diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9616 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.28,
+  "global_step": 1600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7699115044247788e-07,
+      "loss": 2.6342,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5398230088495575e-07,
+      "loss": 2.4942,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.309734513274336e-07,
+      "loss": 2.4458,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.079646017699115e-07,
+      "loss": 2.3624,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.849557522123895e-07,
+      "loss": 2.2731,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0619469026548673e-06,
+      "loss": 2.3463,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2389380530973452e-06,
+      "loss": 2.4769,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.415929203539823e-06,
+      "loss": 2.3047,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.592920353982301e-06,
+      "loss": 2.4436,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.769911504424779e-06,
+      "loss": 2.2426,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9469026548672567e-06,
+      "loss": 2.3656,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1238938053097345e-06,
+      "loss": 2.3785,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3008849557522127e-06,
+      "loss": 2.3527,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4778761061946905e-06,
+      "loss": 2.2126,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6548672566371687e-06,
+      "loss": 2.3341,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.831858407079646e-06,
+      "loss": 2.2448,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0088495575221242e-06,
+      "loss": 2.1304,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.185840707964602e-06,
+      "loss": 2.2165,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.36283185840708e-06,
+      "loss": 2.3042,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.539823008849558e-06,
+      "loss": 2.2978,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7168141592920357e-06,
+      "loss": 2.2188,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8938053097345135e-06,
+      "loss": 2.1372,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.070796460176992e-06,
+      "loss": 2.2388,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.247787610619469e-06,
+      "loss": 2.1211,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.424778761061948e-06,
+      "loss": 2.2362,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6017699115044254e-06,
+      "loss": 2.2629,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.778761061946903e-06,
+      "loss": 2.3319,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.955752212389381e-06,
+      "loss": 2.1444,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.132743362831859e-06,
+      "loss": 2.0985,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.309734513274337e-06,
+      "loss": 2.2057,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.486725663716814e-06,
+      "loss": 2.2332,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.663716814159292e-06,
+      "loss": 2.2764,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.840707964601771e-06,
+      "loss": 2.2456,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0176991150442484e-06,
+      "loss": 2.0918,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.194690265486726e-06,
+      "loss": 2.0425,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.371681415929204e-06,
+      "loss": 2.2086,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.548672566371682e-06,
+      "loss": 2.1651,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.72566371681416e-06,
+      "loss": 2.1239,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.902654867256637e-06,
+      "loss": 2.0779,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.079646017699116e-06,
+      "loss": 2.1271,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.256637168141594e-06,
+      "loss": 2.2926,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4336283185840714e-06,
+      "loss": 2.1001,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.610619469026549e-06,
+      "loss": 2.2959,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.787610619469027e-06,
+      "loss": 2.0291,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.964601769911505e-06,
+      "loss": 2.1234,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.141592920353984e-06,
+      "loss": 2.1356,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.31858407079646e-06,
+      "loss": 2.1785,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.495575221238938e-06,
+      "loss": 2.151,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.672566371681418e-06,
+      "loss": 2.0698,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.849557522123895e-06,
+      "loss": 2.2233,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.026548672566371e-06,
+      "loss": 2.2075,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.203539823008851e-06,
+      "loss": 2.1574,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.380530973451329e-06,
+      "loss": 2.1002,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.557522123893806e-06,
+      "loss": 2.1708,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.734513274336284e-06,
+      "loss": 2.0356,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.911504424778762e-06,
+      "loss": 2.196,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.008849557522124e-05,
+      "loss": 2.1957,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0265486725663717e-05,
+      "loss": 2.113,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0442477876106197e-05,
+      "loss": 2.1256,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0619469026548675e-05,
+      "loss": 2.1362,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.079646017699115e-05,
+      "loss": 2.0819,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0973451327433629e-05,
+      "loss": 2.0522,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1150442477876106e-05,
+      "loss": 2.2226,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1327433628318584e-05,
+      "loss": 2.1568,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1504424778761064e-05,
+      "loss": 1.9969,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1681415929203541e-05,
+      "loss": 2.0662,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1858407079646019e-05,
+      "loss": 2.1751,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2035398230088497e-05,
+      "loss": 2.2176,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2212389380530973e-05,
+      "loss": 2.0555,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2389380530973452e-05,
+      "loss": 2.1516,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.256637168141593e-05,
+      "loss": 2.258,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2743362831858408e-05,
+      "loss": 2.0608,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2920353982300886e-05,
+      "loss": 2.1596,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3097345132743363e-05,
+      "loss": 2.1454,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3274336283185843e-05,
+      "loss": 2.0923,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.345132743362832e-05,
+      "loss": 2.0999,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3628318584070797e-05,
+      "loss": 2.0944,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3805309734513275e-05,
+      "loss": 2.1493,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3982300884955752e-05,
+      "loss": 2.1631,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4159292035398232e-05,
+      "loss": 2.1735,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.433628318584071e-05,
+      "loss": 2.2144,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4513274336283187e-05,
+      "loss": 2.1732,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4690265486725665e-05,
+      "loss": 2.1125,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4867256637168143e-05,
+      "loss": 2.1681,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5044247787610619e-05,
+      "loss": 2.0783,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5221238938053098e-05,
+      "loss": 2.0296,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5398230088495576e-05,
+      "loss": 2.1401,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5575221238938054e-05,
+      "loss": 2.1291,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5752212389380532e-05,
+      "loss": 2.1044,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.592920353982301e-05,
+      "loss": 2.0911,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6106194690265487e-05,
+      "loss": 2.1175,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.628318584070797e-05,
+      "loss": 2.1085,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6460176991150443e-05,
+      "loss": 2.1192,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.663716814159292e-05,
+      "loss": 2.1747,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.68141592920354e-05,
+      "loss": 2.1093,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6991150442477876e-05,
+      "loss": 2.3442,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7168141592920354e-05,
+      "loss": 2.0882,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7345132743362835e-05,
+      "loss": 2.2757,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7522123893805313e-05,
+      "loss": 2.1743,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.769911504424779e-05,
+      "loss": 2.1613,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7876106194690265e-05,
+      "loss": 2.0899,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8053097345132743e-05,
+      "loss": 2.1127,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.823008849557522e-05,
+      "loss": 2.0457,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8407079646017702e-05,
+      "loss": 2.0061,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.858407079646018e-05,
+      "loss": 2.0532,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8761061946902657e-05,
+      "loss": 2.2078,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8938053097345135e-05,
+      "loss": 2.1356,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9115044247787613e-05,
+      "loss": 2.1502,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.929203539823009e-05,
+      "loss": 2.2263,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.946902654867257e-05,
+      "loss": 2.1397,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9646017699115046e-05,
+      "loss": 2.0656,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823008849557524e-05,
+      "loss": 2.0862,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 2.1775,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999626936191e-05,
+      "loss": 2.0169,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999985077450406e-05,
+      "loss": 2.1069,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999966424273852e-05,
+      "loss": 2.1435,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999994030984616e-05,
+      "loss": 2.2267,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999906734186815e-05,
+      "loss": 1.997,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999986569732087e-05,
+      "loss": 2.2769,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999817199278942e-05,
+      "loss": 1.9394,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999761240097216e-05,
+      "loss": 2.121,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999969781981745e-05,
+      "loss": 2.2084,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999626938486954e-05,
+      "loss": 2.0314,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999548596158626e-05,
+      "loss": 2.2046,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999462792890913e-05,
+      "loss": 2.1457,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999369528747838e-05,
+      "loss": 2.1497,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999926880379898e-05,
+      "loss": 2.0739,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999160618119502e-05,
+      "loss": 2.2034,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999904497179012e-05,
+      "loss": 2.1544,
+      "step": 129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998921864897123e-05,
+      "loss": 2.2104,
+      "step": 130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998791297532362e-05,
+      "loss": 2.052,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999865326979326e-05,
+      "loss": 2.2697,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998507781782802e-05,
+      "loss": 2.0682,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998354833609537e-05,
+      "loss": 2.0808,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998194425387588e-05,
+      "loss": 2.0473,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999802655723664e-05,
+      "loss": 2.1415,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997851229281942e-05,
+      "loss": 2.0943,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999766844165431e-05,
+      "loss": 2.162,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997478194490135e-05,
+      "loss": 2.1148,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997280487931355e-05,
+      "loss": 2.3097,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997075322125492e-05,
+      "loss": 2.0729,
+      "step": 141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996862697225624e-05,
+      "loss": 2.2898,
+      "step": 142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999664261339039e-05,
+      "loss": 2.1436,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996415070784007e-05,
+      "loss": 2.0268,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999618006957625e-05,
+      "loss": 2.1917,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995937609942463e-05,
+      "loss": 2.0689,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995687692063547e-05,
+      "loss": 2.0695,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999543031612597e-05,
+      "loss": 2.0363,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995165482321775e-05,
+      "loss": 2.0635,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994893190848556e-05,
+      "loss": 2.2353,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999461344190948e-05,
+      "loss": 2.1787,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994326235713278e-05,
+      "loss": 2.136,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994031572474238e-05,
+      "loss": 2.0413,
+      "step": 153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993729452412214e-05,
+      "loss": 2.1219,
