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9013632110388445e-05,
+      "loss": 1.8115,
+      "step": 1028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9011332681780007e-05,
+      "loss": 1.8721,
+      "step": 1029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900903071544197e-05,
+      "loss": 2.0156,
+      "step": 1030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9006726212022603e-05,
+      "loss": 2.1865,
+      "step": 1031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9004419172170887e-05,
+      "loss": 1.7041,
+      "step": 1032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9002109596536523e-05,
+      "loss": 1.9854,
+      "step": 1033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8999797485769925e-05,
+      "loss": 1.9453,
+      "step": 1034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8997482840522218e-05,
+      "loss": 1.9531,
+      "step": 1035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8995165661445235e-05,
+      "loss": 1.9351,
+      "step": 1036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899284594919154e-05,
+      "loss": 2.0264,
+      "step": 1037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8990523704414388e-05,
+      "loss": 2.3193,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898819892776777e-05,
+      "loss": 2.0322,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898587161990637e-05,
+      "loss": 1.8262,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.89835417814856e-05,
+      "loss": 1.7988,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8981209413161572e-05,
+      "loss": 2.1484,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8978874515591124e-05,
+      "loss": 2.0898,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8976537089431793e-05,
+      "loss": 1.9131,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8974197135341835e-05,
+      "loss": 2.2656,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8971854653980218e-05,
+      "loss": 2.3057,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8969509646006625e-05,
+      "loss": 1.8896,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896716211208144e-05,
+      "loss": 1.9766,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8964812052865764e-05,
+      "loss": 2.1406,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8962459469021416e-05,
+      "loss": 2.1045,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8960104361210915e-05,
+      "loss": 1.8418,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8957746730097497e-05,
+      "loss": 1.8916,
+      "step": 1052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8955386576345103e-05,
+      "loss": 1.96,
+      "step": 1053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8953023900618395e-05,
+      "loss": 1.9521,
+      "step": 1054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8950658703582737e-05,
+      "loss": 2.1514,
+      "step": 1055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8948290985904204e-05,
+      "loss": 2.1719,
+      "step": 1056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8945920748249582e-05,
+      "loss": 1.8936,
+      "step": 1057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8943547991286362e-05,
+      "loss": 2.2334,
+      "step": 1058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8941172715682756e-05,
+      "loss": 1.9351,
+      "step": 1059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8938794922107676e-05,
+      "loss": 2.3887,
+      "step": 1060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8936414611230737e-05,
+      "loss": 2.0625,
+      "step": 1061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8934031783722283e-05,
+      "loss": 2.0684,
+      "step": 1062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.893164644025335e-05,
+      "loss": 1.958,
+      "step": 1063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8929258581495688e-05,
+      "loss": 1.8232,
+      "step": 1064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892686820812175e-05,
+      "loss": 2.1567,
+      "step": 1065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8924475320804706e-05,
+      "loss": 1.9888,
+      "step": 1066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8922079920218433e-05,
+      "loss": 2.2656,
+      "step": 1067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8919682007037504e-05,
+      "loss": 1.8799,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8917281581937216e-05,
+      "loss": 1.7744,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8914878645593565e-05,
+      "loss": 2.2705,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8912473198683248e-05,
+      "loss": 1.8711,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 1.9902,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890765477587298e-05,
+      "loss": 1.748,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8905241801329972e-05,
+      "loss": 2.0068,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8902826318934185e-05,
+      "loss": 2.0693,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8900408329365856e-05,
+      "loss": 2.1113,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889798783330593e-05,
+      "loss": 2.0444,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889556483143605e-05,
+      "loss": 2.1738,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889313932443858e-05,
+      "loss": 1.9326,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889071131299657e-05,
+      "loss": 2.0381,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8888280797793796e-05,
+      "loss": 2.1201,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8885847779514718e-05,
+      "loss": 2.0264,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.888341225884452e-05,
+      "loss": 2.293,
+      "step": 1083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.888097423646907e-05,
+      "loss": 1.9521,
+      "step": 1084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8878533713074967e-05,
+      "loss": 2.4375,
+      "step": 1085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8876090689349493e-05,
+      "loss": 1.8438,
+      "step": 1086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.887364516598064e-05,
+      "loss": 1.9004,
+      "step": 1087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8871197143657105e-05,
+      "loss": 2.1133,
+      "step": 1088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8868746623068292e-05,
+      "loss": 1.9521,
+      "step": 1089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8866293604904304e-05,
+      "loss": 2.0264,
+      "step": 1090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8863838089855943e-05,
+      "loss": 2.2061,
+      "step": 1091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8861380078614728e-05,
+      "loss": 1.938,
+      "step": 1092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8858919571872867e-05,
+      "loss": 2.0063,
+      "step": 1093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885645657032328e-05,
+      "loss": 1.9023,
+      "step": 1094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8853991074659583e-05,
+      "loss": 2.1016,
+      "step": 1095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8851523085576098e-05,
+      "loss": 1.8027,
+      "step": 1096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884905260376785e-05,
+      "loss": 2.1738,
+      "step": 1097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8846579629930565e-05,
+      "loss": 1.9424,
+      "step": 1098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884410416476067e-05,
+      "loss": 1.6426,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8841626208955293e-05,
+      "loss": 1.7959,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8839145763212265e-05,
+      "loss": 2.0264,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883666282823012e-05,
+      "loss": 2.0918,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8834177404708084e-05,
+      "loss": 2.0439,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8831689493346095e-05,
+      "loss": 1.9678,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8829199094844786e-05,
+      "loss": 1.9414,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8826706209905495e-05,
+      "loss": 2.1377,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882421083923025e-05,
+      "loss": 2.062,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8821712983521793e-05,
+      "loss": 2.3174,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 1.8657,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8816709819819663e-05,
+      "loss": 2.0488,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8814204513234964e-05,
+      "loss": 2.0615,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881169672443498e-05,
+      "loss": 2.1807,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8809186454125955e-05,
+      "loss": 1.8928,
+      "step": 1113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8806673703014805e-05,
+      "loss": 1.9746,
+      "step": 1114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8804158471809166e-05,
+      "loss": 2.0596,
+      "step": 1115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.880164076121737e-05,
+      "loss": 2.0244,
+      "step": 1116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8799120571948432e-05,
+      "loss": 1.7725,
+      "step": 1117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8796597904712087e-05,
+      "loss": 2.0264,
+      "step": 1118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8794072760218752e-05,
+      "loss": 1.8301,
+      "step": 1119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879154513917955e-05,
+      "loss": 1.9697,
+      "step": 1120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8789015042306287e-05,
+      "loss": 2.167,
+      "step": 1121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878648247031149e-05,
+      "loss": 2.1826,
+      "step": 1122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8783947423908365e-05,
+      "loss": 1.9258,
+      "step": 1123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8781409903810823e-05,
+      "loss": 1.7783,
+      "step": 1124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8778869910733463e-05,
+      "loss": 1.7944,
+      "step": 1125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8776327445391586e-05,
+      "loss": 2.1553,
+      "step": 1126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8773782508501194e-05,
+      "loss": 1.9043,
+      "step": 1127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877123510077898e-05,
+      "loss": 2.1494,
+      "step": 1128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876868522294233e-05,
+      "loss": 2.1221,
+      "step": 1129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876613287570933e-05,
+      "loss": 2.2158,
+      "step": 1130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876357805979876e-05,
+      "loss": 2.0586,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8761020775930095e-05,
+      "loss": 2.3096,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8758461024823503e-05,
+      "loss": 1.9976,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8755898807199856e-05,
+      "loss": 1.9824,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8753334123780714e-05,
+      "loss": 1.7744,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8750766975288322e-05,
+      "loss": 2.1016,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8748197362445635e-05,
+      "loss": 2.0117,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8745625285976294e-05,
+      "loss": 2.0205,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8743050746604635e-05,
+      "loss": 2.1221,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874047374505569e-05,
+      "loss": 2.3096,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8737894282055177e-05,
+      "loss": 2.1904,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873531235832952e-05,
+      "loss": 1.999,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8732727974605818e-05,
+      "loss": 2.2061,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8730141131611882e-05,
+      "loss": 2.0488,
+      "step": 1144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8727551830076204e-05,
+      "loss": 2.1172,
+      "step": 1145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8724960070727974e-05,
+      "loss": 2.166,
+      "step": 1146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8722365854297064e-05,
+      "loss": 2.3242,
+      "step": 1147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8719769181514053e-05,
+      "loss": 1.8145,
+      "step": 1148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8717170053110198e-05,
+      "loss": 1.6597,
+      "step": 1149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.871456846981745e-05,
+      "loss": 1.9736,
+      "step": 1150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8711964432368467e-05,
+      "loss": 1.8828,
+      "step": 1151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870935794149658e-05,
+      "loss": 1.917,
+      "step": 1152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8706748997935814e-05,
+      "loss": 1.9131,
+      "step": 1153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870413760242089e-05,
+      "loss": 2.126,
+      "step": 1154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8701523755687217e-05,
+      "loss": 1.8291,
+      "step": 1155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8698907458470895e-05,
+      "loss": 2.4893,
+      "step": 1156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8696288711508712e-05,
+      "loss": 1.915,
+      "step": 1157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8693667515538147e-05,
+      "loss": 1.6777,
+      "step": 1158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.869104387129737e-05,
+      "loss": 1.8604,
+      "step": 1159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868841777952524e-05,
+      "loss": 2.1582,
+      "step": 1160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8685789240961303e-05,
+      "loss": 1.7852,
+      "step": 1161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.86831582563458e-05,
+      "loss": 2.0312,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8680524826419652e-05,
+      "loss": 2.2285,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8677888951924473e-05,
+      "loss": 1.8008,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867525063360257e-05,
+      "loss": 1.876,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8672609872196933e-05,
+      "loss": 1.8662,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8669966668451235e-05,
+      "loss": 1.9678,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866732102310985e-05,
+      "loss": 1.9346,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8664672936917828e-05,
+      "loss": 1.5957,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8662022410620917e-05,
+      "loss": 2.1436,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8659369444965535e-05,
+      "loss": 1.8428,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865671404069881e-05,
+      "loss": 2.1191,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8654056198568536e-05,
+      "loss": 2.0361,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8651395919323203e-05,
+      "loss": 2.0322,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.864873320371199e-05,
+      "loss": 2.3369,
+      "step": 1175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8646068052484757e-05,
+      "loss": 1.9062,
+      "step": 1176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864340046639205e-05,
+      "loss": 2.0537,
+      "step": 1177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8640730446185106e-05,
+      "loss": 1.9199,
+      "step": 1178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863805799261584e-05,
+      "loss": 2.0352,
+      "step": 1179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8635383106436854e-05,
+      "loss": 2.1289,
+      "step": 1180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8632705788401444e-05,
+      "loss": 2.0332,
+      "step": 1181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863002603926358e-05,
+      "loss": 1.8926,
+      "step": 1182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862734385977792e-05,
+      "loss": 2.0781,
+      "step": 1183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8624659250699807e-05,
+      "loss": 1.7876,
+      "step": 1184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862197221278527e-05,
+      "loss": 1.6982,
+      "step": 1185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8619282746791017e-05,
+      "loss": 1.9814,
+      "step": 1186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8616590853474443e-05,
+      "loss": 1.9453,
+      "step": 1187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861389653359363e-05,
+      "loss": 1.8076,
+      "step": 1188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861119978790734e-05,
+      "loss": 2.1123,
+      "step": 1189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8608500617175012e-05,
+      "loss": 2.2344,
+      "step": 1190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8605799022156777e-05,
+      "loss": 2.1719,
+      "step": 1191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860309500361345e-05,
+      "loss": 2.1094,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860038856230652e-05,
+      "loss": 2.1016,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8597679698998164e-05,
+      "loss": 1.7959,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8594968414451235e-05,
+      "loss": 2.1348,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859225470942928e-05,
+      "loss": 2.0449,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8589538584696512e-05,
+      "loss": 2.1553,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.858682004101784e-05,
+      "loss": 2.0996,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8584099079158842e-05,
+      "loss": 2.3408,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8581375699885787e-05,
+      "loss": 1.9893,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857864990396562e-05,
+      "loss": 1.9507,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8575921692165963e-05,
+      "loss": 2.1729,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8573191065255126e-05,
+      "loss": 1.9062,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8570458024002094e-05,
+      "loss": 1.9668,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8567722569176534e-05,
+      "loss": 1.8672,
+      "step": 1205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8564984701548793e-05,
+      "loss": 2.0273,
+      "step": 1206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8562244421889896e-05,
+      "loss": 1.8018,
+      "step": 1207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8559501730971544e-05,
+      "loss": 2.0435,
+      "step": 1208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855675662956613e-05,
+      "loss": 1.9316,
+      "step": 1209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855400911844671e-05,
+      "loss": 1.9326,
+      "step": 1210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8551259198387023e-05,
+      "loss": 1.9297,
+      "step": 1211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854850687016149e-05,
+      "loss": 1.7402,
+      "step": 1212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8545752134545214e-05,
+      "loss": 1.7715,
+      "step": 1213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854299499231397e-05,
+      "loss": 1.8389,
+      "step": 1214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8540235444244205e-05,
+      "loss": 2.0518,
+      "step": 1215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8537473491113057e-05,
+      "loss": 1.8018,
+      "step": 1216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853470913369833e-05,
+      "loss": 2.1348,
+      "step": 1217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8531942372778512e-05,
+      "loss": 2.1299,
+      "step": 1218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852917320913276e-05,
+      "loss": 2.0889,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8526401643540924e-05,
+      "loss": 2.0879,
+      "step": 1220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852362767678351e-05,
+      "loss": 2.2451,
+      "step": 1221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8520851309641713e-05,
+      "loss": 1.9355,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.85180725428974e-05,
+      "loss": 1.8965,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8515291377333114e-05,
+      "loss": 1.9502,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8512507813732076e-05,
+      "loss": 2.418,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8509721852878177e-05,
+      "loss": 1.9697,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850693349555599e-05,
+      "loss": 2.0586,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8504142742550755e-05,
+      "loss": 2.0273,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8501349594648394e-05,
+      "loss": 1.8359,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.84985540526355e-05,
+      "loss": 2.0371,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8495756117299338e-05,
+      "loss": 1.8174,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.849295578942785e-05,
+      "loss": 1.9326,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8490153069809656e-05,
+      "loss": 2.0332,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8487347959234042e-05,
+      "loss": 2.0645,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.848454045849097e-05,
+      "loss": 1.957,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8481730568371073e-05,
+      "loss": 1.8877,
+      "step": 1236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847891828966566e-05,
+      "loss": 2.0205,
+      "step": 1237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8476103623166716e-05,
+      "loss": 1.9912,
+      "step": 1238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847328656966689e-05,
+      "loss": 1.7988,
+      "step": 1239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847046712995951e-05,
+      "loss": 1.8701,
+      "step": 1240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8467645304838575e-05,
+      "loss": 2.0322,
+      "step": 1241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8464821095098748e-05,
+      "loss": 2.0234,
+      "step": 1242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846199450153537e-05,
+      "loss": 1.9453,
+      "step": 1243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8459165524944463e-05,
+      "loss": 1.7695,
+      "step": 1244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84563341661227e-05,
+      "loss": 2.0674,
+      "step": 1245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8453500425867443e-05,
+      "loss": 1.9502,
+      "step": 1246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.845066430497671e-05,
+      "loss": 1.8525,
+      "step": 1247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8447825804249203e-05,
+      "loss": 1.9492,
+      "step": 1248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8444984924484278e-05,
+      "loss": 1.8882,
+      "step": 1249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8442141666481975e-05,
+      "loss": 1.9756,
+      "step": 1250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8439296031043006e-05,
+      "loss": 2.3438,
+      "step": 1251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8436448018968733e-05,
+      "loss": 2.0293,
+      "step": 1252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.843359763106121e-05,
+      "loss": 1.8301,
+      "step": 1253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8430744868123146e-05,
+      "loss": 1.8877,
+      "step": 1254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842788973095792e-05,
+      "loss": 2.2168,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842503222036959e-05,
+      "loss": 1.9004,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8422172337162865e-05,
+      "loss": 1.8062,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8419310082143143e-05,
+      "loss": 1.8525,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8416445456116473e-05,
+      "loss": 2.2021,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8413578459889572e-05,
+      "loss": 1.9346,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8410709094269842e-05,
+      "loss": 1.8843,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8407837360065332e-05,
+      "loss": 1.9844,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840496325808477e-05,
+      "loss": 1.8135,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8402086789137547e-05,
+      "loss": 1.8457,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8399207954033714e-05,
+      "loss": 1.8506,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.839632675358401e-05,
+      "loss": 1.8418,
+      "step": 1266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.839344318859981e-05,
+      "loss": 1.9893,
+      "step": 1267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8390557259893177e-05,
+      "loss": 2.1279,
+      "step": 1268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8387668968276836e-05,
+      "loss": 2.1396,
+      "step": 1269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8384778314564168e-05,
+      "loss": 1.8457,
+      "step": 1270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.838188529956923e-05,
+      "loss": 1.9219,
+      "step": 1271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837898992410674e-05,
+      "loss": 2.0791,
+      "step": 1272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837609218899207e-05,
+      "loss": 1.8535,
+      "step": 1273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8373192095041278e-05,
+      "loss": 1.999,
+      "step": 1274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8370289643071072e-05,
+      "loss": 1.9346,
+      "step": 1275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.836738483389883e-05,
+      "loss": 2.0342,
+      "step": 1276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8364477668342582e-05,
+      "loss": 1.7852,
+      "step": 1277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8361568147221038e-05,
+      "loss": 1.8945,
+      "step": 1278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8358656271353558e-05,
+      "loss": 1.8555,
+      "step": 1279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.835574204156018e-05,
+      "loss": 1.7539,
+      "step": 1280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8352825458661588e-05,
+      "loss": 1.9912,
+      "step": 1281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8349906523479136e-05,
+      "loss": 2.0381,
+      "step": 1282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8346985236834845e-05,
+      "loss": 1.9316,
+      "step": 1283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8344061599551397e-05,
+      "loss": 1.708,
+      "step": 1284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8341135612452125e-05,
+      "loss": 1.8789,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833820727636104e-05,
+      "loss": 2.0273,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8335276592102794e-05,
+      "loss": 2.0859,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833234356050273e-05,
+      "loss": 1.9854,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832940818238682e-05,
+      "loss": 1.7129,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832647045858172e-05,
+      "loss": 2.0254,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8323530389914735e-05,
+      "loss": 1.8118,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8320587977213834e-05,
+      "loss": 2.1445,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8317643221307642e-05,
+      "loss": 2.0884,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 2.0283,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8311746683197214e-05,
+      "loss": 2.1904,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8308794902653532e-05,
+      "loss": 2.0713,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8305840782225673e-05,
+      "loss": 2.1055,
+      "step": 1297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.830288432274556e-05,
+      "loss": 2.2627,
+      "step": 1298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8299925525045782e-05,
+      "loss": 2.3154,
+      "step": 1299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829696438995958e-05,
+      "loss": 1.6543,
+      "step": 1300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8294000918320854e-05,
+      "loss": 1.9238,
+      "step": 1301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829103511096417e-05,
+      "loss": 2.1846,
+      "step": 1302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.828806696872473e-05,
+      "loss": 2.1592,
+      "step": 1303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8285096492438424e-05,
+      "loss": 2.1387,
+      "step": 1304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8282123682941776e-05,
+      "loss": 2.1826,
+      "step": 1305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8279148541071972e-05,
+      "loss": 2.1279,
+      "step": 1306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8276171067666867e-05,
+      "loss": 1.4121,
+      "step": 1307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8273191263564956e-05,
+      "loss": 1.8916,
+      "step": 1308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8270209129605397e-05,
+      "loss": 1.9336,
+      "step": 1309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826722466662801e-05,
+      "loss": 1.6909,
+      "step": 1310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8264237875473263e-05,
+      "loss": 1.8389,
+      "step": 1311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826124875698228e-05,
+      "loss": 2.0615,
+      "step": 1312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8258257311996844e-05,
+      "loss": 1.8379,
+      "step": 1313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8255263541359397e-05,
+      "loss": 1.9492,
+      "step": 1314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8252267445913022e-05,
+      "loss": 1.9492,
+      "step": 1315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8249269026501472e-05,
+      "loss": 1.8481,
+      "step": 1316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8246268283969145e-05,
+      "loss": 1.7773,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8243265219161098e-05,
+      "loss": 1.9873,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8240259832923035e-05,
+      "loss": 1.8672,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8237252126101323e-05,
+      "loss": 2.1064,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8234242099542984e-05,
+      "loss": 1.8315,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8231229754095678e-05,
+      "loss": 2.166,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8228215090607728e-05,
+      "loss": 1.894,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8225198109928116e-05,
+      "loss": 1.8467,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8222178812906467e-05,
+      "loss": 2.0186,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8219157200393063e-05,
+      "loss": 1.7578,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8216133273238833e-05,
+      "loss": 1.8574,
+      "step": 1327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8213107032295367e-05,
+      "loss": 2.3467,
+      "step": 1328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8210078478414895e-05,
+      "loss": 2.1123,
+      "step": 1329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8207047612450302e-05,
+      "loss": 2.1357,
+      "step": 1330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8204014435255136e-05,
+      "loss": 1.916,
+      "step": 1331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8200978947683583e-05,
+      "loss": 2.1152,
+      "step": 1332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8197941150590477e-05,
+      "loss": 1.9717,
+      "step": 1333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8194901044831313e-05,
+      "loss": 1.915,
+      "step": 1334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819185863126223e-05,
+      "loss": 1.8438,
+      "step": 1335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818881391074002e-05,
+      "loss": 1.8457,
+      "step": 1336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8185766884122123e-05,
+      "loss": 1.8613,
+      "step": 1337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8182717552266625e-05,
+      "loss": 1.9434,
+      "step": 1338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.817966591603227e-05,
+      "loss": 2.127,
+      "step": 1339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8176611976278443e-05,
+      "loss": 1.957,
+      "step": 1340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8173555733865177e-05,
+      "loss": 1.8027,
+      "step": 1341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8170497189653164e-05,
+      "loss": 2.1631,
+      "step": 1342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816743634450373e-05,
+      "loss": 1.835,
+      "step": 1343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8164373199278858e-05,
+      "loss": 1.8032,
+      "step": 1344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816130775484118e-05,
+      "loss": 1.8389,
+      "step": 1345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8158240012053968e-05,
+      "loss": 1.854,
+      "step": 1346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815516997178115e-05,
+      "loss": 2.1846,
+      "step": 1347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815209763488729e-05,
+      "loss": 2.0195,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8149023002237612e-05,
+      "loss": 1.9707,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8145946074697977e-05,
+      "loss": 2.1797,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.814286685313489e-05,
+      "loss": 2.0947,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8139785338415515e-05,
+      "loss": 1.9805,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8136701531407652e-05,
+      "loss": 1.8281,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8133615432979742e-05,
+      "loss": 2.2134,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813052704400089e-05,
+      "loss": 1.9521,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.812743636534082e-05,
+      "loss": 1.9561,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8124343397869925e-05,
+      "loss": 1.4341,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8121248142459225e-05,
+      "loss": 1.8682,
+      "step": 1358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8118150599980398e-05,
+      "loss": 1.8789,
+      "step": 1359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8115050771305754e-05,
+      "loss": 2.2324,
+      "step": 1360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8111948657308253e-05,
+      "loss": 2.3955,
+      "step": 1361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81088442588615e-05,
+      "loss": 2.5449,
+      "step": 1362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8105737576839747e-05,
+      "loss": 2.0283,
+      "step": 1363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8102628612117868e-05,
+      "loss": 1.9609,
+      "step": 1364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8099517365571405e-05,
+      "loss": 2.084,
+      "step": 1365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8096403838076536e-05,
+      "loss": 1.8633,
+      "step": 1366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8093288030510075e-05,
+      "loss": 1.7959,
+      "step": 1367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.8848,
+      "step": 1368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8087049578672847e-05,
+      "loss": 2.1416,
+      "step": 1369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8083926936158925e-05,
+      "loss": 1.7041,
+      "step": 1370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8080802017087098e-05,
+      "loss": 2.0952,
+      "step": 1371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8077674822337394e-05,
+      "loss": 1.9678,
+      "step": 1372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807454535279047e-05,
+      "loss": 1.7617,
+      "step": 1373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8071413609327638e-05,
+      "loss": 1.9912,
+      "step": 1374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.806827959283085e-05,
+      "loss": 1.8721,
+      "step": 1375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8065143304182683e-05,
+      "loss": 2.1641,
+      "step": 1376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8062004744266372e-05,
+      "loss": 2.1055,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8058863913965782e-05,
+      "loss": 2.0869,
+      "step": 1378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8055720814165415e-05,
+      "loss": 1.9648,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805257544575042e-05,
+      "loss": 1.8633,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.804942780960658e-05,
+      "loss": 1.5781,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8046277906620317e-05,
+      "loss": 2.0508,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.804312573767869e-05,
+      "loss": 2.0957,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8039971303669407e-05,
+      "loss": 2.0547,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8036814605480796e-05,
+      "loss": 1.8623,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8033655644001837e-05,
+      "loss": 1.875,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8030494420122135e-05,
+      "loss": 1.7002,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8027330934731945e-05,
+      "loss": 1.8906,
+      "step": 1388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8024165188722153e-05,
+      "loss": 2.1387,
+      "step": 1389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8020997182984278e-05,
+      "loss": 2.2012,
+      "step": 1390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8017826918410482e-05,
+      "loss": 1.8896,
+      "step": 1391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8014654395893563e-05,
+      "loss": 1.998,
+      "step": 1392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8011479616326947e-05,
+      "loss": 2.3271,
+      "step": 1393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.80083025806047e-05,
+      "loss": 1.9658,
+      "step": 1394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8005123289621528e-05,
+      "loss": 1.916,
+      "step": 1395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8001941744272767e-05,
+      "loss": 1.8838,
+      "step": 1396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7998757945454388e-05,
+      "loss": 2.2256,
+      "step": 1397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7995571894063e-05,
+      "loss": 2.1104,
+      "step": 1398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.799238359099584e-05,
+      "loss": 2.3232,
+      "step": 1399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7989193037150785e-05,
+      "loss": 2.0469,
+      "step": 1400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7986000233426346e-05,
+      "loss": 1.8086,
+      "step": 1401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.798280518072166e-05,
+      "loss": 1.6567,
+      "step": 1402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797960787993651e-05,
+      "loss": 2.1299,
+      "step": 1403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79764083319713e-05,
+      "loss": 1.9111,
+      "step": 1404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797320653772707e-05,
+      "loss": 1.8447,
+      "step": 1405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79700024981055e-05,
+      "loss": 2.0391,
+      "step": 1406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79667962140089e-05,
+      "loss": 2.334,
+      "step": 1407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79635876863402e-05,
+      "loss": 2.123,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7960376916002974e-05,
+      "loss": 1.7046,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.795716390390142e-05,
+      "loss": 1.9365,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7953948650940377e-05,
+      "loss": 2.0195,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.795073115802531e-05,
+      "loss": 1.8682,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7947511426062313e-05,
+      "loss": 1.8574,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7944289455958114e-05,