+      "step": 154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993419875752632e-05,
+      "loss": 2.1356,
+      "step": 155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999310284272647e-05,
+      "loss": 2.0748,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992778353570284e-05,
+      "loss": 2.0701,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992446408526177e-05,
+      "loss": 2.238,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999210700784182e-05,
+      "loss": 2.194,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999176015177046e-05,
+      "loss": 2.2741,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991405840570886e-05,
+      "loss": 2.1712,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991044074507463e-05,
+      "loss": 2.0362,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990674853850113e-05,
+      "loss": 2.0989,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990298178874322e-05,
+      "loss": 2.1951,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989914049861145e-05,
+      "loss": 2.1943,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998952246709718e-05,
+      "loss": 2.1334,
+      "step": 166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989123430874602e-05,
+      "loss": 2.1329,
+      "step": 167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988716941491142e-05,
+      "loss": 2.1719,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99883029992501e-05,
+      "loss": 2.0233,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998788160446032e-05,
+      "loss": 2.2163,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998745275743622e-05,
+      "loss": 2.2141,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998701645849778e-05,
+      "loss": 2.2667,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986572707970527e-05,
+      "loss": 2.052,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986121506185557e-05,
+      "loss": 2.0601,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985662853479525e-05,
+      "loss": 2.089,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985196750194647e-05,
+      "loss": 2.2365,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984723196678694e-05,
+      "loss": 2.167,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984242193284996e-05,
+      "loss": 2.0078,
+      "step": 178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983753740372442e-05,
+      "loss": 2.0686,
+      "step": 179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983257838305487e-05,
+      "loss": 2.1354,
+      "step": 180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982754487454124e-05,
+      "loss": 1.9609,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982243688193935e-05,
+      "loss": 2.1828,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981725440906023e-05,
+      "loss": 2.195,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998119974597708e-05,
+      "loss": 2.1612,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980666603799336e-05,
+      "loss": 2.115,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998012601477058e-05,
+      "loss": 2.0758,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997957797929417e-05,
+      "loss": 2.1763,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979022497779003e-05,
+      "loss": 2.1175,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997845957063954e-05,
+      "loss": 2.104,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977889198295794e-05,
+      "loss": 2.1183,
+      "step": 190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997731138117334e-05,
+      "loss": 2.2253,
+      "step": 191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997672611970331e-05,
+      "loss": 2.0031,
+      "step": 192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997613341432237e-05,
+      "loss": 2.1712,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975533265472756e-05,
+      "loss": 2.187,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974925673602263e-05,
+      "loss": 2.1687,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997431063916423e-05,
+      "loss": 2.0695,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973688162617545e-05,
+      "loss": 2.1737,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973058244426663e-05,
+      "loss": 2.0203,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972420885061576e-05,
+      "loss": 2.2102,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971776084997844e-05,
+      "loss": 2.156,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971123844716562e-05,
+      "loss": 2.0291,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970464164704387e-05,
+      "loss": 2.0157,
+      "step": 202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996979704545353e-05,
+      "loss": 2.1021,
+      "step": 203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996912248746174e-05,
+      "loss": 2.1,
+      "step": 204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968440491232326e-05,
+      "loss": 2.078,
+      "step": 205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967751057274147e-05,
+      "loss": 2.0041,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99670541861016e-05,
+      "loss": 2.1863,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996634987823465e-05,
+      "loss": 2.1564,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965638134198792e-05,
+      "loss": 2.0383,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964918954525086e-05,
+      "loss": 2.073,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996419233975013e-05,
+      "loss": 2.2136,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963458290416066e-05,
+      "loss": 1.9862,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962716807070592e-05,
+      "loss": 2.0693,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961967890266948e-05,
+      "loss": 2.0951,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996121154056392e-05,
+      "loss": 2.1478,
+      "step": 215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960447758525846e-05,
+      "loss": 2.1191,
+      "step": 216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959676544722605e-05,
+      "loss": 2.1407,
+      "step": 217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958897899729616e-05,
+      "loss": 2.2314,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995811182412785e-05,
+      "loss": 2.1224,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995731831850382e-05,
+      "loss": 2.2138,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956517383449587e-05,
+      "loss": 2.0954,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995570901956274e-05,
+      "loss": 2.0866,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995489322744643e-05,
+      "loss": 1.984,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995407000770934e-05,
+      "loss": 2.1659,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953239360965697e-05,
+      "loss": 2.134,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995240128783527e-05,
+      "loss": 2.0198,
+      "step": 226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951555788943364e-05,
+      "loss": 2.1189,
+      "step": 227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950702864920837e-05,
+      "loss": 2.0313,
+      "step": 228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949842516404073e-05,
+      "loss": 2.0241,
+      "step": 229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948974744035002e-05,
+      "loss": 2.003,
+      "step": 230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948099548461098e-05,
+      "loss": 2.1218,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947216930335363e-05,
+      "loss": 2.051,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946326890316345e-05,
+      "loss": 2.09,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945429429068127e-05,
+      "loss": 2.1685,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944524547260334e-05,
+      "loss": 2.1266,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943612245568115e-05,
+      "loss": 2.1289,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994269252467217e-05,
+      "loss": 2.097,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941765385258723e-05,
+      "loss": 2.0301,
+      "step": 238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940830828019547e-05,
+      "loss": 2.1081,
+      "step": 239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939888853651933e-05,
+      "loss": 2.1121,
+      "step": 240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938939462858714e-05,
+      "loss": 2.1643,
+      "step": 241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937982656348262e-05,
+      "loss": 2.0362,
+      "step": 242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937018434834472e-05,
+      "loss": 2.0981,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993604679903678e-05,
+      "loss": 2.0792,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935067749680145e-05,
+      "loss": 2.057,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993408128749507e-05,
+      "loss": 2.0001,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933087413217575e-05,
+      "loss": 2.133,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993208612758922e-05,
+      "loss": 2.0653,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931077431357095e-05,
+      "loss": 2.2018,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930061325273812e-05,
+      "loss": 2.0735,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929037810097516e-05,
+      "loss": 2.1075,
+      "step": 251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992800688659188e-05,
+      "loss": 2.0776,
+      "step": 252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926968555526108e-05,
+      "loss": 2.1681,
+      "step": 253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925922817674923e-05,
+      "loss": 2.0747,
+      "step": 254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992486967381858e-05,
+      "loss": 2.1578,
+      "step": 255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992380912474286e-05,
+      "loss": 2.1087,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922741171239064e-05,
+      "loss": 2.0957,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921665814104027e-05,
+      "loss": 2.1029,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920583054140104e-05,
+      "loss": 2.1479,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919492892155164e-05,
+      "loss": 2.1984,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918395328962615e-05,
+      "loss": 2.2156,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917290365381373e-05,
+      "loss": 2.1825,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916178002235886e-05,
+      "loss": 2.1642,
+      "step": 263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991505824035612e-05,
+      "loss": 2.0837,
+      "step": 264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913931080577554e-05,
+      "loss": 2.0187,
+      "step": 265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99127965237412e-05,
+      "loss": 2.1871,
+      "step": 266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911654570693576e-05,
+      "loss": 2.0616,
+      "step": 267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991050522228673e-05,
+      "loss": 2.1396,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909348479378216e-05,
+      "loss": 2.0462,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990818434283112e-05,
+      "loss": 2.0684,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990701281351403e-05,
+      "loss": 2.0668,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905833892301067e-05,
+      "loss": 2.1311,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990464758007184e-05,
+      "loss": 2.2016,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99034538777115e-05,
+      "loss": 2.2369,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902252786110702e-05,
+      "loss": 2.0509,
+      "step": 275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901044306165607e-05,
+      "loss": 2.1006,
+      "step": 276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98998284387779e-05,
+      "loss": 2.1585,
+      "step": 277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898605184854773e-05,
+      "loss": 2.2726,
+      "step": 278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897374545308928e-05,
+      "loss": 2.2056,
+      "step": 279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989613652105858e-05,
+      "loss": 2.0896,
+      "step": 280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894891113027457e-05,
+      "loss": 2.0929,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989363832214479e-05,
+      "loss": 2.0166,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989237814934531e-05,
+      "loss": 2.1788,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891110595569283e-05,
+      "loss": 1.9928,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889835661762457e-05,
+      "loss": 2.1352,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98885533488761e-05,
+      "loss": 2.1533,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887263657866974e-05,
+      "loss": 2.0889,
+      "step": 287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988596658969736e-05,
+      "loss": 2.1308,
+      "step": 288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884662145335033e-05,
+      "loss": 2.1413,
+      "step": 289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883350325753276e-05,
+      "loss": 2.1864,
+      "step": 290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882031131930876e-05,
+      "loss": 2.042,
+      "step": 291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880704564852112e-05,
+      "loss": 2.2396,
+      "step": 292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879370625506783e-05,
+      "loss": 2.1106,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987802931489017e-05,
+      "loss": 2.1762,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876680634003068e-05,
+      "loss": 2.0019,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987532458385176e-05,
+      "loss": 2.1555,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987396116544803e-05,
+      "loss": 2.0512,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872590379809173e-05,
+      "loss": 2.1658,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871212227957962e-05,
+      "loss": 2.1243,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869826710922676e-05,
+      "loss": 2.1094,
+      "step": 300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868433829737085e-05,
+      "loss": 2.0905,
+      "step": 301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867033585440457e-05,
+      "loss": 2.0737,
+      "step": 302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865625979077557e-05,
+      "loss": 2.1049,
+      "step": 303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864211011698635e-05,
+      "loss": 2.1059,
+      "step": 304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986278868435944e-05,
+      "loss": 2.0882,
+      "step": 305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861358998121207e-05,
+      "loss": 1.9053,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859921954050664e-05,
+      "loss": 2.0445,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858477553220034e-05,
+      "loss": 2.1344,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857025796707018e-05,
+      "loss": 2.0787,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855566685594817e-05,
+      "loss": 2.1858,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854100220972112e-05,
+      "loss": 2.0423,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985262640393307e-05,
+      "loss": 2.1353,
+      "step": 312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851145235577352e-05,
+      "loss": 2.157,
+      "step": 313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849656717010094e-05,
+      "loss": 2.0816,
+      "step": 314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848160849341927e-05,
+      "loss": 2.032,
+      "step": 315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984665763368895e-05,
+      "loss": 1.9813,
+      "step": 316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984514707117276e-05,
+      "loss": 2.1839,
+      "step": 317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843629162920428e-05,
+      "loss": 2.0142,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9842103910064507e-05,
+      "loss": 2.1252,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984057131374303e-05,
+      "loss": 2.1124,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839031375099514e-05,
+      "loss": 2.1058,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983748409528294e-05,
+      "loss": 2.1472,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835929475447786e-05,
+      "loss": 2.2478,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834367516753992e-05,
+      "loss": 2.0825,
+      "step": 324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832798220366977e-05,
+      "loss": 2.1809,
+      "step": 325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831221587457642e-05,
+      "loss": 2.081,
+      "step": 326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982963761920235e-05,
+      "loss": 2.1381,
+      "step": 327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982804631678295e-05,
+      "loss": 2.0337,
+      "step": 328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982644768138675e-05,
+      "loss": 1.9488,
+      "step": 329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982484171420654e-05,
+      "loss": 1.9757,
+      "step": 330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823228416440577e-05,
+      "loss": 1.9545,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821607789292584e-05,
+      "loss": 2.1547,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819979833971756e-05,
+      "loss": 2.0237,
+      "step": 333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981834455169276e-05,
+      "loss": 2.1029,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981670194367572e-05,
+      "loss": 2.0786,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815052011146233e-05,
+      "loss": 2.1815,
+      "step": 336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981339475533536e-05,
+      "loss": 2.1501,
+      "step": 337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811730177479624e-05,
+      "loss": 2.0851,
+      "step": 338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810058278821017e-05,
+      "loss": 2.0512,
+      "step": 339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980837906060698e-05,
+      "loss": 2.1895,
+      "step": 340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806692524090434e-05,
+      "loss": 2.0738,
+      "step": 341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804998670529742e-05,
+      "loss": 2.0929,
+      "step": 342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980329750118874e-05,
+      "loss": 2.1284,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801589017336715e-05,
+      "loss": 2.0856,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799873220248417e-05,
+      "loss": 2.1048,
+      "step": 345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798150111204045e-05,
+      "loss": 1.9984,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796419691489263e-05,
+      "loss": 2.2915,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794681962395182e-05,
+      "loss": 2.2062,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979293692521837e-05,
+      "loss": 2.2924,
+      "step": 349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979118458126085e-05,
+      "loss": 2.1264,
+      "step": 350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978942493183009e-05,
+      "loss": 2.0657,
+      "step": 351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787657978239014e-05,
+      "loss": 2.1687,
+      "step": 352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785883721805997e-05,
+      "loss": 2.1672,
+      "step": 353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784102163854862e-05,
+      "loss": 2.0972,
+      "step": 354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782313305714873e-05,
+      "loss": 2.217,
+      "step": 355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780517148720752e-05,
+      "loss": 2.1013,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778713694212662e-05,
+      "loss": 2.0173,
+      "step": 357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776902943536203e-05,
+      "loss": 2.094,
+      "step": 358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977508489804243e-05,
+      "loss": 2.2898,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773259559087838e-05,
+      "loss": 2.1117,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771426928034362e-05,
+      "loss": 2.2154,
+      "step": 361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976958700624938e-05,
+      "loss": 2.1857,
+      "step": 362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767739795105708e-05,
+      "loss": 2.089,
+      "step": 363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97658852959816e-05,
+      "loss": 2.0843,
+      "step": 364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976402351026075e-05,
+      "loss": 2.1967,
+      "step": 365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976215443933229e-05,
+      "loss": 2.0014,
+      "step": 366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976027808459078e-05,
+      "loss": 2.0905,
+      "step": 367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975839444743622e-05,
+      "loss": 2.1205,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756503529274047e-05,
+      "loss": 2.283,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975460533151513e-05,
+      "loss": 2.1516,
+      "step": 370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975269985557576e-05,
+      "loss": 2.0907,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750787102877673e-05,
+      "loss": 2.0709,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748867074848022e-05,
+      "loss": 2.0006,
+      "step": 373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746939772919393e-05,
+      "loss": 2.1337,
+      "step": 374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97450051985298e-05,
+      "loss": 1.9718,
+      "step": 375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974306335312268e-05,
+      "loss": 2.0384,
+      "step": 376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97411142381469e-05,
+      "loss": 2.2065,
+      "step": 377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739157855056746e-05,
+      "loss": 2.052,
+      "step": 378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737194205311935e-05,
+      "loss": 2.0993,
+      "step": 379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735223290377595e-05,
+      "loss": 2.1628,
+      "step": 380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733245111724282e-05,
+      "loss": 2.2113,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731259670827973e-05,
+      "loss": 2.071,
+      "step": 382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729266969170048e-05,
+      "loss": 2.0015,
+      "step": 383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727267008237334e-05,
+      "loss": 2.1402,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725259789522043e-05,
+      "loss": 2.2953,
+      "step": 385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723245314521827e-05,
+      "loss": 2.0347,
+      "step": 386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721223584739735e-05,
+      "loss": 2.1246,
+      "step": 387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719194601684236e-05,
+      "loss": 2.2483,
+      "step": 388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971715836686921e-05,
+      "loss": 2.0186,
+      "step": 389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971511488181395e-05,
+      "loss": 2.0507,
+      "step": 390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713064148043156e-05,
+      "loss": 2.0041,
+      "step": 391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971100616708694e-05,
+      "loss": 2.2381,
+      "step": 392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708940940480813e-05,
+      "loss": 1.9932,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706868469765697e-05,
+      "loss": 2.0435,
+      "step": 394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704788756487926e-05,
+      "loss": 2.1168,
+      "step": 395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970270180219923e-05,
+      "loss": 2.2486,
+      "step": 396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700607608456734e-05,
+      "loss": 2.074,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969850617682299e-05,
+      "loss": 2.003,
+      "step": 398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696397508865917e-05,
+      "loss": 2.2014,
+      "step": 399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694281606158864e-05,
+      "loss": 2.0637,
+      "step": 400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969215847028056e-05,
+      "loss": 2.2759,
+      "step": 401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690028102815132e-05,
+      "loss": 2.0274,
+      "step": 402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968789050535211e-05,
+      "loss": 1.9691,
+      "step": 403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685745679486408e-05,
+      "loss": 2.2409,
+      "step": 404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968359362681835e-05,
+      "loss": 2.1545,
+      "step": 405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681434348953638e-05,
+      "loss": 2.1214,
+      "step": 406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679267847503362e-05,
+      "loss": 2.0591,
+      "step": 407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967709412408402e-05,
+      "loss": 2.102,
+      "step": 408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674913180317478e-05,
+      "loss": 2.123,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672725017831003e-05,
+      "loss": 2.1348,
+      "step": 410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670529638257242e-05,
+      "loss": 2.1212,
+      "step": 411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668327043234225e-05,
+      "loss": 2.0202,
+      "step": 412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666117234405378e-05,
+      "loss": 2.1056,
+      "step": 413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663900213419492e-05,
+      "loss": 2.1624,
+      "step": 414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661675981930747e-05,
+      "loss": 2.1116,
+      "step": 415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965944454159871e-05,
+      "loss": 2.0525,
+      "step": 416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965720589408832e-05,
+      "loss": 2.0346,
+      "step": 417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965496004106989e-05,
+      "loss": 2.0764,
+      "step": 418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652706984219114e-05,
+      "loss": 2.0636,
+      "step": 419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650446725217056e-05,
+      "loss": 2.058,
+      "step": 420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648179265750165e-05,
+      "loss": 2.1546,
+      "step": 421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964590460751025e-05,
+      "loss": 1.9647,
+      "step": 422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643622752194496e-05,
+      "loss": 1.9524,
+      "step": 423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641333701505465e-05,
+      "loss": 2.0231,
+      "step": 424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639037457151072e-05,
+      "loss": 2.1501,
+      "step": 425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636734020844614e-05,
+      "loss": 2.0024,
+      "step": 426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963442339430475e-05,
+      "loss": 2.0111,
+      "step": 427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632105579255497e-05,
+      "loss": 1.993,
+      "step": 428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962978057742624e-05,
+      "loss": 2.0784,
+      "step": 429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627448390551736e-05,
+      "loss": 2.1103,
+      "step": 430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625109020372085e-05,
+      "loss": 2.0631,
+      "step": 431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962276246863276e-05,
+      "loss": 2.0654,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620408737084586e-05,
+      "loss": 2.1347,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618047827483745e-05,
+      "loss": 2.1274,
+      "step": 434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615679741591784e-05,
+      "loss": 1.9769,
+      "step": 435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613304481175594e-05,
+      "loss": 1.9048,
+      "step": 436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610922048007418e-05,
+      "loss": 2.0066,
+      "step": 437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960853244386486e-05,
+      "loss": 2.1428,
+      "step": 438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606135670530872e-05,