+      "loss": 2.1953,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.794106524862006e-05,
+      "loss": 2.3584,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7937838804956152e-05,
+      "loss": 1.9453,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793461012587499e-05,
+      "loss": 1.8545,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793137921228583e-05,
+      "loss": 1.9429,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.792814606509855e-05,
+      "loss": 2.1816,
+      "step": 1419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7924910685223643e-05,
+      "loss": 2.5273,
+      "step": 1420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.792167307357225e-05,
+      "loss": 1.8418,
+      "step": 1421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7918433231056125e-05,
+      "loss": 1.9062,
+      "step": 1422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7915191158587668e-05,
+      "loss": 2.2158,
+      "step": 1423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7911946857079886e-05,
+      "loss": 1.8369,
+      "step": 1424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7908700327446427e-05,
+      "loss": 1.8457,
+      "step": 1425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790545157060157e-05,
+      "loss": 2.1143,
+      "step": 1426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7902200587460205e-05,
+      "loss": 2.0605,
+      "step": 1427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7898947378937862e-05,
+      "loss": 1.9111,
+      "step": 1428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7895691945950696e-05,
+      "loss": 2.1621,
+      "step": 1429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789243428941549e-05,
+      "loss": 2.167,
+      "step": 1430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7889174410249644e-05,
+      "loss": 2.2393,
+      "step": 1431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.788591230937119e-05,
+      "loss": 1.6855,
+      "step": 1432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7882647987698796e-05,
+      "loss": 2.1406,
+      "step": 1433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787938144615173e-05,
+      "loss": 1.9229,
+      "step": 1434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7876112685649914e-05,
+      "loss": 2.0293,
+      "step": 1435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787284170711387e-05,
+      "loss": 2.165,
+      "step": 1436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.786956851146476e-05,
+      "loss": 2.1689,
+      "step": 1437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.786629309962437e-05,
+      "loss": 1.8955,
+      "step": 1438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.78630154725151e-05,
+      "loss": 2.0791,
+      "step": 1439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7859735631059983e-05,
+      "loss": 1.8232,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785645357618268e-05,
+      "loss": 2.0205,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785316930880745e-05,
+      "loss": 1.916,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.784988282985921e-05,
+      "loss": 2.1455,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7846594140263475e-05,
+      "loss": 1.8428,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7843303240946393e-05,
+      "loss": 1.9023,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.784001013283473e-05,
+      "loss": 2.0664,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7836714816855877e-05,
+      "loss": 1.8906,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.783341729393785e-05,
+      "loss": 1.6553,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.783011756500927e-05,
+      "loss": 1.6001,
+      "step": 1449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7826815630999398e-05,
+      "loss": 1.9102,
+      "step": 1450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7823511492838117e-05,
+      "loss": 1.6982,
+      "step": 1451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7820205151455914e-05,
+      "loss": 1.9541,
+      "step": 1452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7816896607783905e-05,
+      "loss": 1.9365,
+      "step": 1453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7813585862753832e-05,
+      "loss": 1.7344,
+      "step": 1454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7810272917298045e-05,
+      "loss": 1.7773,
+      "step": 1455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7806957772349527e-05,
+      "loss": 2.0625,
+      "step": 1456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.780364042884187e-05,
+      "loss": 1.9736,
+      "step": 1457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.780032088770929e-05,
+      "loss": 2.1016,
+      "step": 1458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779699914988662e-05,
+      "loss": 2.1172,
+      "step": 1459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779367521630931e-05,
+      "loss": 2.1074,
+      "step": 1460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779034908791344e-05,
+      "loss": 1.8789,
+      "step": 1461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778702076563569e-05,
+      "loss": 2.1191,
+      "step": 1462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7783690250413364e-05,
+      "loss": 1.9224,
+      "step": 1463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7780357543184396e-05,
+      "loss": 2.1611,
+      "step": 1464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777702264488732e-05,
+      "loss": 1.8447,
+      "step": 1465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.77736855564613e-05,
+      "loss": 1.9678,
+      "step": 1466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7770346278846107e-05,
+      "loss": 2.1094,
+      "step": 1467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7767004812982133e-05,
+      "loss": 1.9844,
+      "step": 1468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.776366115981039e-05,
+      "loss": 2.1816,
+      "step": 1469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.77603153202725e-05,
+      "loss": 1.8613,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7756967295310698e-05,
+      "loss": 2.0586,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775361708586785e-05,
+      "loss": 1.835,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7750264692887415e-05,
+      "loss": 1.958,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7746910117313482e-05,
+      "loss": 2.0596,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7743553360090754e-05,
+      "loss": 1.9639,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.774019442216454e-05,
+      "loss": 1.9863,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.773683330448078e-05,
+      "loss": 1.9268,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7733470007986003e-05,
+      "loss": 1.9238,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.773010453362737e-05,
+      "loss": 1.9707,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7726736882352652e-05,
+      "loss": 1.7158,
+      "step": 1480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.772336705511023e-05,
+      "loss": 1.8955,
+      "step": 1481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7719995052849095e-05,
+      "loss": 1.9697,
+      "step": 1482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7716620876518863e-05,
+      "loss": 2.1514,
+      "step": 1483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.771324452706975e-05,
+      "loss": 1.7051,
+      "step": 1484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7709866005452583e-05,
+      "loss": 2.1533,
+      "step": 1485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7706485312618813e-05,
+      "loss": 1.8164,
+      "step": 1486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.770310244952049e-05,
+      "loss": 1.8994,
+      "step": 1487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769971741711028e-05,
+      "loss": 1.7495,
+      "step": 1488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7696330216341465e-05,
+      "loss": 1.9785,
+      "step": 1489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7692940848167928e-05,
+      "loss": 1.8779,
+      "step": 1490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.768954931354417e-05,
+      "loss": 2.168,
+      "step": 1491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7686155613425296e-05,
+      "loss": 2.0137,
+      "step": 1492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7682759748767026e-05,
+      "loss": 1.7939,
+      "step": 1493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767936172052569e-05,
+      "loss": 2.1914,
+      "step": 1494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767596152965822e-05,
+      "loss": 1.5957,
+      "step": 1495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7672559177122167e-05,
+      "loss": 1.9785,
+      "step": 1496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7669154663875684e-05,
+      "loss": 2.0332,
+      "step": 1497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766574799087753e-05,
+      "loss": 1.8496,
+      "step": 1498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7662339159087077e-05,
+      "loss": 2.0537,
+      "step": 1499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.765892816946431e-05,
+      "loss": 1.9297,
+      "step": 1500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7655515022969817e-05,
+      "loss": 2.0371,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7652099720564783e-05,
+      "loss": 2.0576,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.764868226321102e-05,
+      "loss": 1.7021,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7645262651870926e-05,
+      "loss": 1.8652,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7641840887507523e-05,
+      "loss": 1.8972,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7638416971084433e-05,
+      "loss": 1.7832,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763499090356588e-05,
+      "loss": 1.6025,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7631562685916705e-05,
+      "loss": 1.9766,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.762813231910233e-05,
+      "loss": 1.8662,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7624699804088815e-05,
+      "loss": 2.1641,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7621265141842804e-05,
+      "loss": 2.0303,
+      "step": 1511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7617828333331547e-05,
+      "loss": 2.0586,
+      "step": 1512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7614389379522906e-05,
+      "loss": 1.7119,
+      "step": 1513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.761094828138534e-05,
+      "loss": 1.8926,
+      "step": 1514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.760750503988792e-05,
+      "loss": 1.9404,
+      "step": 1515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7604059656000313e-05,
+      "loss": 2.1416,
+      "step": 1516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7600612130692788e-05,
+      "loss": 1.6987,
+      "step": 1517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7597162464936226e-05,
+      "loss": 2.0,
+      "step": 1518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7593710659702105e-05,
+      "loss": 1.7114,
+      "step": 1519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7590256715962506e-05,
+      "loss": 1.9912,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7586800634690114e-05,
+      "loss": 1.7502,
+      "step": 1521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758334241685821e-05,
+      "loss": 1.877,
+      "step": 1522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757988206344069e-05,
+      "loss": 2.2178,
+      "step": 1523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7576419575412028e-05,
+      "loss": 2.083,
+      "step": 1524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757295495374733e-05,
+      "loss": 2.0439,
+      "step": 1525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7569488199422275e-05,
+      "loss": 2.0283,
+      "step": 1526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.756601931341316e-05,
+      "loss": 1.8613,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7562548296696876e-05,
+      "loss": 2.1895,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7559075150250913e-05,
+      "loss": 2.1592,
+      "step": 1529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7555599875053362e-05,
+      "loss": 2.0518,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7552122472082915e-05,
+      "loss": 1.9824,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754864294231886e-05,
+      "loss": 1.916,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754516128674109e-05,
+      "loss": 2.3936,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754167750633009e-05,
+      "loss": 1.9717,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7538191602066946e-05,
+      "loss": 2.2461,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.753470357493335e-05,
+      "loss": 1.6499,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7531213425911566e-05,
+      "loss": 1.7783,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752772115598449e-05,
+      "loss": 1.9795,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7524226766135587e-05,
+      "loss": 1.9189,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7520730257348947e-05,
+      "loss": 1.8994,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7517231630609223e-05,
+      "loss": 2.0439,
+      "step": 1541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.751373088690169e-05,
+      "loss": 1.877,
+      "step": 1542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.751022802721221e-05,
+      "loss": 1.7217,
+      "step": 1543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7506723052527243e-05,
+      "loss": 1.7432,
+      "step": 1544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7503215963833847e-05,
+      "loss": 1.9023,
+      "step": 1545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7499706762119664e-05,
+      "loss": 2.3486,
+      "step": 1546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7496195448372947e-05,
+      "loss": 1.8613,
+      "step": 1547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749268202358253e-05,
+      "loss": 2.1426,
+      "step": 1548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7489166488737847e-05,
+      "loss": 2.0723,
+      "step": 1549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7485648844828932e-05,
+      "loss": 1.8711,
+      "step": 1550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7482129092846404e-05,
+      "loss": 2.0518,
+      "step": 1551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7478607233781478e-05,
+      "loss": 1.874,
+      "step": 1552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747508326862597e-05,
+      "loss": 1.7822,
+      "step": 1553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7471557198372277e-05,
+      "loss": 2.2217,
+      "step": 1554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.746802902401339e-05,
+      "loss": 1.7559,
+      "step": 1555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.746449874654291e-05,
+      "loss": 2.0312,
+      "step": 1556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7460966366955003e-05,
+      "loss": 1.8564,
+      "step": 1557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7457431886244452e-05,
+      "loss": 1.8057,
+      "step": 1558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7453895305406615e-05,
+      "loss": 2.542,
+      "step": 1559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7450356625437454e-05,
+      "loss": 2.0938,
+      "step": 1560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7446815847333506e-05,
+      "loss": 1.8516,
+      "step": 1561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7443272972091917e-05,
+      "loss": 1.8887,
+      "step": 1562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743972800071041e-05,
+      "loss": 1.8799,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7436180934187307e-05,
+      "loss": 2.2227,
+      "step": 1564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7432631773521517e-05,
+      "loss": 2.0303,
+      "step": 1565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7429080519712532e-05,
+      "loss": 1.4761,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7425527173760442e-05,
+      "loss": 1.7432,
+      "step": 1567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7421971736665927e-05,
+      "loss": 1.8926,
+      "step": 1568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.741841420943025e-05,
+      "loss": 1.6128,
+      "step": 1569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7414854593055267e-05,
+      "loss": 1.8613,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7411292888543415e-05,
+      "loss": 1.8242,
+      "step": 1571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7407729096897735e-05,
+      "loss": 2.209,
+      "step": 1572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7404163219121837e-05,
+      "loss": 1.8975,
+      "step": 1573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.740059525621993e-05,
+      "loss": 1.9424,
+      "step": 1574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7397025209196807e-05,
+      "loss": 1.9009,
+      "step": 1575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7393453079057847e-05,
+      "loss": 1.9277,
+      "step": 1576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7389878866809016e-05,
+      "loss": 2.0498,
+      "step": 1577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.738630257345687e-05,
+      "loss": 1.999,
+      "step": 1578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7382724200008546e-05,
+      "loss": 2.1797,
+      "step": 1579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7379143747471768e-05,
+      "loss": 2.0117,
+      "step": 1580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737556121685485e-05,
+      "loss": 1.9688,
+      "step": 1581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737197660916669e-05,
+      "loss": 2.041,
+      "step": 1582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7368389925416757e-05,
+      "loss": 2.3389,
+      "step": 1583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7364801166615124e-05,
+      "loss": 1.8496,
+      "step": 1584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736121033377244e-05,
+      "loss": 2.0762,
+      "step": 1585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7357617427899938e-05,
+      "loss": 1.9248,
+      "step": 1586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.735402245000944e-05,
+      "loss": 1.957,
+      "step": 1587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7350425401113334e-05,
+      "loss": 2.2939,
+      "step": 1588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734682628222462e-05,
+      "loss": 1.8516,
+      "step": 1589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7343225094356857e-05,
+      "loss": 2.0732,
+      "step": 1590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7339621838524196e-05,
+      "loss": 1.8398,
+      "step": 1591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733601651574137e-05,
+      "loss": 1.8701,
+      "step": 1592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7332409127023688e-05,
+      "loss": 2.0811,
+      "step": 1593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7328799673387053e-05,
+      "loss": 2.1367,
+      "step": 1594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.732518815584794e-05,
+      "loss": 2.0908,
+      "step": 1595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7321574575423406e-05,
+      "loss": 2.043,
+      "step": 1596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731795893313109e-05,
+      "loss": 1.9346,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731434122998922e-05,
+      "loss": 1.5601,
+      "step": 1598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7310721467016587e-05,
+      "loss": 1.7651,
+      "step": 1599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730709964523258e-05,
+      "loss": 2.1533,
+      "step": 1600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7303475765657153e-05,
+      "loss": 1.917,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7299849829310847e-05,
+      "loss": 2.0713,
+      "step": 1602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7296221837214785e-05,