+      "loss": 2.0388,
+      "step": 439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603731729793746e-05,
+      "loss": 1.9821,
+      "step": 440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960132062344713e-05,
+      "loss": 2.1089,
+      "step": 441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598902353290024e-05,
+      "loss": 2.061,
+      "step": 442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596476921126757e-05,
+      "loss": 2.0722,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9594044328767015e-05,
+      "loss": 2.1043,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591604578025825e-05,
+      "loss": 2.2685,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958915767072355e-05,
+      "loss": 2.1525,
+      "step": 446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958670360868589e-05,
+      "loss": 2.2932,
+      "step": 447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95842423937439e-05,
+      "loss": 2.2267,
+      "step": 448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581774027733947e-05,
+      "loss": 1.9721,
+      "step": 449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579298512497758e-05,
+      "loss": 1.9634,
+      "step": 450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957681584988238e-05,
+      "loss": 2.1615,
+      "step": 451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957432604174019e-05,
+      "loss": 2.1338,
+      "step": 452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571829089928913e-05,
+      "loss": 2.1771,
+      "step": 453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569324996311586e-05,
+      "loss": 2.0738,
+      "step": 454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566813762756584e-05,
+      "loss": 2.0316,
+      "step": 455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956429539113761e-05,
+      "loss": 2.1739,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956176988333369e-05,
+      "loss": 2.1869,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559237241229175e-05,
+      "loss": 2.1589,
+      "step": 458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556697466713738e-05,
+      "loss": 2.2783,
+      "step": 459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554150561682374e-05,
+      "loss": 2.1501,
+      "step": 460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95515965280354e-05,
+      "loss": 1.9589,
+      "step": 461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549035367678453e-05,
+      "loss": 2.0964,
+      "step": 462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546467082522486e-05,
+      "loss": 2.2925,
+      "step": 463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543891674483767e-05,
+      "loss": 2.0813,
+      "step": 464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541309145483872e-05,
+      "loss": 2.1646,
+      "step": 465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953871949744971e-05,
+      "loss": 2.1302,
+      "step": 466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536122732313476e-05,
+      "loss": 2.0852,
+      "step": 467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533518852012692e-05,
+      "loss": 2.1051,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530907858490195e-05,
+      "loss": 2.0434,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528289753694108e-05,
+      "loss": 2.044,
+      "step": 470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525664539577877e-05,
+      "loss": 2.0711,
+      "step": 471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952303221810024e-05,
+      "loss": 2.2477,
+      "step": 472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520392791225255e-05,
+      "loss": 2.167,
+      "step": 473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951774626092226e-05,
+      "loss": 2.2673,
+      "step": 474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951509262916591e-05,
+      "loss": 2.024,
+      "step": 475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512431897936156e-05,
+      "loss": 2.1577,
+      "step": 476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950976406921824e-05,
+      "loss": 1.9125,
+      "step": 477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95070891450027e-05,
+      "loss": 2.0845,
+      "step": 478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504407127285377e-05,
+      "loss": 1.9416,
+      "step": 479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501718018067395e-05,
+      "loss": 1.9172,
+      "step": 480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499021819355168e-05,
+      "loss": 2.0909,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949631853316041e-05,
+      "loss": 2.2726,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949360816150012e-05,
+      "loss": 2.0058,
+      "step": 483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490890706396577e-05,
+      "loss": 2.0582,
+      "step": 484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948816616987735e-05,
+      "loss": 2.1043,
+      "step": 485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948543455397529e-05,
+      "loss": 1.9863,
+      "step": 486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482695860728534e-05,
+      "loss": 2.1207,
+      "step": 487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947995009218049e-05,
+      "loss": 2.2377,
+      "step": 488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477197250379862e-05,
+      "loss": 2.0327,
+      "step": 489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474437337380608e-05,
+      "loss": 2.1055,
+      "step": 490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947167035524199e-05,
+      "loss": 1.9996,
+      "step": 491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946889630602852e-05,
+      "loss": 2.0112,
+      "step": 492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466115191809995e-05,
+      "loss": 2.026,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9463327014661484e-05,
+      "loss": 2.1515,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946053177666332e-05,
+      "loss": 2.1027,
+      "step": 495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457729479901103e-05,
+      "loss": 2.0827,
+      "step": 496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454920126465716e-05,
+      "loss": 2.0898,
+      "step": 497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452103718453283e-05,
+      "loss": 2.0573,
+      "step": 498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944928025796521e-05,
+      "loss": 2.2742,
+      "step": 499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944644974710816e-05,
+      "loss": 2.0683,
+      "step": 500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944361218799405e-05,
+      "loss": 2.1139,
+      "step": 501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944076758274007e-05,
+      "loss": 2.0449,
+      "step": 502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437915933468648e-05,
+      "loss": 2.1037,
+      "step": 503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943505724230748e-05,
+      "loss": 1.959,
+      "step": 504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432191511389524e-05,
+      "loss": 2.2189,
+      "step": 505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942931874285297e-05,
+      "loss": 2.1891,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426438938841278e-05,
+      "loss": 2.0744,
+      "step": 507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423552101503143e-05,
+      "loss": 2.0531,
+      "step": 508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942065823299252e-05,
+      "loss": 2.0824,
+      "step": 509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417757335468596e-05,
+      "loss": 2.13,
+      "step": 510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941484941109582e-05,
+      "loss": 2.0355,
+      "step": 511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941193446204387e-05,
+      "loss": 2.073,
+      "step": 512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940901249048767e-05,
+      "loss": 2.0244,
+      "step": 513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406083498607385e-05,
+      "loss": 2.0352,
+      "step": 514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403147488588414e-05,
+      "loss": 2.0722,
+      "step": 515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400204462621398e-05,
+      "loss": 2.1696,
+      "step": 516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397254422902208e-05,
+      "loss": 2.2138,
+      "step": 517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394297371631955e-05,
+      "loss": 1.9755,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9391333311016968e-05,
+      "loss": 2.1022,
+      "step": 519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388362243268823e-05,
+      "loss": 2.1782,
+      "step": 520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938538417060431e-05,
+      "loss": 2.2433,
+      "step": 521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382399095245453e-05,
+      "loss": 2.1096,
+      "step": 522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93794070194195e-05,
+      "loss": 2.1201,
+      "step": 523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376407945358923e-05,
+      "loss": 2.2764,
+      "step": 524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373401875301407e-05,
+      "loss": 1.987,
+      "step": 525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370388811489873e-05,
+      "loss": 2.1447,
+      "step": 526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367368756172444e-05,
+      "loss": 1.8911,
+      "step": 527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936434171160247e-05,
+      "loss": 2.0658,
+      "step": 528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361307680038517e-05,
+      "loss": 2.208,
+      "step": 529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935826666374435e-05,
+      "loss": 2.29,
+      "step": 530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355218664988958e-05,
+      "loss": 2.0086,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352163686046546e-05,
+      "loss": 1.946,
+      "step": 532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349101729196505e-05,
+      "loss": 2.2165,
+      "step": 533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346032796723454e-05,
+      "loss": 2.087,
+      "step": 534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934295689091721e-05,
+      "loss": 2.1251,
+      "step": 535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339874014072785e-05,
+      "loss": 2.0044,
+      "step": 536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336784168490395e-05,
+      "loss": 2.0714,
+      "step": 537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933368735647547e-05,
+      "loss": 2.1978,
+      "step": 538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330583580338624e-05,
+      "loss": 2.1544,
+      "step": 539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327472842395666e-05,
+      "loss": 1.9759,
+      "step": 540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324355144967606e-05,
+      "loss": 2.1178,
+      "step": 541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321230490380644e-05,
+      "loss": 2.055,
+      "step": 542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318098880966173e-05,
+      "loss": 2.026,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314960319060768e-05,
+      "loss": 1.9408,
+      "step": 544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93118148070062e-05,
+      "loss": 2.1526,
+      "step": 545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308662347149423e-05,
+      "loss": 2.0109,
+      "step": 546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9305502941842574e-05,
+      "loss": 2.123,
+      "step": 547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930233659344297e-05,
+      "loss": 2.0683,
+      "step": 548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929916330431312e-05,
+      "loss": 2.0783,
+      "step": 549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295983076820687e-05,
+      "loss": 2.179,
+      "step": 550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292795913338543e-05,
+      "loss": 2.1003,
+      "step": 551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289601816244708e-05,
+      "loss": 2.0289,
+      "step": 552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928640078792239e-05,
+      "loss": 2.1366,
+      "step": 553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928319283075996e-05,
+      "loss": 2.0295,
+      "step": 554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927997794715097e-05,
+      "loss": 2.0243,
+      "step": 555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276756139494134e-05,
+      "loss": 2.2167,
+      "step": 556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273527410193325e-05,
+      "loss": 2.1146,
+      "step": 557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270291761657592e-05,
+      "loss": 2.0717,
+      "step": 558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267049196301137e-05,
+      "loss": 1.9456,
+      "step": 559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263799716543335e-05,
+      "loss": 2.0217,
+      "step": 560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260543324808706e-05,
+      "loss": 1.9718,
+      "step": 561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257280023526938e-05,
+      "loss": 2.0528,
+      "step": 562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254009815132867e-05,
+      "loss": 2.1248,
+      "step": 563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250732702066487e-05,
+      "loss": 1.9551,
+      "step": 564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247448686772944e-05,
+      "loss": 2.0266,
+      "step": 565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244157771702533e-05,
+      "loss": 2.1102,
+      "step": 566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924085995931069e-05,
+      "loss": 2.0152,
+      "step": 567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237555252058015e-05,
+      "loss": 2.1487,
+      "step": 568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923424365241023e-05,
+      "loss": 2.0398,
+      "step": 569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230925162838223e-05,
+      "loss": 2.1375,
+      "step": 570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227599785817997e-05,
+      "loss": 2.0095,
+      "step": 571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922426752383072e-05,
+      "loss": 2.1953,
+      "step": 572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220928379362673e-05,
+      "loss": 2.0316,
+      "step": 573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217582354905295e-05,
+      "loss": 2.0101,
+      "step": 574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921422945295514e-05,
+      "loss": 2.1603,
+      "step": 575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9210869676013906e-05,
+      "loss": 2.1103,
+      "step": 576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920750302658841e-05,
+      "loss": 2.1324,
+      "step": 577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204129507190604e-05,
+      "loss": 2.3168,
+      "step": 578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920074912033757e-05,
+      "loss": 2.0659,
+      "step": 579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197361868551498e-05,
+      "loss": 1.9578,
+      "step": 580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193967754359715e-05,
+      "loss": 1.9142,
+      "step": 581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919056678029466e-05,
+      "loss": 2.1447,
+      "step": 582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187158948893904e-05,
+      "loss": 2.0808,
+      "step": 583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183744262700114e-05,
+      "loss": 2.2585,
+      "step": 584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918032272426108e-05,
+      "loss": 2.1271,
+      "step": 585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176894336129717e-05,
+      "loss": 2.0481,
+      "step": 586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173459100864033e-05,
+      "loss": 2.0968,
+      "step": 587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170017021027153e-05,
+      "loss": 2.0043,
+      "step": 588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166568099187307e-05,
+      "loss": 2.0674,
+      "step": 589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916311233791783e-05,
+      "loss": 2.0377,
+      "step": 590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159649739797163e-05,
+      "loss": 2.0513,
+      "step": 591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156180307408846e-05,
+      "loss": 1.9481,
+      "step": 592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152704043341522e-05,
+      "loss": 2.0555,
+      "step": 593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9149220950188917e-05,
+      "loss": 2.0321,
+      "step": 594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145731030549873e-05,
+      "loss": 2.0564,
+      "step": 595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142234287028313e-05,
+      "loss": 2.0679,
+      "step": 596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913873072223325e-05,
+      "loss": 2.0356,
+      "step": 597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135220338778797e-05,
+      "loss": 2.0611,
+      "step": 598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131703139284143e-05,
+      "loss": 2.0958,
+      "step": 599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912817912637357e-05,
+      "loss": 2.0945,
+      "step": 600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124648302676437e-05,
+      "loss": 2.1156,
+      "step": 601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121110670827193e-05,
+      "loss": 2.1326,
+      "step": 602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117566233465362e-05,
+      "loss": 2.0849,
+      "step": 603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114014993235552e-05,
+      "loss": 2.0759,
+      "step": 604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110456952787432e-05,
+      "loss": 1.9964,
+      "step": 605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106892114775763e-05,
+      "loss": 2.0984,
+      "step": 606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103320481860362e-05,
+      "loss": 2.1734,
+      "step": 607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099742056706123e-05,
+      "loss": 2.1178,
+      "step": 608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096156841983013e-05,
+      "loss": 2.0401,
+      "step": 609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909256484036606e-05,
+      "loss": 2.1282,
+      "step": 610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908896605453535e-05,
+      "loss": 2.0475,
+      "step": 611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085360487176037e-05,
+      "loss": 1.8798,
+      "step": 612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908174814097834e-05,
+      "loss": 2.149,
+      "step": 613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907812901863753e-05,
+      "loss": 1.9444,
+      "step": 614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074503122853923e-05,
+      "loss": 2.2001,
+      "step": 615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070870456332914e-05,
+      "loss": 2.1119,
+      "step": 616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906723102178493e-05,
+      "loss": 2.0627,
+      "step": 617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063584821925452e-05,
+      "loss": 2.0413,
+      "step": 618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9059931859475013e-05,
+      "loss": 2.0902,
+      "step": 619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056272137159187e-05,
+      "loss": 2.1869,
+      "step": 620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052605657708596e-05,
+      "loss": 2.0562,
+      "step": 621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048932423858903e-05,
+      "loss": 1.9537,
+      "step": 622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045252438350803e-05,
+      "loss": 2.0217,
+      "step": 623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904156570393004e-05,
+      "loss": 2.0087,
+      "step": 624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9037872223347385e-05,
+      "loss": 1.9944,
+      "step": 625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9034171999358655e-05,
+      "loss": 2.1688,
+      "step": 626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9030465034724676e-05,
+      "loss": 2.1464,
+      "step": 627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026751332211324e-05,
+      "loss": 2.0621,
+      "step": 628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023030894589496e-05,
+      "loss": 2.0693,
+      "step": 629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901930372463511e-05,
+      "loss": 2.1817,
+      "step": 630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015569825129112e-05,
+      "loss": 2.0721,
+      "step": 631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011829198857467e-05,
+      "loss": 2.0774,
+      "step": 632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900808184861116e-05,
+      "loss": 2.078,
+      "step": 633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004327777186195e-05,
+      "loss": 2.04,
+      "step": 634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000566987383582e-05,
+      "loss": 2.0963,
+      "step": 635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996799482009355e-05,
+      "loss": 2.0912,
+      "step": 636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993025263874552e-05,
+      "loss": 2.1518,
+      "step": 637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989244335795222e-05,
+      "loss": 2.1706,
+      "step": 638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898545670059242e-05,
+      "loss": 1.9684,
+      "step": 639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981662361092206e-05,
+      "loss": 2.215,
+      "step": 640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897786132012564e-05,
+      "loss": 2.028,
+      "step": 641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974053580528786e-05,
+      "loss": 2.1806,
+      "step": 642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89702391451427e-05,
+      "loss": 2.1355,
+      "step": 643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8966418016813443e-05,
+      "loss": 2.1247,
+      "step": 644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962590198392057e-05,
+      "loss": 1.9697,
+      "step": 645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895875569273459e-05,
+      "loss": 2.0094,
+      "step": 646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895491450270207e-05,
+      "loss": 1.9919,
+      "step": 647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895106663116051e-05,
+      "loss": 1.9849,
+      "step": 648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894721208098092e-05,
+      "loss": 1.9958,
+      "step": 649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943350855039288e-05,
+      "loss": 2.1585,
+      "step": 650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939482956216573e-05,
+      "loss": 2.1583,
+      "step": 651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893560838739873e-05,
+      "loss": 2.1848,
+      "step": 652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893172715147667e-05,
+      "loss": 2.2405,
+      "step": 653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927839251346302e-05,
+      "loss": 2.1478,
+      "step": 654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8923944689908492e-05,
+      "loss": 2.1029,
+      "step": 655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892004347006908e-05,
+      "loss": 2.0691,
+      "step": 656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891613559473887e-05,
+      "loss": 2.0549,
+      "step": 657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891222106683364e-05,
+      "loss": 2.039,
+      "step": 658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890829988927413e-05,
+      "loss": 1.9734,
+      "step": 659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904372064986033e-05,
+      "loss": 1.9615,
+      "step": 660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890043759690001e-05,
+      "loss": 1.9322,
+      "step": 661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889649648795167e-05,
+      "loss": 2.0899,
+      "step": 662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889254874108159e-05,
+      "loss": 2.0542,
+      "step": 663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888594359235297e-05,
+      "loss": 2.141,
+      "step": 664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884633345363257e-05,
+      "loss": 2.045,
+      "step": 665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880665702420894e-05,
+      "loss": 2.094,
+      "step": 666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887669143336858e-05,
+      "loss": 1.9949,
+      "step": 667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872710541171614e-05,
+      "loss": 2.2024,
+      "step": 668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8868723028800263e-05,
+      "loss": 1.983,
+      "step": 669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886472889922972e-05,
+      "loss": 2.1915,
+      "step": 670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860728155440108e-05,
+      "loss": 1.9994,
+      "step": 671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856720800416496e-05,
+      "loss": 2.1001,
+      "step": 672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885270683714888e-05,
+      "loss": 2.132,
+      "step": 673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848686268632193e-05,
+      "loss": 2.1833,
+      "step": 674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884465909786629e-05,
+      "loss": 2.2095,
+      "step": 675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884062532785595e-05,
+      "loss": 2.0305,
+      "step": 676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883658496161089e-05,
+      "loss": 2.1877,
+      "step": 677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832538002145728e-05,
+      "loss": 2.0565,
+      "step": 678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8828484452480024e-05,
+      "loss": 2.0808,
+      "step": 679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824424315638233e-05,
+      "loss": 2.066,
+      "step": 680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882035759464974e-05,
+      "loss": 2.237,
+      "step": 681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816284292548833e-05,
+      "loss": 2.0739,
+      "step": 682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812204412374724e-05,
+      "loss": 2.1036,
+      "step": 683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880811795717152e-05,
+      "loss": 1.992,
+      "step": 684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8804024929988234e-05,
+      "loss": 1.9937,
+      "step": 685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799925333878793e-05,
+      "loss": 2.1947,
+      "step": 686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795819171902015e-05,
+      "loss": 2.1933,
+      "step": 687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791706447121623e-05,
+      "loss": 2.1005,
+      "step": 688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878758716260623e-05,
+      "loss": 2.1701,
+      "step": 689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783461321429356e-05,
+      "loss": 2.1513,
+      "step": 690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779328926669397e-05,
+      "loss": 1.9733,
+      "step": 691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775189981409655e-05,
+      "loss": 1.9488,
+      "step": 692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87710444887383e-05,
+      "loss": 2.1372,
+      "step": 693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876689245174841e-05,
+      "loss": 2.0654,
+      "step": 694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876273387353793e-05,
+      "loss": 2.0267,
+      "step": 695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758568757209687e-05,
+      "loss": 1.9816,
+      "step": 696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875439710587139e-05,
+      "loss": 2.0952,
+      "step": 697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750218922635633e-05,
+      "loss": 2.0556,
+      "step": 698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874603421061986e-05,
+      "loss": 1.9524,
+      "step": 699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874184297294641e-05,
+      "loss": 2.1978,
+      "step": 700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737645212742474e-05,
+      "loss": 2.1391,
+      "step": 701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733440933140127e-05,
+      "loss": 1.9921,
+      "step": 702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729230137276287e-05,
+      "loss": 2.1958,
+      "step": 703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872501282829275e-05,