+      "loss": 2.1924,
+      "step": 1603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7292591790390668e-05,
+      "loss": 1.7725,
+      "step": 1604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7288959689860764e-05,
+      "loss": 1.9619,
+      "step": 1605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7285325536647933e-05,
+      "loss": 1.8398,
+      "step": 1606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7281689331775606e-05,
+      "loss": 1.8516,
+      "step": 1607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7278051076267798e-05,
+      "loss": 2.4717,
+      "step": 1608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7274410771149094e-05,
+      "loss": 2.0537,
+      "step": 1609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7270768417444667e-05,
+      "loss": 2.2646,
+      "step": 1610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.726712401618025e-05,
+      "loss": 2.0498,
+      "step": 1611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7263477568382166e-05,
+      "loss": 2.1387,
+      "step": 1612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7259829075077314e-05,
+      "loss": 2.1455,
+      "step": 1613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.725617853729316e-05,
+      "loss": 2.3076,
+      "step": 1614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7252525956057756e-05,
+      "loss": 2.209,
+      "step": 1615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.724887133239972e-05,
+      "loss": 2.0117,
+      "step": 1616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7245214667348257e-05,
+      "loss": 1.8643,
+      "step": 1617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7241555961933137e-05,
+      "loss": 1.8818,
+      "step": 1618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7237895217184702e-05,
+      "loss": 2.1133,
+      "step": 1619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7234232434133885e-05,
+      "loss": 2.0527,
+      "step": 1620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.723056761381217e-05,
+      "loss": 1.8799,
+      "step": 1621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7226900757251635e-05,
+      "loss": 2.1572,
+      "step": 1622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7223231865484923e-05,
+      "loss": 1.8896,
+      "step": 1623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7219560939545246e-05,
+      "loss": 2.5576,
+      "step": 1624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7215887980466393e-05,
+      "loss": 1.8955,
+      "step": 1625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7212212989282733e-05,
+      "loss": 1.7148,
+      "step": 1626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720853596702919e-05,
+      "loss": 1.8779,
+      "step": 1627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7204856914741276e-05,
+      "loss": 1.9961,
+      "step": 1628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7201175833455066e-05,
+      "loss": 1.9072,
+      "step": 1629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719749272420721e-05,
+      "loss": 2.3145,
+      "step": 1630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7193807588034928e-05,
+      "loss": 2.0962,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719012042597601e-05,
+      "loss": 1.8809,
+      "step": 1632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7186431239068822e-05,
+      "loss": 1.8564,
+      "step": 1633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.718274002835229e-05,
+      "loss": 1.8928,
+      "step": 1634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7179046794865918e-05,
+      "loss": 1.8965,
+      "step": 1635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7175351539649777e-05,
+      "loss": 1.9971,
+      "step": 1636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7171654263744506e-05,
+      "loss": 1.8867,
+      "step": 1637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7167954968191317e-05,
+      "loss": 1.627,
+      "step": 1638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7164253654031986e-05,
+      "loss": 2.2305,
+      "step": 1639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7160550322308863e-05,
+      "loss": 1.8135,
+      "step": 1640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7156844974064862e-05,
+      "loss": 1.9453,
+      "step": 1641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7153137610343466e-05,
+      "loss": 2.1816,
+      "step": 1642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7149428232188724e-05,
+      "loss": 2.1484,
+      "step": 1643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7145716840645253e-05,
+      "loss": 2.0879,
+      "step": 1644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7142003436758242e-05,
+      "loss": 2.0078,
+      "step": 1645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.713828802157344e-05,
+      "loss": 1.7793,
+      "step": 1646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7134570596137165e-05,
+      "loss": 2.0059,
+      "step": 1647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71308511614963e-05,
+      "loss": 2.002,
+      "step": 1648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7127129718698298e-05,
+      "loss": 2.0332,
+      "step": 1649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712340626879117e-05,
+      "loss": 1.9697,
+      "step": 1650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71196808128235e-05,
+      "loss": 2.1689,
+      "step": 1651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7115953351844433e-05,
+      "loss": 2.3828,
+      "step": 1652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7112223886903677e-05,
+      "loss": 1.7197,
+      "step": 1653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.710849241905151e-05,
+      "loss": 2.126,
+      "step": 1654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.710475894933877e-05,
+      "loss": 1.8496,
+      "step": 1655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7101023478816857e-05,
+      "loss": 1.9834,
+      "step": 1656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.709728600853774e-05,
+      "loss": 1.6772,
+      "step": 1657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7093546539553947e-05,
+      "loss": 1.7891,
+      "step": 1658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7089805072918567e-05,
+      "loss": 1.7559,
+      "step": 1659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7086061609685258e-05,
+      "loss": 2.0498,
+      "step": 1660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7082316150908235e-05,
+      "loss": 2.2119,
+      "step": 1661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7078568697642273e-05,
+      "loss": 2.1074,
+      "step": 1662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7074819250942726e-05,
+      "loss": 2.0215,
+      "step": 1663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 1.7773,
+      "step": 1664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7067314381467e-05,
+      "loss": 1.8223,
+      "step": 1665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7063558960804318e-05,
+      "loss": 1.8813,
+      "step": 1666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705980155093501e-05,
+      "loss": 1.7495,
+      "step": 1667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705604215291723e-05,
+      "loss": 2.1924,
+      "step": 1668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7052280767809672e-05,
+      "loss": 1.8379,
+      "step": 1669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7048517396671605e-05,
+      "loss": 2.0254,
+      "step": 1670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.704475204056285e-05,
+      "loss": 2.0117,
+      "step": 1671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.704098470054379e-05,
+      "loss": 2.0801,
+      "step": 1672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7037215377675367e-05,
+      "loss": 2.0098,
+      "step": 1673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7033444073019077e-05,
+      "loss": 1.8799,
+      "step": 1674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7029670787636983e-05,
+      "loss": 1.8574,
+      "step": 1675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7025895522591694e-05,
+      "loss": 1.8965,
+      "step": 1676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7022118278946384e-05,
+      "loss": 2.0186,
+      "step": 1677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7018339057764787e-05,
+      "loss": 2.0283,
+      "step": 1678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7014557860111184e-05,
+      "loss": 1.9893,
+      "step": 1679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701077468705042e-05,
+      "loss": 1.875,
+      "step": 1680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7006989539647894e-05,
+      "loss": 1.8682,
+      "step": 1681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7003202418969566e-05,
+      "loss": 2.2588,
+      "step": 1682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6999413326081942e-05,
+      "loss": 2.0688,
+      "step": 1683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6995622262052093e-05,
+      "loss": 2.0078,
+      "step": 1684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6991829227947637e-05,
+      "loss": 1.5962,
+      "step": 1685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6988034224836754e-05,
+      "loss": 1.9268,
+      "step": 1686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698423725378817e-05,
+      "loss": 1.8018,
+      "step": 1687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6980438315871178e-05,
+      "loss": 1.7686,
+      "step": 1688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.697663741215561e-05,
+      "loss": 1.9512,
+      "step": 1689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6972834543711866e-05,
+      "loss": 1.9292,
+      "step": 1690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6969029711610884e-05,
+      "loss": 1.999,
+      "step": 1691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696522291692417e-05,
+      "loss": 1.9727,
+      "step": 1692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696141416072377e-05,
+      "loss": 1.9326,
+      "step": 1693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6957603444082295e-05,
+      "loss": 2.1074,
+      "step": 1694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6953790768072895e-05,
+      "loss": 2.0811,
+      "step": 1695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.694997613376928e-05,
+      "loss": 1.645,
+      "step": 1696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.694615954224571e-05,
+      "loss": 2.0244,
+      "step": 1697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6942340994576998e-05,
+      "loss": 1.9492,
+      "step": 1698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6938520491838502e-05,
+      "loss": 1.5913,
+      "step": 1699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6934698035106135e-05,
+      "loss": 1.915,
+      "step": 1700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6930873625456362e-05,
+      "loss": 1.9648,
+      "step": 1701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.692704726396619e-05,
+      "loss": 2.084,
+      "step": 1702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.692321895171319e-05,
+      "loss": 2.1553,
+      "step": 1703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6919388689775463e-05,
+      "loss": 2.0547,
+      "step": 1704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691555647923168e-05,
+      "loss": 1.9307,
+      "step": 1705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6911722321161042e-05,
+      "loss": 1.8818,
+      "step": 1706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.690788621664331e-05,
+      "loss": 1.791,
+      "step": 1707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6904048166758792e-05,
+      "loss": 1.7695,
+      "step": 1708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6900208172588333e-05,
+      "loss": 1.9814,
+      "step": 1709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6896366235213348e-05,
+      "loss": 1.7598,
+      "step": 1710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6892522355715774e-05,
+      "loss": 2.3877,
+      "step": 1711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6888676535178114e-05,
+      "loss": 2.1133,
+      "step": 1712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6884828774683402e-05,
+      "loss": 1.9082,
+      "step": 1713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6880979075315238e-05,
+      "loss": 1.9785,
+      "step": 1714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687712743815775e-05,
+      "loss": 1.75,
+      "step": 1715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687327386429561e-05,
+      "loss": 2.1562,
+      "step": 1716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686941835481406e-05,
+      "loss": 2.0752,
+      "step": 1717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686556091079886e-05,
+      "loss": 2.209,
+      "step": 1718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6861701533336322e-05,
+      "loss": 2.0654,
+      "step": 1719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6857840223513315e-05,
+      "loss": 1.8584,
+      "step": 1720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685397698241724e-05,
+      "loss": 1.9492,
+      "step": 1721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685011181113605e-05,
+      "loss": 1.8438,
+      "step": 1722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6846244710758223e-05,
+      "loss": 2.1367,
+      "step": 1723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6842375682372803e-05,
+      "loss": 2.0684,
+      "step": 1724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.683850472706937e-05,
+      "loss": 1.8774,
+      "step": 1725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.683463184593804e-05,
+      "loss": 1.9736,
+      "step": 1726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6830757040069477e-05,
+      "loss": 2.0713,
+      "step": 1727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6826880310554888e-05,
+      "loss": 2.085,
+      "step": 1728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6823001658486013e-05,
+      "loss": 1.8604,
+      "step": 1729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6819121084955145e-05,
+      "loss": 2.166,
+      "step": 1730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6815238591055117e-05,
+      "loss": 2.2197,
+      "step": 1731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6811354177879288e-05,
+      "loss": 1.7197,
+      "step": 1732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.680746784652158e-05,
+      "loss": 2.2705,
+      "step": 1733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6803579598076434e-05,
+      "loss": 1.8232,
+      "step": 1734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6799689433638847e-05,
+      "loss": 1.876,
+      "step": 1735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6795797354304345e-05,
+      "loss": 1.9111,
+      "step": 1736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6791903361169e-05,
+      "loss": 1.5737,
+      "step": 1737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678800745532942e-05,
+      "loss": 2.2598,
+      "step": 1738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678410963788275e-05,
+      "loss": 1.9355,
+      "step": 1739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6780209909926675e-05,
+      "loss": 2.1309,
+      "step": 1740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.677630827255942e-05,
+      "loss": 1.7808,
+      "step": 1741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6772404726879745e-05,
+      "loss": 2.7109,
+      "step": 1742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.676849927398695e-05,
+      "loss": 1.7983,
+      "step": 1743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.676459191498087e-05,
+      "loss": 2.1953,
+      "step": 1744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6760682650961877e-05,
+      "loss": 1.9355,
+      "step": 1745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6756771483030878e-05,
+      "loss": 2.1689,
+      "step": 1746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6752858412289318e-05,
+      "loss": 2.0859,
+      "step": 1747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674894343983918e-05,
+      "loss": 1.751,
+      "step": 1748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674502656678298e-05,
+      "loss": 1.9229,
+      "step": 1749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674110779422377e-05,
+      "loss": 2.084,
+      "step": 1750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6737187123265134e-05,
+      "loss": 2.1729,
+      "step": 1751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6733264555011196e-05,
+      "loss": 2.0225,
+      "step": 1752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6729340090566604e-05,
+      "loss": 1.8813,
+      "step": 1753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6725413731036562e-05,
+      "loss": 1.7695,
+      "step": 1754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.672148547752678e-05,
+      "loss": 1.8921,
+      "step": 1755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.671755533114352e-05,
+      "loss": 1.9541,
+      "step": 1756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.671362329299357e-05,
+      "loss": 2.2578,
+      "step": 1757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6709689364184255e-05,
+      "loss": 1.917,
+      "step": 1758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6705753545823423e-05,
+      "loss": 2.2998,
+      "step": 1759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6701815839019468e-05,
+      "loss": 1.8096,
+      "step": 1760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6697876244881307e-05,
+      "loss": 2.291,
+      "step": 1761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6693934764518388e-05,
+      "loss": 1.9482,
+      "step": 1762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6689991399040694e-05,
+      "loss": 2.0547,
+      "step": 1763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6686046149558736e-05,
+      "loss": 1.8564,
+      "step": 1764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6682099017183562e-05,
+      "loss": 1.9199,
+      "step": 1765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6678150003026736e-05,
+      "loss": 1.9453,
+      "step": 1766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6674199108200366e-05,
+      "loss": 1.8438,
+      "step": 1767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6670246333817088e-05,
+      "loss": 2.1426,
+      "step": 1768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6666291680990056e-05,
+      "loss": 1.8818,
+      "step": 1769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.666233515083297e-05,
+      "loss": 2.2129,
+      "step": 1770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6658376744460043e-05,
+      "loss": 2.0645,
+      "step": 1771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665441646298602e-05,
+      "loss": 2.0742,
+      "step": 1772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6650454307526187e-05,
+      "loss": 2.0576,
+      "step": 1773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6646490279196344e-05,
+      "loss": 1.7998,
+      "step": 1774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664252437911282e-05,
+      "loss": 1.9814,
+      "step": 1775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6638556608392473e-05,
+      "loss": 2.0225,
+      "step": 1776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663458696815269e-05,
+      "loss": 1.8936,
+      "step": 1777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663061545951138e-05,
+      "loss": 1.7617,
+      "step": 1778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6626642083586986e-05,
+      "loss": 2.001,
+      "step": 1779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6622666841498464e-05,
+      "loss": 1.7812,
+      "step": 1780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661868973436531e-05,
+      "loss": 1.8594,
+      "step": 1781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661471076330754e-05,
+      "loss": 1.7646,