+      "loss": 2.2318,
+      "step": 704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720789009336165e-05,
+      "loss": 2.18,
+      "step": 705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716558683558046e-05,
+      "loss": 2.1911,
+      "step": 706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712321854114747e-05,
+      "loss": 2.1474,
+      "step": 707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870807852416749e-05,
+      "loss": 1.9531,
+      "step": 708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870382869688234e-05,
+      "loss": 2.1122,
+      "step": 709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699572375430206e-05,
+      "loss": 2.0319,
+      "step": 710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869530956298685e-05,
+      "loss": 2.0601,
+      "step": 711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691040262732877e-05,
+      "loss": 2.1528,
+      "step": 712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686764477853726e-05,
+      "loss": 2.105,
+      "step": 713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8682482211539675e-05,
+      "loss": 2.0976,
+      "step": 714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678193466985853e-05,
+      "loss": 1.9635,
+      "step": 715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673898247392197e-05,
+      "loss": 2.0271,
+      "step": 716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86695965559635e-05,
+      "loss": 2.1241,
+      "step": 717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665288395909362e-05,
+      "loss": 2.0422,
+      "step": 718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8660973770444227e-05,
+      "loss": 2.1433,
+      "step": 719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656652682787356e-05,
+      "loss": 2.172,
+      "step": 720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652325136162836e-05,
+      "loss": 2.061,
+      "step": 721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647991133799558e-05,
+      "loss": 2.0486,
+      "step": 722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643650678931248e-05,
+      "loss": 2.079,
+      "step": 723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863930377479644e-05,
+      "loss": 2.1625,
+      "step": 724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8634950424638477e-05,
+      "loss": 2.0032,
+      "step": 725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630590631705514e-05,
+      "loss": 2.1242,
+      "step": 726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626224399250515e-05,
+      "loss": 2.0079,
+      "step": 727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862185173053124e-05,
+      "loss": 2.1512,
+      "step": 728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861747262881027e-05,
+      "loss": 2.1447,
+      "step": 729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861308709735496e-05,
+      "loss": 2.1407,
+      "step": 730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608695139437486e-05,
+      "loss": 2.1484,
+      "step": 731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8604296758334804e-05,
+      "loss": 1.9151,
+      "step": 732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859989195732867e-05,
+      "loss": 2.2456,
+      "step": 733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859548073970563e-05,
+      "loss": 2.0356,
+      "step": 734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859106310875701e-05,
+      "loss": 2.1163,
+      "step": 735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8586639067778925e-05,
+      "loss": 2.0882,
+      "step": 736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8582208620072283e-05,
+      "loss": 1.9685,
+      "step": 737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577771768942755e-05,
+      "loss": 2.1537,
+      "step": 738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8573328517700805e-05,
+      "loss": 2.2408,
+      "step": 739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856887886966166e-05,
+      "loss": 2.1632,
+      "step": 740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564422828145327e-05,
+      "loss": 2.1166,
+      "step": 741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855996039647658e-05,
+      "loss": 1.992,
+      "step": 742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555491577984967e-05,
+      "loss": 1.9231,
+      "step": 743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8551016376004796e-05,
+      "loss": 2.0601,
+      "step": 744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546534793875134e-05,
+      "loss": 2.1245,
+      "step": 745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542046834939816e-05,
+      "loss": 2.079,
+      "step": 746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537552502547435e-05,
+      "loss": 2.038,
+      "step": 747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533051800051333e-05,
+      "loss": 2.0449,
+      "step": 748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852854473080961e-05,
+      "loss": 2.0672,
+      "step": 749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524031298185114e-05,
+      "loss": 1.9162,
+      "step": 750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851951150554544e-05,
+      "loss": 1.9797,
+      "step": 751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514985356262932e-05,
+      "loss": 2.0481,
+      "step": 752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851045285371468e-05,
+      "loss": 2.0876,
+      "step": 753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85059140012825e-05,
+      "loss": 1.9302,
+      "step": 754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850136880235296e-05,
+      "loss": 2.0301,
+      "step": 755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849681726031736e-05,
+      "loss": 2.0945,
+      "step": 756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8492259378571727e-05,
+      "loss": 2.0323,
+      "step": 757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8487695160516825e-05,
+      "loss": 1.9973,
+      "step": 758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483124609558145e-05,
+      "loss": 2.0284,
+      "step": 759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478547729105898e-05,
+      "loss": 2.0892,
+      "step": 760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847396452257502e-05,
+      "loss": 2.037,
+      "step": 761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469374993385175e-05,
+      "loss": 1.9884,
+      "step": 762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464779144960726e-05,
+      "loss": 1.9746,
+      "step": 763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8460176980730777e-05,
+      "loss": 2.1478,
+      "step": 764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455568504129117e-05,
+      "loss": 2.0941,
+      "step": 765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8450953718594263e-05,
+      "loss": 2.2028,
+      "step": 766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844633262756943e-05,
+      "loss": 2.0477,
+      "step": 767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844170523450255e-05,
+      "loss": 2.0261,
+      "step": 768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8437071542846242e-05,
+      "loss": 2.2724,
+      "step": 769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432431556057832e-05,
+      "loss": 2.2646,
+      "step": 770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427785277599347e-05,
+      "loss": 2.0803,
+      "step": 771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423132710937498e-05,
+      "loss": 2.0853,
+      "step": 772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418473859543694e-05,
+      "loss": 2.101,
+      "step": 773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413808726894038e-05,
+      "loss": 2.1327,
+      "step": 774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8409137316469308e-05,
+      "loss": 2.1332,
+      "step": 775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404459631754978e-05,
+      "loss": 2.0031,
+      "step": 776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399775676241192e-05,
+      "loss": 2.0672,
+      "step": 777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839508545342278e-05,
+      "loss": 2.0958,
+      "step": 778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390388966799252e-05,
+      "loss": 2.0105,
+      "step": 779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838568621987478e-05,
+      "loss": 1.9901,
+      "step": 780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380977216158217e-05,
+      "loss": 2.1736,
+      "step": 781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376261959163076e-05,
+      "loss": 1.9927,
+      "step": 782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837154045240755e-05,
+      "loss": 2.1751,
+      "step": 783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366812699414476e-05,
+      "loss": 1.9944,
+      "step": 784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362078703711366e-05,
+      "loss": 1.9317,
+      "step": 785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835733846883038e-05,
+      "loss": 1.9408,
+      "step": 786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8352591998308346e-05,
+      "loss": 1.9305,
+      "step": 787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834783929568673e-05,
+      "loss": 2.0034,
+      "step": 788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834308036451166e-05,
+      "loss": 1.898,
+      "step": 789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338315208333904e-05,
+      "loss": 2.2036,
+      "step": 790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833354383070887e-05,
+      "loss": 2.2859,
+      "step": 791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328766235196628e-05,
+      "loss": 2.0754,
+      "step": 792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8323982425361864e-05,
+      "loss": 2.1682,
+      "step": 793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8319192404773912e-05,
+      "loss": 1.899,
+      "step": 794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831439617700674e-05,
+      "loss": 2.142,
+      "step": 795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309593745638945e-05,
+      "loss": 2.0643,
+      "step": 796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304785114253757e-05,
+      "loss": 2.0347,
+      "step": 797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299970286439023e-05,
+      "loss": 2.0681,
+      "step": 798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295149265787224e-05,
+      "loss": 2.2443,
+      "step": 799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290322055895454e-05,
+      "loss": 1.9751,
+      "step": 800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828548866036543e-05,
+      "loss": 2.1085,
+      "step": 801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280649082803478e-05,
+      "loss": 2.0898,
+      "step": 802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275803326820545e-05,
+      "loss": 2.0397,
+      "step": 803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827095139603218e-05,
+      "loss": 2.0453,
+      "step": 804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266093294058542e-05,
+      "loss": 2.1116,
+      "step": 805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82612290245244e-05,
+      "loss": 2.0554,
+      "step": 806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8256358591059115e-05,
+      "loss": 2.0964,
+      "step": 807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251481997296654e-05,
+      "loss": 2.177,
+      "step": 808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824659924687558e-05,
+      "loss": 2.2019,
+      "step": 809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8241710343439042e-05,
+      "loss": 2.0446,
+      "step": 810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8236815290634794e-05,
+      "loss": 2.2175,
+      "step": 811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231914092115164e-05,
+      "loss": 2.068,
+      "step": 812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822700675153707e-05,
+      "loss": 2.2024,
+      "step": 813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8222093272562023e-05,
+      "loss": 2.1152,
+      "step": 814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82171736588561e-05,
+      "loss": 2.1658,
+      "step": 815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8212247914089954e-05,
+      "loss": 2.0257,
+      "step": 816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207316041938835e-05,
+      "loss": 2.1251,
+      "step": 817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202378046082533e-05,
+      "loss": 2.0856,
+      "step": 818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8197433930205433e-05,
+      "loss": 2.0725,
+      "step": 819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192483697996473e-05,
+      "loss": 2.1683,
+      "step": 820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818752735314916e-05,
+      "loss": 2.0949,
+      "step": 821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182564899361558e-05,
+      "loss": 1.9965,
+      "step": 822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817759634033629e-05,
+      "loss": 2.0271,
+      "step": 823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8172621679780532e-05,
+      "loss": 1.9172,
+      "step": 824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167640921406026e-05,
+      "loss": 2.1983,
+      "step": 825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162654068929043e-05,
+      "loss": 2.3338,
+      "step": 826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815766112607042e-05,
+      "loss": 2.0754,
+      "step": 827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815266209655552e-05,
+      "loss": 2.2493,
+      "step": 828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8147656984114268e-05,
+      "loss": 2.1235,
+      "step": 829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814264579248111e-05,
+      "loss": 2.0683,
+      "step": 830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137628525395032e-05,
+      "loss": 2.1036,
+      "step": 831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813260518659956e-05,
+      "loss": 2.0402,
+      "step": 832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127575779842744e-05,
+      "loss": 2.0176,
+      "step": 833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8122540308877165e-05,
+      "loss": 2.088,
+      "step": 834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8117498777459926e-05,
+      "loss": 2.1039,