+      "step": 1782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6610729929445688e-05,
+      "loss": 2.0312,
+      "step": 1783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6606747233900816e-05,
+      "loss": 1.8008,
+      "step": 1784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6602762677794517e-05,
+      "loss": 1.9434,
+      "step": 1785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.65987762622489e-05,
+      "loss": 1.8408,
+      "step": 1786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659478798838661e-05,
+      "loss": 1.8047,
+      "step": 1787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659079785733079e-05,
+      "loss": 1.9951,
+      "step": 1788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6586805870205135e-05,
+      "loss": 1.5767,
+      "step": 1789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6582812028133843e-05,
+      "loss": 1.9521,
+      "step": 1790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6578816332241645e-05,
+      "loss": 1.605,
+      "step": 1791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657481878365379e-05,
+      "loss": 1.9033,
+      "step": 1792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657081938349605e-05,
+      "loss": 1.7646,
+      "step": 1793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656681813289471e-05,
+      "loss": 2.0537,
+      "step": 1794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6562815032976593e-05,
+      "loss": 1.9238,
+      "step": 1795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.655881008486903e-05,
+      "loss": 1.7397,
+      "step": 1796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6554803289699875e-05,
+      "loss": 1.8857,
+      "step": 1797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.65507946485975e-05,
+      "loss": 1.874,
+      "step": 1798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654678416269081e-05,
+      "loss": 1.8027,
+      "step": 1799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654277183310921e-05,
+      "loss": 1.8779,
+      "step": 1800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6538757660982635e-05,
+      "loss": 1.9785,
+      "step": 1801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.653474164744154e-05,
+      "loss": 1.7007,
+      "step": 1802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6530723793616895e-05,
+      "loss": 1.9805,
+      "step": 1803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.652670410064019e-05,
+      "loss": 1.665,
+      "step": 1804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6522682569643427e-05,
+      "loss": 1.6909,
+      "step": 1805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.651865920175914e-05,
+      "loss": 2.5488,
+      "step": 1806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6514633998120365e-05,
+      "loss": 1.8164,
+      "step": 1807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6510606959860664e-05,
+      "loss": 2.0391,
+      "step": 1808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6506578088114105e-05,
+      "loss": 2.3184,
+      "step": 1809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6502547384015294e-05,
+      "loss": 2.0342,
+      "step": 1810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6498514848699332e-05,
+      "loss": 1.9287,
+      "step": 1811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6494480483301836e-05,
+      "loss": 1.9795,
+      "step": 1812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649044428895896e-05,
+      "loss": 1.9746,
+      "step": 1813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6486406266807343e-05,
+      "loss": 2.0498,
+      "step": 1814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6482366417984163e-05,
+      "loss": 1.9551,
+      "step": 1815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6478324743627104e-05,
+      "loss": 2.04,
+      "step": 1816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6474281244874356e-05,
+      "loss": 1.7246,
+      "step": 1817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647023592286464e-05,
+      "loss": 1.915,
+      "step": 1818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646618877873717e-05,
+      "loss": 1.8223,
+      "step": 1819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6462139813631693e-05,
+      "loss": 1.7798,
+      "step": 1820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.645808902868846e-05,
+      "loss": 1.7725,
+      "step": 1821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.645403642504823e-05,
+      "loss": 1.9502,
+      "step": 1822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644998200385228e-05,
+      "loss": 2.1562,
+      "step": 1823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6445925766242392e-05,
+      "loss": 2.2041,
+      "step": 1824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644186771336087e-05,
+      "loss": 1.8589,
+      "step": 1825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6437807846350526e-05,
+      "loss": 2.0449,
+      "step": 1826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.643374616635468e-05,
+      "loss": 1.9531,
+      "step": 1827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6429682674517158e-05,
+      "loss": 1.9805,
+      "step": 1828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6425617371982302e-05,
+      "loss": 2.0981,
+      "step": 1829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6421550259894972e-05,
+      "loss": 1.916,
+      "step": 1830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6417481339400523e-05,
+      "loss": 1.7578,
+      "step": 1831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6413410611644827e-05,
+      "loss": 2.2402,
+      "step": 1832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.640933807777426e-05,
+      "loss": 2.1406,
+      "step": 1833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6405263738935716e-05,
+      "loss": 1.8701,
+      "step": 1834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6401187596276588e-05,
+      "loss": 1.8037,
+      "step": 1835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.639710965094478e-05,
+      "loss": 1.7388,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6393029904088703e-05,
+      "loss": 1.8149,
+      "step": 1837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.638894835685728e-05,
+      "loss": 2.0996,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6384865010399935e-05,
+      "loss": 1.7432,
+      "step": 1839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6380779865866603e-05,
+      "loss": 2.0225,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.637669292440772e-05,
+      "loss": 1.7422,
+      "step": 1841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.637260418717423e-05,
+      "loss": 2.0771,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6368513655317593e-05,
+      "loss": 2.4746,
+      "step": 1843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6364421329989758e-05,
+      "loss": 1.668,
+      "step": 1844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6360327212343188e-05,
+      "loss": 1.9121,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.635623130353085e-05,
+      "loss": 2.1348,
+      "step": 1846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.635213360470622e-05,
+      "loss": 1.9902,
+      "step": 1847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634803411702326e-05,
+      "loss": 2.0283,
+      "step": 1848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6343932841636455e-05,
+      "loss": 1.874,
+      "step": 1849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6339829779700792e-05,
+      "loss": 1.7012,
+      "step": 1850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6335724932371753e-05,
+      "loss": 1.5703,
+      "step": 1851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6331618300805327e-05,
+      "loss": 2.0635,
+      "step": 1852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6327509886158003e-05,
+      "loss": 1.7715,
+      "step": 1853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.632339968958677e-05,
+      "loss": 1.7686,
+      "step": 1854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6319287712249127e-05,
+      "loss": 1.833,
+      "step": 1855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.631517395530307e-05,
+      "loss": 1.6982,
+      "step": 1856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6311058419907096e-05,
+      "loss": 2.0439,
+      "step": 1857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6306941107220202e-05,
+      "loss": 1.7905,
+      "step": 1858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6302822018401885e-05,
+      "loss": 1.7373,
+      "step": 1859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629870115461215e-05,
+      "loss": 2.0244,
+      "step": 1860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6294578517011484e-05,
+      "loss": 1.7573,
+      "step": 1861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6290454106760893e-05,
+      "loss": 1.9014,
+      "step": 1862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6286327925021877e-05,
+      "loss": 2.0156,
+      "step": 1863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6282199972956425e-05,
+      "loss": 2.6406,
+      "step": 1864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6278070251727035e-05,
+      "loss": 1.7725,
+      "step": 1865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62739387624967e-05,
+      "loss": 1.8535,
+      "step": 1866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6269805506428912e-05,
+      "loss": 1.8135,
+      "step": 1867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6265670484687655e-05,
+      "loss": 2.0576,
+      "step": 1868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6261533698437416e-05,
+      "loss": 1.8691,
+      "step": 1869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625739514884318e-05,
+      "loss": 2.1045,
+      "step": 1870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625325483707043e-05,
+      "loss": 1.7627,
+      "step": 1871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6249112764285133e-05,
+      "loss": 2.1182,
+      "step": 1872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624496893165376e-05,
+      "loss": 1.8994,
+      "step": 1873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6240823340343285e-05,
+      "loss": 1.916,
+      "step": 1874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6236675991521167e-05,
+      "loss": 1.75,
+      "step": 1875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6232526886355363e-05,
+      "loss": 1.8433,
+      "step": 1876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.622837602601432e-05,
+      "loss": 1.8066,
+      "step": 1877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6224223411666993e-05,
+      "loss": 2.043,
+      "step": 1878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6220069044482815e-05,
+      "loss": 1.9414,
+      "step": 1879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6215912925631724e-05,
+      "loss": 1.8701,
+      "step": 1880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6211755056284143e-05,
+      "loss": 1.7529,
+      "step": 1881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6207595437610994e-05,
+      "loss": 2.1777,
+      "step": 1882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6203434070783685e-05,
+      "loss": 2.1514,
+      "step": 1883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6199270956974128e-05,
+      "loss": 2.1592,
+      "step": 1884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.619510609735472e-05,
+      "loss": 1.749,
+      "step": 1885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6190939493098344e-05,
+      "loss": 2.1777,
+      "step": 1886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618677114537838e-05,
+      "loss": 2.0322,
+      "step": 1887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61826010553687e-05,
+      "loss": 1.7363,
+      "step": 1888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6178429224243665e-05,
+      "loss": 1.6504,
+      "step": 1889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6174255653178126e-05,
+      "loss": 2.0547,
+      "step": 1890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.617008034334743e-05,
+      "loss": 1.9707,
+      "step": 1891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6165903295927405e-05,
+      "loss": 2.2305,
+      "step": 1892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.616172451209437e-05,
+      "loss": 1.8184,
+      "step": 1893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6157543993025134e-05,
+      "loss": 1.8252,
+      "step": 1894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6153361739896995e-05,
+      "loss": 1.9453,
+      "step": 1895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6149177753887744e-05,
+      "loss": 2.1162,
+      "step": 1896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6144992036175656e-05,
+      "loss": 2.0703,
+      "step": 1897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6140804587939488e-05,
+      "loss": 1.9854,
+      "step": 1898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6136615410358493e-05,
+      "loss": 1.7822,
+      "step": 1899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6132424504612408e-05,
+      "loss": 1.7656,
+      "step": 1900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6128231871881452e-05,
+      "loss": 1.7637,
+      "step": 1901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6124037513346342e-05,
+      "loss": 2.1729,
+      "step": 1902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6119841430188267e-05,
+      "loss": 1.9668,
+      "step": 1903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6115643623588915e-05,
+      "loss": 1.7227,
+      "step": 1904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6111444094730447e-05,
+      "loss": 1.9922,
+      "step": 1905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.610724284479552e-05,
+      "loss": 2.04,
+      "step": 1906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6103039874967262e-05,
+      "loss": 1.4253,
+      "step": 1907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6098835186429304e-05,
+      "loss": 1.8652,
+      "step": 1908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6094628780365745e-05,
+      "loss": 1.8496,
+      "step": 1909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6090420657961172e-05,
+      "loss": 1.832,
+      "step": 1910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6086210820400662e-05,
+      "loss": 1.9482,
+      "step": 1911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6081999268869765e-05,
+      "loss": 1.8281,
+      "step": 1912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6077786004554522e-05,
+      "loss": 1.8525,
+      "step": 1913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6073571028641452e-05,
+      "loss": 2.0312,
+      "step": 1914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6069354342317557e-05,
+      "loss": 1.876,
+      "step": 1915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6065135946770323e-05,
+      "loss": 1.9238,
+      "step": 1916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6060915843187707e-05,
+      "loss": 1.8706,
+      "step": 1917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6056694032758164e-05,
+      "loss": 1.9033,
+      "step": 1918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6052470516670613e-05,
+      "loss": 1.8643,
+      "step": 1919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6048245296114468e-05,
+      "loss": 1.748,
+      "step": 1920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6044018372279613e-05,
+      "loss": 1.7441,
+      "step": 1921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6039789746356412e-05,
+      "loss": 1.957,
+      "step": 1922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6035559419535714e-05,
+      "loss": 1.8906,
+      "step": 1923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6031327393008848e-05,
+      "loss": 2.0088,
+      "step": 1924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6027093667967606e-05,
+      "loss": 2.0488,
+      "step": 1925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6022858245604278e-05,
+      "loss": 1.8936,
+      "step": 1926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6018621127111626e-05,
+      "loss": 1.9805,
+      "step": 1927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.601438231368288e-05,
+      "loss": 1.875,
+      "step": 1928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6010141806511765e-05,
+      "loss": 2.2041,
+      "step": 1929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6005899606792465e-05,
+      "loss": 1.96,
+      "step": 1930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6001655715719652e-05,
+      "loss": 2.1572,
+      "step": 1931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5997410134488464e-05,
+      "loss": 1.856,
+      "step": 1932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.599316286429453e-05,
+      "loss": 2.0718,
+      "step": 1933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.598891390633395e-05,
+      "loss": 1.6333,
+      "step": 1934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5984663261803283e-05,
+      "loss": 1.8496,
+      "step": 1935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5980410931899585e-05,
+      "loss": 2.0498,
+      "step": 1936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5976156917820373e-05,
+      "loss": 1.7402,
+      "step": 1937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5971901220763645e-05,
+      "loss": 1.7666,
+      "step": 1938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.596764384192787e-05,
+      "loss": 1.7905,
+      "step": 1939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.596338478251199e-05,
+      "loss": 1.7412,
+      "step": 1940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5959124043715428e-05,
+      "loss": 1.7407,
+      "step": 1941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5954861626738067e-05,
+      "loss": 1.75,
+      "step": 1942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.595059753278027e-05,
+      "loss": 1.8965,
+      "step": 1943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.594633176304287e-05,
+      "loss": 1.7861,
+      "step": 1944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5942064318727174e-05,
+      "loss": 1.8984,
+      "step": 1945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.593779520103496e-05,
+      "loss": 1.7444,
+      "step": 1946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.593352441116848e-05,
+      "loss": 1.5898,
+      "step": 1947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.592925195033045e-05,
+      "loss": 1.7051,
+      "step": 1948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5924977819724068e-05,
+      "loss": 2.2617,
+      "step": 1949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5920702020552984e-05,
+      "loss": 2.2061,
+      "step": 1950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591642455402133e-05,
+      "loss": 1.7959,
+      "step": 1951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591214542133372e-05,
+      "loss": 2.29,
+      "step": 1952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5907864623695206e-05,
+      "loss": 2.1367,
+      "step": 1953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.590358216231134e-05,
+      "loss": 1.7939,
+      "step": 1954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5899298038388116e-05,
+      "loss": 2.3154,
+      "step": 1955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5895012253132017e-05,
+      "loss": 1.9453,
+      "step": 1956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5890724807749988e-05,
+      "loss": 1.915,
+      "step": 1957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.588643570344943e-05,
+      "loss": 1.9131,
+      "step": 1958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5882144941438234e-05,
+      "loss": 2.0566,
+      "step": 1959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 1.8325,
+      "step": 1960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5873558449117745e-05,
+      "loss": 2.0459,
+      "step": 1961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.586926272122654e-05,
+      "loss": 2.1436,
+      "step": 1962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5864965340460864e-05,