+      "step": 835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112451189352653e-05,
+      "loss": 1.9656,
+      "step": 836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107397548321487e-05,
+      "loss": 2.0934,
+      "step": 837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102337858137094e-05,
+      "loss": 2.0962,
+      "step": 838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8097272122574653e-05,
+      "loss": 2.1393,
+      "step": 839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809220034541384e-05,
+      "loss": 2.0145,
+      "step": 840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087122530438847e-05,
+      "loss": 2.1078,
+      "step": 841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808203868143838e-05,
+      "loss": 2.0474,
+      "step": 842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8076948802205637e-05,
+      "loss": 2.0055,
+      "step": 843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8071852896538314e-05,
+      "loss": 2.0778,
+      "step": 844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806675096823861e-05,
+      "loss": 2.0673,
+      "step": 845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061643021113215e-05,
+      "loss": 1.96,
+      "step": 846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805652905897331e-05,
+      "loss": 2.0562,
+      "step": 847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8051409085634556e-05,
+      "loss": 2.0278,
+      "step": 848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046283104917116e-05,
+      "loss": 2.1826,
+      "step": 849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8041151120645622e-05,
+      "loss": 2.1399,
+      "step": 850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036013136649186e-05,
+      "loss": 2.2628,
+      "step": 851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030869156761403e-05,
+      "loss": 1.994,
+      "step": 852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802571918482034e-05,
+      "loss": 2.1652,
+      "step": 853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020563224668534e-05,
+      "loss": 2.0373,
+      "step": 854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015401280152986e-05,
+      "loss": 2.0345,
+      "step": 855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8010233355125163e-05,
+      "loss": 2.0333,
+      "step": 856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005059453441002e-05,
+      "loss": 2.0996,
+      "step": 857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799987957896089e-05,
+      "loss": 2.1884,
+      "step": 858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799469373554968e-05,
+      "loss": 2.0926,
+      "step": 859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7989501927076663e-05,
+      "loss": 2.0505,
+      "step": 860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984304157415602e-05,
+      "loss": 1.956,
+      "step": 861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797910043044469e-05,
+      "loss": 1.9716,
+      "step": 862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973890750046574e-05,
+      "loss": 2.0674,
+      "step": 863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7968675120108338e-05,
+      "loss": 2.0069,
+      "step": 864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796345354452151e-05,
+      "loss": 2.0547,
+      "step": 865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795822602718205e-05,
+      "loss": 2.1247,
+      "step": 866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952992571990352e-05,
+      "loss": 2.1307,
+      "step": 867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7947753182851248e-05,
+      "loss": 2.1847,
+      "step": 868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794250786367398e-05,
+      "loss": 1.9865,
+      "step": 869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793725661837223e-05,
+      "loss": 2.0754,
+      "step": 870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931999450864106e-05,
+      "loss": 2.0803,
+      "step": 871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7926736365072116e-05,
+      "loss": 1.9959,
+      "step": 872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921467364923194e-05,
+      "loss": 2.1058,
+      "step": 873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7916192454348687e-05,
+      "loss": 2.1385,
+      "step": 874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791091163728436e-05,
+      "loss": 1.9837,
+      "step": 875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905624917670365e-05,
+      "loss": 2.0095,
+      "step": 876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900332299451276e-05,
+      "loss": 2.1201,
+      "step": 877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7895033786576056e-05,
+      "loss": 2.0723,
+      "step": 878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788972938299808e-05,
+      "loss": 2.0061,
+      "step": 879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7884419092675104e-05,
+      "loss": 1.993,
+      "step": 880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879102919569283e-05,
+      "loss": 2.0725,
+      "step": 881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873780867647165e-05,
+      "loss": 2.148,
+      "step": 882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868452940879673e-05,
+      "loss": 1.9938,
+      "step": 883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7863119143242125e-05,
+      "loss": 2.1881,
+      "step": 884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857779478714215e-05,
+      "loss": 1.9317,
+      "step": 885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7852433951280012e-05,
+      "loss": 2.0024,
+      "step": 886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847082564927958e-05,
+      "loss": 1.9247,
+      "step": 887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7841725323650876e-05,
+      "loss": 2.0414,
+      "step": 888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7836362231445953e-05,
+      "loss": 1.9478,
+      "step": 889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7830993292314736e-05,
+      "loss": 2.126,
+      "step": 890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825618510263144e-05,
+      "loss": 2.0988,
+      "step": 891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782023788930144e-05,
+      "loss": 1.9235,
+      "step": 892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7814851433444263e-05,
+      "loss": 1.9372,
+      "step": 893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7809459146710596e-05,
+      "loss": 1.9239,
+      "step": 894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7804061033123767e-05,
+      "loss": 1.8896,
+      "step": 895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798657096711466e-05,
+      "loss": 2.1105,
+      "step": 896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779324734150571e-05,
+      "loss": 1.8998,
+      "step": 897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778783177154287e-05,
+      "loss": 2.1331,
+      "step": 898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7782410390863664e-05,
+      "loss": 1.9915,
+      "step": 899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7776983203513113e-05,
+      "loss": 2.0835,
+      "step": 900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777155021354061e-05,
+      "loss": 1.9979,
+      "step": 901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7766111424999844e-05,
+      "loss": 2.1237,
+      "step": 902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760666841948857e-05,
+      "loss": 2.1668,
+      "step": 903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7755216468449995e-05,
+      "loss": 1.8196,
+      "step": 904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774976030856994e-05,
+      "loss": 1.9662,
+      "step": 905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7744298366379673e-05,
+      "loss": 1.9102,
+      "step": 906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773883064595451e-05,
+      "loss": 2.1534,
+      "step": 907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733357151374062e-05,
+      "loss": 2.1858,
+      "step": 908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772787788672226e-05,
+      "loss": 2.2984,
+      "step": 909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722392856087326e-05,
+      "loss": 2.0689,
+      "step": 910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716902063561798e-05,
+      "loss": 2.134,
+      "step": 911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711405513242513e-05,
+      "loss": 2.1491,
+      "step": 912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770590320923059e-05,
+      "loss": 2.1084,
+      "step": 913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7700395155631456e-05,
+      "loss": 2.07,
+      "step": 914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769488135655482e-05,
+      "loss": 2.1265,
+      "step": 915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768936181611468e-05,
+      "loss": 2.0257,
+      "step": 916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683836538429314e-05,
+      "loss": 2.2208,
+      "step": 917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678305527621292e-05,
+      "loss": 2.036,
+      "step": 918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7672768787817444e-05,
+      "loss": 2.1872,
+      "step": 919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7667226323148894e-05,
+      "loss": 2.0627,
+      "step": 920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766167813775102e-05,
+      "loss": 2.0086,
+      "step": 921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656124235763487e-05,
+      "loss": 2.0519,
+      "step": 922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650564621330202e-05,
+      "loss": 2.0584,
+      "step": 923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644999298599353e-05,
+      "loss": 2.1629,
+      "step": 924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639428271723385e-05,
+      "loss": 2.178,
+      "step": 925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633851544858988e-05,
+      "loss": 2.0371,
+      "step": 926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7628269122167115e-05,
+      "loss": 2.0742,
+      "step": 927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622681007812965e-05,
+      "loss": 2.0686,
+      "step": 928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7617087205965987e-05,
+      "loss": 1.9865,
+      "step": 929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7611487720799868e-05,
+      "loss": 2.2099,
+      "step": 930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7605882556492536e-05,
+      "loss": 2.0271,
+      "step": 931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600271717226167e-05,
+      "loss": 2.025,
+      "step": 932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594655207187155e-05,
+      "loss": 2.0317,
+      "step": 933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758903303056614e-05,
+      "loss": 1.9743,
+      "step": 934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758340519155798e-05,
+      "loss": 2.0186,
+      "step": 935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757777169436176e-05,
+      "loss": 2.0269,
+      "step": 936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572132543180787e-05,
+      "loss": 2.0942,
+      "step": 937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566487742222594e-05,
+      "loss": 1.949,
+      "step": 938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560837295698915e-05,
+      "loss": 2.0108,
+      "step": 939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555181207825712e-05,
+      "loss": 1.9715,
+      "step": 940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754951948282314e-05,
+      "loss": 2.1552,
+      "step": 941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754385212491557e-05,
+      "loss": 2.1164,
+      "step": 942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753817913833158e-05,
+      "loss": 2.1771,
+      "step": 943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.753250052730394e-05,
+      "loss": 2.0059,
+      "step": 944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526816296069616e-05,
+      "loss": 2.1509,
+      "step": 945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752112644886977e-05,
+      "loss": 2.1531,
+      "step": 946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515430989949754e-05,
+      "loss": 2.0267,
+      "step": 947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7509729923559113e-05,
+      "loss": 2.2791,
+      "step": 948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750402325395156e-05,
+      "loss": 2.0161,
+      "step": 949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749831098538501e-05,
+      "loss": 2.142,
+      "step": 950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492593122121534e-05,
+      "loss": 2.0069,
+      "step": 951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7486869668427396e-05,
+      "loss": 2.1469,
+      "step": 952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481140628573017e-05,
+      "loss": 2.0611,
+      "step": 953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7475406006832996e-05,
+      "loss": 1.9623,
+      "step": 954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7469665807486088e-05,
+      "loss": 1.9778,
+      "step": 955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7463920034815214e-05,
+      "loss": 2.1141,
+      "step": 956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7458168693107465e-05,
+      "loss": 2.232,
+      "step": 957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7452411786654063e-05,
+      "loss": 1.8763,
+      "step": 958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74466493197504e-05,
+      "loss": 1.9351,
+      "step": 959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744088129669601e-05,
+      "loss": 2.0438,
+      "step": 960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435107721794577e-05,
+      "loss": 1.9983,
+      "step": 961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429328599353926e-05,
+      "loss": 2.033,
+      "step": 962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742354393368601e-05,
+      "loss": 2.1257,
+      "step": 963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741775372910694e-05,
+      "loss": 2.054,
+      "step": 964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411957989936944e-05,
+      "loss": 2.0819,
+      "step": 965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406156720500376e-05,
+      "loss": 2.0178,