+      "loss": 1.8438,
+      "step": 1963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5860666308030933e-05,
+      "loss": 1.79,
+      "step": 1964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585636562514741e-05,
+      "loss": 2.1748,
+      "step": 1965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5852063293021435e-05,
+      "loss": 2.0469,
+      "step": 1966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5847759312864615e-05,
+      "loss": 2.0923,
+      "step": 1967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5843453685889015e-05,
+      "loss": 1.9785,
+      "step": 1968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5839146413307167e-05,
+      "loss": 2.0137,
+      "step": 1969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.583483749633206e-05,
+      "loss": 1.8545,
+      "step": 1970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5830526936177156e-05,
+      "loss": 2.0737,
+      "step": 1971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5826214734056376e-05,
+      "loss": 1.9443,
+      "step": 1972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5821900891184098e-05,
+      "loss": 2.0684,
+      "step": 1973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5817585408775168e-05,
+      "loss": 1.8447,
+      "step": 1974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5813268288044896e-05,
+      "loss": 1.9268,
+      "step": 1975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580894953020904e-05,
+      "loss": 1.9238,
+      "step": 1976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580462913648384e-05,
+      "loss": 1.9893,
+      "step": 1977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5800307108085976e-05,
+      "loss": 2.1221,
+      "step": 1978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5795983446232602e-05,
+      "loss": 1.7793,
+      "step": 1979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5791658152141326e-05,
+      "loss": 2.1338,
+      "step": 1980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.578733122703022e-05,
+      "loss": 1.8311,
+      "step": 1981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5783002672117808e-05,
+      "loss": 1.7881,
+      "step": 1982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577867248862308e-05,
+      "loss": 1.9014,
+      "step": 1983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5774340677765483e-05,
+      "loss": 2.0957,
+      "step": 1984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5770007240764923e-05,
+      "loss": 2.0615,
+      "step": 1985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5765672178841754e-05,
+      "loss": 1.8145,
+      "step": 1986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5761335493216806e-05,
+      "loss": 1.8965,
+      "step": 1987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.575699718511135e-05,
+      "loss": 2.0283,
+      "step": 1988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5752657255747122e-05,
+      "loss": 1.8242,
+      "step": 1989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.574831570634631e-05,
+      "loss": 1.7842,
+      "step": 1990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5743972538131572e-05,
+      "loss": 1.7383,
+      "step": 1991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5739627752325995e-05,
+      "loss": 1.9326,
+      "step": 1992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.573528135015315e-05,
+      "loss": 2.127,
+      "step": 1993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5730933332837045e-05,
+      "loss": 2.124,
+      "step": 1994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5726583701602147e-05,
+      "loss": 2.0381,
+      "step": 1995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.572223245767338e-05,
+      "loss": 1.4873,
+      "step": 1996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5717879602276123e-05,
+      "loss": 1.9429,
+      "step": 1997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5713525136636206e-05,
+      "loss": 1.9111,
+      "step": 1998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5709169061979915e-05,
+      "loss": 1.7812,
+      "step": 1999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5704811379533986e-05,
+      "loss": 1.917,
+      "step": 2000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5700452090525608e-05,
+      "loss": 2.0195,
+      "step": 2001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5696091196182425e-05,
+      "loss": 2.0068,
+      "step": 2002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.569172869773253e-05,
+      "loss": 1.9668,
+      "step": 2003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.568736459640447e-05,
+      "loss": 1.8838,
+      "step": 2004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5682998893427246e-05,
+      "loss": 2.0864,
+      "step": 2005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5678631590030304e-05,
+      "loss": 1.9766,
+      "step": 2006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5674262687443547e-05,
+      "loss": 1.6602,
+      "step": 2007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5669892186897317e-05,
+      "loss": 1.9375,
+      "step": 2008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5665520089622424e-05,
+      "loss": 1.8896,
+      "step": 2009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5661146396850115e-05,
+      "loss": 1.9902,
+      "step": 2010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5656771109812082e-05,
+      "loss": 2.2666,
+      "step": 2011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5652394229740485e-05,
+      "loss": 1.9487,
+      "step": 2012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.564801575786791e-05,
+      "loss": 1.7627,
+      "step": 2013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5643635695427405e-05,
+      "loss": 1.9561,
+      "step": 2014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563925404365246e-05,
+      "loss": 2.3447,
+      "step": 2015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5634870803777027e-05,
+      "loss": 1.543,
+      "step": 2016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5630485977035484e-05,
+      "loss": 1.7925,
+      "step": 2017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5626099564662666e-05,
+      "loss": 1.9736,
+      "step": 2018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5621711567893853e-05,
+      "loss": 2.1553,
+      "step": 2019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5617321987964777e-05,
+      "loss": 1.7949,
+      "step": 2020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.561293082611161e-05,
+      "loss": 2.0879,
+      "step": 2021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5608538083570963e-05,
+      "loss": 2.0771,
+      "step": 2022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.560414376157991e-05,
+      "loss": 2.1875,
+      "step": 2023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5599747861375957e-05,
+      "loss": 1.7891,
+      "step": 2024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559535038419705e-05,
+      "loss": 2.4443,
+      "step": 2025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5590951331281595e-05,
+      "loss": 1.8906,
+      "step": 2026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5586550703868432e-05,
+      "loss": 1.7061,
+      "step": 2027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.558214850319684e-05,
+      "loss": 1.75,
+      "step": 2028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5577744730506545e-05,
+      "loss": 1.8003,
+      "step": 2029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5573339387037725e-05,
+      "loss": 1.8711,
+      "step": 2030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5568932474030988e-05,
+      "loss": 1.8262,
+      "step": 2031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5564523992727387e-05,
+      "loss": 1.9268,
+      "step": 2032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.556011394436842e-05,
+      "loss": 1.8301,
+      "step": 2033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 2.1982,
+      "step": 2034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.555128915145258e-05,
+      "loss": 2.0571,
+      "step": 2035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5546874409380897e-05,
+      "loss": 1.7646,
+      "step": 2036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5542458105224244e-05,
+      "loss": 2.0107,
+      "step": 2037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.553804024022632e-05,
+      "loss": 2.2295,
+      "step": 2038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5533620815631255e-05,
+      "loss": 1.8999,
+      "step": 2039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5529199832683634e-05,
+      "loss": 1.8633,
+      "step": 2040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552477729262848e-05,
+      "loss": 1.6011,
+      "step": 2041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552035319671123e-05,
+      "loss": 1.9043,
+      "step": 2042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.551592754617779e-05,
+      "loss": 2.041,
+      "step": 2043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.551150034227449e-05,
+      "loss": 1.9482,
+      "step": 2044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5507071586248098e-05,
+      "loss": 1.731,
+      "step": 2045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5502641279345816e-05,
+      "loss": 1.8965,
+      "step": 2046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5498209422815294e-05,
+      "loss": 1.8765,
+      "step": 2047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.54937760179046e-05,
+      "loss": 2.1738,
+      "step": 2048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5489341065862263e-05,
+      "loss": 2.0713,
+      "step": 2049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5484904567937223e-05,
+      "loss": 1.5721,
+      "step": 2050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5480466525378862e-05,
+      "loss": 1.8408,
+      "step": 2051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5476026939437015e-05,
+      "loss": 1.9473,
+      "step": 2052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5471585811361925e-05,
+      "loss": 1.8643,
+      "step": 2053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546714314240429e-05,
+      "loss": 1.9277,
+      "step": 2054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546269893381523e-05,
+      "loss": 2.0264,
+      "step": 2055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5458253186846302e-05,
+      "loss": 2.0645,
+      "step": 2056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.54538059027495e-05,
+      "loss": 1.8169,
+      "step": 2057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.544935708277724e-05,
+      "loss": 2.0693,
+      "step": 2058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5444906728182388e-05,
+      "loss": 1.8271,
+      "step": 2059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5440454840218225e-05,
+      "loss": 2.0312,
+      "step": 2060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.543600142013847e-05,
+      "loss": 1.9883,
+      "step": 2061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.543154646919728e-05,
+      "loss": 1.8809,
+      "step": 2062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.542708998864924e-05,
+      "loss": 2.0518,
+      "step": 2063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5422631979749354e-05,
+      "loss": 1.8398,
+      "step": 2064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5418172443753067e-05,
+      "loss": 1.9341,
+      "step": 2065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.541371138191626e-05,
+      "loss": 1.8057,
+      "step": 2066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5409248795495234e-05,
+      "loss": 1.6865,
+      "step": 2067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5404784685746715e-05,
+      "loss": 1.9307,
+      "step": 2068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5400319053927875e-05,
+      "loss": 2.1045,
+      "step": 2069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5395851901296296e-05,
+      "loss": 1.9268,
+      "step": 2070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5391383229110005e-05,
+      "loss": 1.7222,
+      "step": 2071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.538691303862744e-05,
+      "loss": 2.0391,
+      "step": 2072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5382441331107482e-05,
+      "loss": 2.2686,
+      "step": 2073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5377968107809425e-05,
+      "loss": 1.9463,
+      "step": 2074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5373493369993006e-05,
+      "loss": 1.8262,
+      "step": 2075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5369017118918375e-05,
+      "loss": 1.916,
+      "step": 2076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5364539355846115e-05,
+      "loss": 1.8555,
+      "step": 2077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.536006008203723e-05,
+      "loss": 1.9619,
+      "step": 2078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5355579298753154e-05,
+      "loss": 2.0283,
+      "step": 2079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5351097007255743e-05,
+      "loss": 2.1719,
+      "step": 2080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5346613208807284e-05,
+      "loss": 2.208,
+      "step": 2081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534212790467048e-05,
+      "loss": 1.9512,
+      "step": 2082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5337641096108457e-05,
+      "loss": 2.375,
+      "step": 2083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5333152784384777e-05,
+      "loss": 1.8462,
+      "step": 2084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5328662970763414e-05,
+      "loss": 1.8749,
+      "step": 2085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5324171656508768e-05,
+      "loss": 1.7886,
+      "step": 2086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5319678842885662e-05,
+      "loss": 1.6777,
+      "step": 2087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.531518453115934e-05,
+      "loss": 1.8408,
+      "step": 2088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5310688722595472e-05,
+      "loss": 1.7134,
+      "step": 2089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.530619141846015e-05,
+      "loss": 2.002,
+      "step": 2090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5301692620019875e-05,
+      "loss": 1.7529,
+      "step": 2091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5297192328541584e-05,
+      "loss": 2.1123,
+      "step": 2092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5292690545292625e-05,
+      "loss": 1.9194,
+      "step": 2093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528818727154077e-05,
+      "loss": 1.793,
+      "step": 2094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528368250855421e-05,
+      "loss": 2.2471,
+      "step": 2095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527917625760156e-05,
+      "loss": 2.0479,
+      "step": 2096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527466851995184e-05,
+      "loss": 2.0225,
+      "step": 2097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5270159296874507e-05,
+      "loss": 2.1162,
+      "step": 2098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5265648589639424e-05,
+      "loss": 1.7959,
+      "step": 2099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5261136399516873e-05,
+      "loss": 1.8438,
+      "step": 2100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5256622727777562e-05,
+      "loss": 1.7778,
+      "step": 2101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5252107575692605e-05,
+      "loss": 2.0332,
+      "step": 2102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5247590944533542e-05,
+      "loss": 1.8843,
+      "step": 2103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5243072835572319e-05,
+      "loss": 2.0889,
+      "step": 2104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5238553250081309e-05,
+      "loss": 2.1323,
+      "step": 2105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5234032189333301e-05,
+      "loss": 1.7871,
+      "step": 2106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5229509654601488e-05,
+      "loss": 1.8521,
+      "step": 2107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5224985647159489e-05,
+      "loss": 2.0195,
+      "step": 2108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5220460168281335e-05,
+      "loss": 1.5,
+      "step": 2109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5215933219241468e-05,
+      "loss": 2.0518,
+      "step": 2110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5211404801314743e-05,
+      "loss": 2.2256,
+      "step": 2111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5206874915776441e-05,
+      "loss": 1.9453,
+      "step": 2112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.520234356390224e-05,
+      "loss": 1.8311,
+      "step": 2113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.519781074696824e-05,
+      "loss": 1.752,
+      "step": 2114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.519327646625095e-05,
+      "loss": 1.4844,
+      "step": 2115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5188740723027295e-05,
+      "loss": 1.707,
+      "step": 2116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5184203518574613e-05,
+      "loss": 1.8535,
+      "step": 2117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5179664854170648e-05,
+      "loss": 1.749,
+      "step": 2118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5175124731093553e-05,
+      "loss": 1.8408,
+      "step": 2119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5170583150621904e-05,
+      "loss": 2.3301,
+      "step": 2120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5166040114034676e-05,
+      "loss": 2.1396,
+      "step": 2121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5161495622611255e-05,
+      "loss": 1.8516,
+      "step": 2122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5156949677631448e-05,
+      "loss": 1.9844,
+      "step": 2123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5152402280375454e-05,
+      "loss": 1.8086,
+      "step": 2124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5147853432123891e-05,
+      "loss": 2.0088,
+      "step": 2125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.514330313415779e-05,
+      "loss": 2.2471,
+      "step": 2126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.513875138775858e-05,
+      "loss": 2.2871,
+      "step": 2127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5134198194208107e-05,
+      "loss": 1.8257,
+      "step": 2128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5129643554788614e-05,
+      "loss": 1.8735,
+      "step": 2129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5125087470782762e-05,
+      "loss": 2.124,
+      "step": 2130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5120529943473614e-05,
+      "loss": 1.9668,
+      "step": 2131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5115970974144636e-05,
+      "loss": 2.2988,
+      "step": 2132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5111410564079707e-05,
+      "loss": 2.2041,
+      "step": 2133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5106848714563112e-05,
+      "loss": 1.9014,
+      "step": 2134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.510228542687953e-05,
+      "loss": 1.9404,
+      "step": 2135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5097720702314055e-05,
+      "loss": 2.0,
+      "step": 2136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5093154542152187e-05,
+      "loss": 1.9141,
+      "step": 2137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5088586947679824e-05,
+      "loss": 2.1553,
+      "step": 2138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5084017920183271e-05,
+      "loss": 2.2266,
+      "step": 2139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.507944746094924e-05,
+      "loss": 2.1035,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5074875571264836e-05,
+      "loss": 1.7598,
+      "step": 2141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5070302252417578e-05,
+      "loss": 1.6343,
+      "step": 2142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.506572750569538e-05,
+      "loss": 2.0732,
+      "step": 2143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5061151332386565e-05,
+      "loss": 1.999,
+      "step": 2144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5056573733779848e-05,
+      "loss": 1.708,
+      "step": 2145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5051994711164356e-05,
+      "loss": 1.7764,
+      "step": 2146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5047414265829607e-05,
+      "loss": 1.8662,
+      "step": 2147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.504283239906553e-05,
+      "loss": 1.998,
+      "step": 2148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5038249112162446e-05,
+      "loss": 1.8252,
+      "step": 2149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.503366440641108e-05,
+      "loss": 1.8438,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5029078283102554e-05,
+      "loss": 1.875,
+      "step": 2151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5024490743528393e-05,
+      "loss": 1.8691,
+      "step": 2152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5019901788980517e-05,
+      "loss": 1.9307,
+      "step": 2153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5015311420751243e-05,
+      "loss": 1.8184,
+      "step": 2154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.50107196401333e-05,
+      "loss": 2.0791,
+      "step": 2155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5006126448419792e-05,
+      "loss": 2.0576,
+      "step": 2156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5001531846904236e-05,
+      "loss": 1.959,
+      "step": 2157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4996935836880544e-05,
+      "loss": 1.748,
+      "step": 2158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4992338419643022e-05,
+      "loss": 1.7061,
+      "step": 2159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4987739596486375e-05,
+      "loss": 2.165,
+      "step": 2160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4983139368705699e-05,
+      "loss": 2.0029,
+      "step": 2161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.497853773759649e-05,
+      "loss": 1.8647,
+      "step": 2162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4973934704454638e-05,
+      "loss": 2.1143,
+      "step": 2163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4969330270576428e-05,
+      "loss": 1.9648,
+      "step": 2164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.496472443725854e-05,
+      "loss": 2.0879,
+      "step": 2165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4960117205798048e-05,
+      "loss": 2.0273,
+      "step": 2166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4955508577492416e-05,
+      "loss": 1.6577,
+      "step": 2167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4950898553639505e-05,
+      "loss": 1.8076,
+      "step": 2168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4946287135537571e-05,
+      "loss": 1.9854,
+      "step": 2169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4941674324485258e-05,
+      "loss": 2.0479,
+      "step": 2170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4937060121781607e-05,
+      "loss": 1.8354,
+      "step": 2171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4932444528726045e-05,
+      "loss": 1.7729,
+      "step": 2172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4927827546618394e-05,
+      "loss": 2.6543,
+      "step": 2173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4923209176758872e-05,
+      "loss": 2.2266,
+      "step": 2174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4918589420448074e-05,
+      "loss": 1.668,
+      "step": 2175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4913968278987004e-05,
+      "loss": 1.8643,
+      "step": 2176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4909345753677041e-05,
+      "loss": 1.6904,
+      "step": 2177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4904721845819961e-05,
+      "loss": 2.3223,
+      "step": 2178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4900096556717923e-05,
+      "loss": 1.9512,
+      "step": 2179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4895469887673483e-05,
+      "loss": 1.9385,
+      "step": 2180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4890841839989582e-05,
+      "loss": 1.7998,
+      "step": 2181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4886212414969551e-05,
+      "loss": 1.5938,
+      "step": 2182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4881581613917105e-05,
+      "loss": 1.9019,
+      "step": 2183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4876949438136348e-05,
+      "loss": 1.8965,
+      "step": 2184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4872315888931772e-05,
+      "loss": 1.6836,
+      "step": 2185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4867680967608255e-05,
+      "loss": 1.6719,
+      "step": 2186
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4863044675471067e-05,
+      "loss": 2.0688,
+      "step": 2187
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4858407013825854e-05,
+      "loss": 1.8652,
+      "step": 2188
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.485376798397865e-05,
+      "loss": 1.9131,
+      "step": 2189
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4849127587235886e-05,
+      "loss": 2.1514,
+      "step": 2190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.484448582490436e-05,
+      "loss": 1.6377,
+      "step": 2191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.483984269829127e-05,
+      "loss": 1.8799,
+      "step": 2192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4835198208704186e-05,
+      "loss": 1.9736,
+      "step": 2193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4830552357451075e-05,
+      "loss": 1.9014,
+      "step": 2194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4825905145840273e-05,
+      "loss": 2.1475,
+      "step": 2195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4821256575180507e-05,
+      "loss": 1.96,
+      "step": 2196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4816606646780892e-05,
+      "loss": 1.8428,
+      "step": 2197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4811955361950914e-05,
+      "loss": 1.8584,
+      "step": 2198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4807302722000447e-05,
+      "loss": 1.9204,
+      "step": 2199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4802648728239743e-05,
+      "loss": 2.0176,
+      "step": 2200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4797993381979444e-05,
+      "loss": 2.0977,
+      "step": 2201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4793336684530562e-05,
+      "loss": 1.9199,
+      "step": 2202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4788678637204496e-05,
+      "loss": 1.9443,
+      "step": 2203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4784019241313025e-05,
+      "loss": 2.04,
+      "step": 2204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4779358498168307e-05,
+      "loss": 1.7651,
+      "step": 2205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4774696409082876e-05,
+      "loss": 2.1436,
+      "step": 2206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4770032975369648e-05,
+      "loss": 1.7998,
+      "step": 2207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4765368198341918e-05,
+      "loss": 2.2451,
+      "step": 2208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4760702079313363e-05,
+      "loss": 2.0898,
+      "step": 2209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.475603461959803e-05,
+      "loss": 1.8848,
+      "step": 2210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4751365820510342e-05,
+      "loss": 2.0146,
+      "step": 2211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4746695683365113e-05,
+      "loss": 1.9609,
+      "step": 2212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4742024209477524e-05,
+      "loss": 1.8057,
+      "step": 2213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.473735140016313e-05,
+      "loss": 2.083,
+      "step": 2214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4732677256737869e-05,
+      "loss": 1.8965,
+      "step": 2215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4728001780518048e-05,
+      "loss": 2.1006,
+      "step": 2216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4723324972820358e-05,
+      "loss": 1.979,
+      "step": 2217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4718646834961854e-05,
+      "loss": 2.1807,
+      "step": 2218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4713967368259981e-05,
+      "loss": 2.0566,
+      "step": 2219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4709286574032536e-05,
+      "loss": 1.7695,
+      "step": 2220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4704604453597708e-05,
+      "loss": 2.0332,
+      "step": 2221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4699921008274055e-05,
+      "loss": 1.9834,
+      "step": 2222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4695236239380511e-05,
+      "loss": 1.9639,
+      "step": 2223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4690550148236371e-05,
+      "loss": 2.1367,
+      "step": 2224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4685862736161314e-05,
+      "loss": 1.7412,
+      "step": 2225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.468117400447539e-05,
+      "loss": 2.0645,
+      "step": 2226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.467648395449901e-05,
+      "loss": 1.8682,
+      "step": 2227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4671792587552971e-05,
+      "loss": 1.9561,
+      "step": 2228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.466709990495843e-05,
+      "loss": 1.7607,
+      "step": 2229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4662405908036925e-05,
+      "loss": 1.9199,
+      "step": 2230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4657710598110348e-05,
+      "loss": 2.0195,
+      "step": 2231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4653013976500977e-05,
+      "loss": 1.7764,
+      "step": 2232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.464831604453145e-05,
+      "loss": 1.6709,
+      "step": 2233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4643616803524778e-05,
+      "loss": 2.2598,
+      "step": 2234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4638916254804338e-05,
+      "loss": 1.8164,
+      "step": 2235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.463421439969388e-05,
+      "loss": 2.0342,
+      "step": 2236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4629511239517513e-05,
+      "loss": 2.1855,
+      "step": 2237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4624806775599725e-05,
+      "loss": 1.8721,
+      "step": 2238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.462010100926536e-05,
+      "loss": 2.043,
+      "step": 2239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4615393941839637e-05,
+      "loss": 1.9697,
+      "step": 2240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4610685574648142e-05,
+      "loss": 1.9619,
+      "step": 2241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4605975909016816e-05,
+      "loss": 1.7568,
+      "step": 2242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4601264946271976e-05,
+      "loss": 1.8242,
+      "step": 2243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4596552687740304e-05,
+      "loss": 1.6104,
+      "step": 2244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4591839134748842e-05,
+      "loss": 2.0381,
+      "step": 2245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4587124288624996e-05,
+      "loss": 2.1289,
+      "step": 2246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4582408150696546e-05,
+      "loss": 1.8809,
+      "step": 2247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4577690722291623e-05,
+      "loss": 2.0693,
+      "step": 2248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4572972004738732e-05,
+      "loss": 2.0713,
+      "step": 2249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4568251999366729e-05,
+      "loss": 1.8271,
+      "step": 2250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.456353070750485e-05,
+      "loss": 2.2861,
+      "step": 2251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4558808130482675e-05,
+      "loss": 2.1602,
+      "step": 2252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4554084269630158e-05,
+      "loss": 1.8652,
+      "step": 2253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.454935912627761e-05,
+      "loss": 1.9062,
+      "step": 2254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4544632701755701e-05,
+      "loss": 1.8389,
+      "step": 2255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 2.0127,
+      "step": 2256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4535176014528308e-05,
+      "loss": 2.042,
+      "step": 2257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4530445754485969e-05,
+      "loss": 1.8584,
+      "step": 2258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4525714218600566e-05,
+      "loss": 1.8633,
+      "step": 2259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4520981408204574e-05,
+      "loss": 2.2793,
+      "step": 2260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4516247324630825e-05,
+      "loss": 2.1104,
+      "step": 2261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4511511969212513e-05,
+      "loss": 1.8213,
+      "step": 2262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.450677534328318e-05,
+      "loss": 1.8848,
+      "step": 2263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4502037448176734e-05,
+      "loss": 1.7539,
+      "step": 2264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.449729828522744e-05,
+      "loss": 2.0117,
+      "step": 2265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4492557855769924e-05,
+      "loss": 1.7524,
+      "step": 2266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4487816161139156e-05,
+      "loss": 1.9971,
+      "step": 2267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4483073202670474e-05,
+      "loss": 1.9502,
+      "step": 2268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4478328981699568e-05,
+      "loss": 1.8164,
+      "step": 2269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4473583499562482e-05,
+      "loss": 2.0176,
+      "step": 2270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4468836757595614e-05,
+      "loss": 1.9072,
+      "step": 2271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4464088757135728e-05,
+      "loss": 2.0537,
+      "step": 2272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4459339499519926e-05,
+      "loss": 2.0215,
+      "step": 2273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4454588986085677e-05,
+      "loss": 1.8662,
+      "step": 2274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4449837218170791e-05,
+      "loss": 1.6714,
+      "step": 2275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4445084197113443e-05,
+      "loss": 2.1943,
+      "step": 2276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4440329924252161e-05,
+      "loss": 2.1406,
+      "step": 2277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4435574400925815e-05,
+      "loss": 2.1348,
+      "step": 2278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.443081762847364e-05,
+      "loss": 2.0029,
+      "step": 2279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4426059608235208e-05,
+      "loss": 1.8555,
+      "step": 2280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4421300341550454e-05,
+      "loss": 1.9043,
+      "step": 2281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4416539829759662e-05,
+      "loss": 1.8711,
+      "step": 2282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4411778074203464e-05,
+      "loss": 1.8779,
+      "step": 2283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4407015076222845e-05,
+      "loss": 1.9717,
+      "step": 2284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4402250837159136e-05,
+      "loss": 2.0146,
+      "step": 2285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.439748535835402e-05,
+      "loss": 1.7988,
+      "step": 2286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.439271864114953e-05,
+      "loss": 1.8135,
+      "step": 2287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4387950686888048e-05,
+      "loss": 1.9785,
+      "step": 2288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4383181496912301e-05,
+      "loss": 1.9297,
+      "step": 2289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4378411072565368e-05,
+      "loss": 1.7451,
+      "step": 2290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.437363941519067e-05,
+      "loss": 2.1621,
+      "step": 2291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4368866526131982e-05,
+      "loss": 1.7266,
+      "step": 2292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.436409240673342e-05,
+      "loss": 1.8154,
+      "step": 2293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4359317058339457e-05,
+      "loss": 1.7993,
+      "step": 2294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4354540482294896e-05,
+      "loss": 1.7734,
+      "step": 2295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4349762679944896e-05,
+      "loss": 2.1396,
+      "step": 2296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.434498365263496e-05,
+      "loss": 1.8999,
+      "step": 2297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4340203401710939e-05,
+      "loss": 1.9121,
+      "step": 2298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4335421928519022e-05,
+      "loss": 1.9971,
+      "step": 2299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4330639234405743e-05,
+      "loss": 2.0029,
+      "step": 2300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4325855320717987e-05,
+      "loss": 1.8706,
+      "step": 2301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4321070188802972e-05,
+      "loss": 1.6846,
+      "step": 2302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4316283840008266e-05,
+      "loss": 1.7803,
+      "step": 2303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4311496275681785e-05,
+      "loss": 1.9971,
+      "step": 2304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4306707497171774e-05,
+      "loss": 2.0439,
+      "step": 2305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4301917505826825e-05,
+      "loss": 2.085,
+      "step": 2306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4297126302995876e-05,
+      "loss": 1.9082,
+      "step": 2307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4292333890028205e-05,
+      "loss": 1.8018,
+      "step": 2308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4287540268273428e-05,
+      "loss": 1.7236,
+      "step": 2309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.42827454390815e-05,
+      "loss": 1.9326,
+      "step": 2310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.427794940380272e-05,
+      "loss": 1.9053,
+      "step": 2311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4273152163787726e-05,
+      "loss": 1.7354,
+      "step": 2312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4268353720387495e-05,
+      "loss": 1.7437,
+      "step": 2313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4263554074953338e-05,
+      "loss": 1.9502,
+      "step": 2314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4258753228836914e-05,
+      "loss": 1.9795,
+      "step": 2315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4253951183390216e-05,
+      "loss": 2.1152,
+      "step": 2316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4249147939965566e-05,
+      "loss": 1.8418,
+      "step": 2317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4244343499915636e-05,
+      "loss": 2.0869,
+      "step": 2318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4239537864593432e-05,
+      "loss": 1.8965,
+      "step": 2319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4234731035352291e-05,
+      "loss": 1.9072,
+      "step": 2320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4229923013545892e-05,
+      "loss": 1.9521,
+      "step": 2321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4225113800528247e-05,
+      "loss": 1.8867,
+      "step": 2322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4220303397653703e-05,
+      "loss": 1.8862,
+      "step": 2323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4215491806276944e-05,
+      "loss": 2.1523,
+      "step": 2324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4210679027752989e-05,
+      "loss": 1.6123,
+      "step": 2325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4205865063437192e-05,
+      "loss": 1.9365,
+      "step": 2326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4201049914685236e-05,
+      "loss": 1.6602,
+      "step": 2327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.419623358285314e-05,
+      "loss": 1.915,
+      "step": 2328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4191416069297261e-05,
+      "loss": 1.8916,
+      "step": 2329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4186597375374283e-05,
+      "loss": 1.832,
+      "step": 2330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4181777502441222e-05,
+      "loss": 1.9678,
+      "step": 2331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4176956451855431e-05,
+      "loss": 1.8945,
+      "step": 2332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.417213422497459e-05,
+      "loss": 1.9727,
+      "step": 2333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4167310823156713e-05,
+      "loss": 1.7725,
+      "step": 2334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4162486247760143e-05,
+      "loss": 1.8945,
+      "step": 2335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4157660500143555e-05,
+      "loss": 2.0771,
+      "step": 2336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4152833581665956e-05,
+      "loss": 1.9043,
+      "step": 2337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4148005493686681e-05,
+      "loss": 1.8125,
+      "step": 2338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4143176237565386e-05,
+      "loss": 1.9238,
+      "step": 2339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4138345814662068e-05,
+      "loss": 1.7959,
+      "step": 2340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4133514226337052e-05,
+      "loss": 1.9102,
+      "step": 2341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4128681473950984e-05,
+      "loss": 1.7695,
+      "step": 2342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4123847558864841e-05,
+      "loss": 2.0371,
+      "step": 2343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4119012482439929e-05,
+      "loss": 1.8198,
+      "step": 2344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.411417624603788e-05,
+      "loss": 2.0908,
+      "step": 2345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4109338851020646e-05,
+      "loss": 1.8965,
+      "step": 2346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4104500298750521e-05,
+      "loss": 2.1797,
+      "step": 2347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4099660590590115e-05,
+      "loss": 1.9043,
+      "step": 2348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4094819727902354e-05,
+      "loss": 1.6201,
+      "step": 2349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4089977712050508e-05,
+      "loss": 2.0498,
+      "step": 2350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4085134544398163e-05,
+      "loss": 2.0029,
+      "step": 2351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4080290226309225e-05,
+      "loss": 1.8057,
+      "step": 2352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4075444759147932e-05,
+      "loss": 1.5518,
+      "step": 2353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.407059814427884e-05,
+      "loss": 1.6802,
+      "step": 2354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4065750383066832e-05,
+      "loss": 1.814,
+      "step": 2355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4060901476877107e-05,
+      "loss": 1.8643,
+      "step": 2356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4056051427075196e-05,
+      "loss": 1.8955,
+      "step": 2357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4051200235026952e-05,
+      "loss": 1.7227,
+      "step": 2358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4046347902098535e-05,
+      "loss": 1.9424,
+      "step": 2359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4041494429656442e-05,
+      "loss": 2.0215,
+      "step": 2360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4036639819067484e-05,
+      "loss": 1.6729,
+      "step": 2361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4031784071698796e-05,
+      "loss": 1.8496,
+      "step": 2362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4026927188917827e-05,
+      "loss": 2.043,
+      "step": 2363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4022069172092354e-05,
+      "loss": 1.9697,
+      "step": 2364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4017210022590466e-05,
+      "loss": 1.6729,
+      "step": 2365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4012349741780576e-05,
+      "loss": 1.9854,
+      "step": 2366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4007488331031409e-05,
+      "loss": 1.9336,
+      "step": 2367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4002625791712021e-05,
+      "loss": 1.7637,
+      "step": 2368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3997762125191774e-05,
+      "loss": 2.3672,
+      "step": 2369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.399289733284035e-05,
+      "loss": 1.8994,
+      "step": 2370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3988031416027747e-05,
+      "loss": 1.8799,
+      "step": 2371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3983164376124287e-05,
+      "loss": 1.9414,
+      "step": 2372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3978296214500601e-05,
+      "loss": 1.9375,
+      "step": 2373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3973426932527637e-05,
+      "loss": 2.2285,
+      "step": 2374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3968556531576663e-05,
+      "loss": 1.9316,
+      "step": 2375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3963685013019253e-05,
+      "loss": 1.7803,
+      "step": 2376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3958812378227306e-05,
+      "loss": 1.7705,
+      "step": 2377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3953938628573027e-05,
+      "loss": 1.9385,
+      "step": 2378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3949063765428943e-05,
+      "loss": 2.1934,
+      "step": 2379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3944187790167889e-05,
+      "loss": 2.0762,
+      "step": 2380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3939310704163012e-05,
+      "loss": 2.1582,
+      "step": 2381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3934432508787776e-05,
+      "loss": 1.9697,
+      "step": 2382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3929553205415958e-05,
+      "loss": 1.7129,
+      "step": 2383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3924672795421638e-05,
+      "loss": 1.9561,
+      "step": 2384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3919791280179223e-05,
+      "loss": 1.8701,
+      "step": 2385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3914908661063416e-05,
+      "loss": 1.8047,
+      "step": 2386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.391002493944924e-05,
+      "loss": 2.0391,
+      "step": 2387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3905140116712026e-05,
+      "loss": 1.8389,
+      "step": 2388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3900254194227417e-05,
+      "loss": 1.7568,
+      "step": 2389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3895367173371356e-05,
+      "loss": 1.7466,
+      "step": 2390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3890479055520116e-05,
+      "loss": 1.6641,
+      "step": 2391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3885589842050253e-05,
+      "loss": 1.8955,
+      "step": 2392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3880699534338656e-05,
+      "loss": 1.7861,
+      "step": 2393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.38758081337625e-05,
+      "loss": 1.9824,
+      "step": 2394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3870915641699285e-05,
+      "loss": 1.8867,
+      "step": 2395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3866022059526812e-05,
+      "loss": 1.8955,
+      "step": 2396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3861127388623186e-05,
+      "loss": 1.7832,
+      "step": 2397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3856231630366823e-05,
+      "loss": 2.1621,
+      "step": 2398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.385133478613644e-05,
+      "loss": 1.6953,
+      "step": 2399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3846436857311069e-05,
+      "loss": 1.636,
+      "step": 2400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.384153784527004e-05,
+      "loss": 2.0332,
+      "step": 2401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3836637751392994e-05,
+      "loss": 1.875,
+      "step": 2402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3831736577059862e-05,
+      "loss": 1.8955,
+      "step": 2403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 1.7559,
+      "step": 2404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3821930992546653e-05,
+      "loss": 1.8359,
+      "step": 2405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3817026585127972e-05,
+      "loss": 1.9346,
+      "step": 2406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3812121102776024e-05,
+      "loss": 1.8506,
+      "step": 2407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3807214546872256e-05,
+      "loss": 1.5283,
+      "step": 2408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3802306918798435e-05,
+      "loss": 2.1045,
+      "step": 2409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3797398219936623e-05,
+      "loss": 1.6733,
+      "step": 2410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3792488451669188e-05,
+      "loss": 2.0713,
+      "step": 2411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3787577615378793e-05,
+      "loss": 1.6816,
+      "step": 2412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3782665712448406e-05,
+      "loss": 2.002,
+      "step": 2413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3777752744261295e-05,
+      "loss": 2.0391,
+      "step": 2414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3772838712201027e-05,
+      "loss": 2.1113,
+      "step": 2415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3767923617651464e-05,
+      "loss": 1.4922,
+      "step": 2416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3763007461996778e-05,
+      "loss": 2.1553,
+      "step": 2417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3758090246621434e-05,
+      "loss": 1.7158,
+      "step": 2418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3753171972910191e-05,
+      "loss": 1.9717,
+      "step": 2419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3748252642248115e-05,
+      "loss": 1.7363,
+      "step": 2420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3743332256020559e-05,
+      "loss": 1.9814,
+      "step": 2421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3738410815613186e-05,
+      "loss": 1.9785,
+      "step": 2422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3733488322411945e-05,
+      "loss": 1.6987,
+      "step": 2423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3728564777803089e-05,
+      "loss": 2.0098,
+      "step": 2424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.372364018317316e-05,
+      "loss": 1.7598,
+      "step": 2425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3718714539909e-05,
+      "loss": 1.6953,
+      "step": 2426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3713787849397748e-05,
+      "loss": 2.0723,
+      "step": 2427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3708860113026834e-05,
+      "loss": 1.9434,
+      "step": 2428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3703931332183987e-05,
+      "loss": 1.6182,
+      "step": 2429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.369900150825722e-05,
+      "loss": 1.8887,
+      "step": 2430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3694070642634854e-05,
+      "loss": 2.2764,
+      "step": 2431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3689138736705496e-05,
+      "loss": 1.9297,
+      "step": 2432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3684205791858043e-05,
+      "loss": 1.7295,
+      "step": 2433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3679271809481693e-05,
+      "loss": 1.8315,
+      "step": 2434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3674336790965925e-05,
+      "loss": 1.7988,
+      "step": 2435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3669400737700523e-05,
+      "loss": 1.4824,
+      "step": 2436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3664463651075547e-05,
+      "loss": 1.8584,
+      "step": 2437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3659525532481362e-05,
+      "loss": 1.8179,
+      "step": 2438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3654586383308619e-05,
+      "loss": 1.8662,
+      "step": 2439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3649646204948254e-05,
+      "loss": 1.7271,
+      "step": 2440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3644704998791501e-05,
+      "loss": 1.8926,
+      "step": 2441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3639762766229872e-05,
+      "loss": 1.835,
+      "step": 2442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3634819508655186e-05,
+      "loss": 1.959,
+      "step": 2443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3629875227459532e-05,
+      "loss": 1.7305,
+      "step": 2444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3624929924035301e-05,
+      "loss": 1.687,
+      "step": 2445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3619983599775167e-05,
+      "loss": 1.5718,
+      "step": 2446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3615036256072083e-05,
+      "loss": 1.998,
+      "step": 2447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3610087894319303e-05,
+      "loss": 1.7539,
+      "step": 2448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3605138515910362e-05,
+      "loss": 1.8301,
+      "step": 2449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.360018812223908e-05,
+      "loss": 1.8135,
+      "step": 2450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3595236714699562e-05,
+      "loss": 2.0771,
+      "step": 2451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3590284294686202e-05,
+      "loss": 1.7319,
+      "step": 2452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3585330863593674e-05,
+      "loss": 1.585,
+      "step": 2453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3580376422816945e-05,
+      "loss": 1.9512,
+      "step": 2454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3575420973751263e-05,
+      "loss": 1.7964,
+      "step": 2455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3570464517792155e-05,
+      "loss": 2.0732,
+      "step": 2456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3565507056335431e-05,
+      "loss": 1.9453,
+      "step": 2457
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3560548590777196e-05,
+      "loss": 2.0664,
+      "step": 2458
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3555589122513828e-05,
+      "loss": 1.9111,
+      "step": 2459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3550628652941983e-05,
+      "loss": 1.895,
+      "step": 2460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3545667183458616e-05,
+      "loss": 1.7412,
+      "step": 2461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3540704715460946e-05,
+      "loss": 1.9756,
+      "step": 2462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.353574125034648e-05,
+      "loss": 1.7754,
+      "step": 2463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3530776789513009e-05,
+      "loss": 1.9668,
+      "step": 2464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.35258113343586e-05,
+      "loss": 1.8037,
+      "step": 2465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3520844886281601e-05,
+      "loss": 1.8213,
+      "step": 2466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.351587744668064e-05,
+      "loss": 2.0449,
+      "step": 2467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3510909016954626e-05,
+      "loss": 1.9004,
+      "step": 2468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3505939598502742e-05,
+      "loss": 1.9971,
+      "step": 2469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3500969192724451e-05,
+      "loss": 1.7881,
+      "step": 2470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3495997801019506e-05,
+      "loss": 2.2051,
+      "step": 2471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3491025424787916e-05,
+      "loss": 1.624,
+      "step": 2472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3486052065429985e-05,
+      "loss": 2.0107,
+      "step": 2473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3481077724346279e-05,
+      "loss": 2.0186,
+      "step": 2474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.347610240293766e-05,
+      "loss": 2.0244,
+      "step": 2475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3471126102605245e-05,
+      "loss": 2.2588,
+      "step": 2476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3466148824750444e-05,
+      "loss": 1.9141,
+      "step": 2477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.346117057077493e-05,
+      "loss": 1.8975,
+      "step": 2478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.345619134208066e-05,
+      "loss": 1.7188,
+      "step": 2479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3451211140069857e-05,
+      "loss": 1.752,
+      "step": 2480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3446229966145024e-05,
+      "loss": 1.7915,
+      "step": 2481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3441247821708939e-05,
+      "loss": 1.7695,
+      "step": 2482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3436264708164645e-05,
+      "loss": 1.8184,
+      "step": 2483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3431280626915466e-05,
+      "loss": 2.085,
+      "step": 2484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3426295579364995e-05,
+      "loss": 1.5869,
+      "step": 2485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3421309566917098e-05,
+      "loss": 1.834,
+      "step": 2486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3416322590975911e-05,
+      "loss": 1.9561,
+      "step": 2487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3411334652945852e-05,
+      "loss": 1.9277,
+      "step": 2488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.340634575423159e-05,
+      "loss": 1.833,
+      "step": 2489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3401355896238077e-05,
+      "loss": 1.7402,
+      "step": 2490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3396365080370539e-05,
+      "loss": 1.8203,
+      "step": 2491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3391373308034461e-05,
+      "loss": 1.9072,
+      "step": 2492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3386380580635606e-05,
+      "loss": 1.998,
+      "step": 2493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3381386899580005e-05,
+      "loss": 1.7832,
+      "step": 2494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3376392266273947e-05,
+      "loss": 1.8018,
+      "step": 2495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3371396682124006e-05,
+      "loss": 1.9062,
+      "step": 2496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3366400148537008e-05,
+      "loss": 1.9062,
+      "step": 2497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.336140266692006e-05,
+      "loss": 2.0273,
+      "step": 2498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3356404238680528e-05,
+      "loss": 1.8623,
+      "step": 2499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3351404865226041e-05,
+      "loss": 2.082,
+      "step": 2500
+    }
+  ],
+  "max_steps": 6104,
+  "num_train_epochs": 2,
+  "total_flos": 3.721559290905887e+17,
+  "trial_name": null,
+  "trial_params": null
+}