+      "step": 966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7400349925125733e-05,
+      "loss": 2.1249,
+      "step": 967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739453760814562e-05,
+      "loss": 2.015,
+      "step": 968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738871977389677e-05,
+      "loss": 2.0478,
+      "step": 969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7382896426720025e-05,
+      "loss": 1.927,
+      "step": 970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377067570960352e-05,
+      "loss": 2.1537,
+      "step": 971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371233210966814e-05,
+      "loss": 2.0347,
+      "step": 972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365393351092598e-05,
+      "loss": 2.0972,
+      "step": 973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359547995694975e-05,
+      "loss": 2.1136,
+      "step": 974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353697149135327e-05,
+      "loss": 2.0218,
+      "step": 975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734784081577914e-05,
+      "loss": 2.0246,
+      "step": 976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7341978999995975e-05,
+      "loss": 1.9411,
+      "step": 977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336111706159508e-05,
+      "loss": 2.0379,
+      "step": 978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330238938647475e-05,
+      "loss": 1.9772,
+      "step": 979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732436070184172e-05,
+      "loss": 2.0398,
+      "step": 980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7318477000128153e-05,
+      "loss": 2.1369,
+      "step": 981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7312587837896767e-05,
+      "loss": 1.9949,
+      "step": 982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306693219541632e-05,
+      "loss": 2.1325,
+      "step": 983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300793149460884e-05,
+      "loss": 2.1197,
+      "step": 984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7294887632056724e-05,
+      "loss": 1.9793,
+      "step": 985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7288976671735428e-05,
+      "loss": 1.9091,
+      "step": 986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7283060272907322e-05,
+      "loss": 2.0585,
+      "step": 987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277138439986797e-05,
+      "loss": 2.0667,
+      "step": 988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271211177392296e-05,
+      "loss": 2.1258,
+      "step": 989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726527848954631e-05,
+      "loss": 2.0552,
+      "step": 990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259340380875384e-05,
+      "loss": 2.2264,
+      "step": 991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7253396855810108e-05,
+      "loss": 2.0313,
+      "step": 992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7247447918785104e-05,
+      "loss": 2.1198,
+      "step": 993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7241493574239043e-05,
+      "loss": 2.1439,
+      "step": 994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723553382661462e-05,
+      "loss": 2.1,
+      "step": 995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722956868035858e-05,
+      "loss": 2.0662,
+      "step": 996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223598139921667e-05,
+      "loss": 2.135,
+      "step": 997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7217622209758676e-05,
+      "loss": 2.0709,
+      "step": 998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7211640894328413e-05,
+      "loss": 2.1558,
+      "step": 999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72056541980937e-05,
+      "loss": 1.9143,
+      "step": 1000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199662125521375e-05,
+      "loss": 2.0896,
+      "step": 1001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193664681082296e-05,
+      "loss": 2.0557,
+      "step": 1002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187661869251314e-05,
+      "loss": 2.009,
+      "step": 1003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181653694507297e-05,
+      "loss": 2.0489,
+      "step": 1004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717564016133311e-05,
+      "loss": 2.1122,
+      "step": 1005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7169621274215614e-05,
+      "loss": 2.047,
+      "step": 1006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7163597037645666e-05,
+      "loss": 1.9579,
+      "step": 1007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7157567456118124e-05,
+      "loss": 2.0689,
+      "step": 1008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715153253413181e-05,
+      "loss": 1.9044,
+      "step": 1009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7145492276189565e-05,
+      "loss": 2.0589,
+      "step": 1010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7139446686798176e-05,
+      "loss": 1.98,
+      "step": 1011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713339577046843e-05,
+      "loss": 1.9663,
+      "step": 1012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127339531715085e-05,
+      "loss": 2.1015,
+      "step": 1013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121277975056863e-05,
+      "loss": 2.1151,
+      "step": 1014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7115211105016465e-05,
+      "loss": 2.0571,
+      "step": 1015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109138926120548e-05,
+      "loss": 1.9339,
+      "step": 1016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7103061442899727e-05,
+      "loss": 2.0227,
+      "step": 1017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709697865988859e-05,
+      "loss": 2.0826,
+      "step": 1018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709089058162566e-05,
+      "loss": 2.139,
+      "step": 1019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7084797212653427e-05,
+      "loss": 2.0555,
+      "step": 1020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707869855751832e-05,
+      "loss": 1.9024,
+      "step": 1021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707259462077071e-05,
+      "loss": 1.9507,
+      "step": 1022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066485406964917e-05,
+      "loss": 2.1505,
+      "step": 1023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060370920659194e-05,
+      "loss": 1.9748,
+      "step": 1024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7054251166415726e-05,
+      "loss": 2.1809,
+      "step": 1025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048126148800634e-05,
+      "loss": 1.9567,
+      "step": 1026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704199587238396e-05,
+      "loss": 2.0266,
+      "step": 1027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7035860341739676e-05,
+      "loss": 1.972,
+      "step": 1028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029719561445665e-05,
+      "loss": 2.093,
+      "step": 1029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7023573536083734e-05,
+      "loss": 2.0699,
+      "step": 1030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017422270239608e-05,
+      "loss": 2.0017,
+      "step": 1031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011265768502912e-05,
+      "loss": 2.0849,
+      "step": 1032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7005104035467182e-05,
+      "loss": 1.9899,
+      "step": 1033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699893707572986e-05,
+      "loss": 2.0793,
+      "step": 1034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6992764893892283e-05,
+      "loss": 2.0402,
+      "step": 1035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986587494559684e-05,
+      "loss": 1.8528,
+      "step": 1036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6980404882341192e-05,
+      "loss": 2.0223,
+      "step": 1037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974217061849828e-05,
+      "loss": 2.036,
+      "step": 1038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6968024037702493e-05,
+      "loss": 2.0401,
+      "step": 1039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961825814519976e-05,
+      "loss": 2.181,
+      "step": 1040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695562239692694e-05,
+      "loss": 2.1262,
+      "step": 1041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6949413789551926e-05,
+      "loss": 2.0758,
+      "step": 1042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694319999702735e-05,
+      "loss": 2.1068,
+      "step": 1043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693698102398949e-05,
+      "loss": 2.1767,
+      "step": 1044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6930756875078496e-05,
+      "loss": 2.0053,
+      "step": 1045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6924527554938383e-05,
+      "loss": 1.9366,
+      "step": 1046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691829306821701e-05,
+      "loss": 2.1567,
+      "step": 1047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69120534195661e-05,
+      "loss": 2.1365,
+      "step": 1048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6905808613641233e-05,
+      "loss": 2.1564,
+      "step": 1049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6899558655101832e-05,
+      "loss": 1.9482,
+      "step": 1050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893303548611152e-05,
+      "loss": 2.078,
+      "step": 1051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6887043298836318e-05,
+      "loss": 1.9897,
+      "step": 1052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688077791044826e-05,
+      "loss": 2.0035,
+      "step": 1053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6874507388121767e-05,
+      "loss": 1.9257,
+      "step": 1054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686823173653544e-05,
+      "loss": 1.9665,
+      "step": 1055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861950960371728e-05,
+      "loss": 1.9981,
+      "step": 1056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855665064316878e-05,
+      "loss": 1.9285,
+      "step": 1057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6849374053060984e-05,
+      "loss": 2.1091,
+      "step": 1058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843077931297935e-05,
+      "loss": 2.0266,
+      "step": 1059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683677670372544e-05,
+      "loss": 2.2318,
+      "step": 1060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6830470375045026e-05,
+      "loss": 2.0764,
+      "step": 1061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682415894996201e-05,
+      "loss": 1.9991,
+      "step": 1062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681784243318553e-05,
+      "loss": 2.1128,
+      "step": 1063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681152082942851e-05,
+      "loss": 2.0251,
+      "step": 1064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6805194143407672e-05,
+      "loss": 2.0431,
+      "step": 1065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679886237984353e-05,
+      "loss": 2.0584,
+      "step": 1066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679252554346039e-05,
+      "loss": 2.0676,
+      "step": 1067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6786183638986336e-05,
+      "loss": 2.0422,
+      "step": 1068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6779836671153247e-05,
+      "loss": 2.1321,
+      "step": 1069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773484644696764e-05,
+      "loss": 2.0751,
+      "step": 1070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6767127564356312e-05,
+      "loss": 2.2097,
+      "step": 1071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6760765434875082e-05,
+      "loss": 1.9918,
+      "step": 1072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754398261000038e-05,
+      "loss": 2.0792,
+      "step": 1073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67480260474819e-05,
+      "loss": 1.9113,
+      "step": 1074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674164879907516e-05,
+      "loss": 2.0272,
+      "step": 1075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6735266520538046e-05,
+      "loss": 1.996,
+      "step": 1076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6728879216632567e-05,
+      "loss": 2.0135,
+      "step": 1077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6722486892124458e-05,
+      "loss": 1.9634,
+      "step": 1078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6716089551783212e-05,
+      "loss": 2.0898,
+      "step": 1079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6709687200382057e-05,
+      "loss": 2.0836,
+      "step": 1080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6703279842697974e-05,
+      "loss": 2.0394,
+      "step": 1081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6696867483511657e-05,
+      "loss": 2.0271,
+      "step": 1082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690450127607555e-05,
+      "loss": 2.1467,
+      "step": 1083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6684027779773827e-05,
+      "loss": 2.1812,
+      "step": 1084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6677600444802365e-05,
+      "loss": 2.103,
+      "step": 1085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671168127488785e-05,
+      "loss": 2.0507,
+      "step": 1086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664730832632417e-05,
+      "loss": 1.8555,
+      "step": 1087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66582885650363e-05,
+      "loss": 2.0102,
+      "step": 1088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665184132950719e-05,
+      "loss": 2.2401,
+      "step": 1089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6645389130855547e-05,
+      "loss": 1.9668,
+      "step": 1090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6638931973895537e-05,
+      "loss": 2.1121,
+      "step": 1091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663246986344502e-05,
+      "loss": 2.1751,
+      "step": 1092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626002804325555e-05,
+      "loss": 2.0091,
+      "step": 1093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6619530801362396e-05,
+      "loss": 2.0214,
+      "step": 1094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6613053859384483e-05,
+      "loss": 1.9509,
+      "step": 1095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660657198322444e-05,
+      "loss": 1.9866,
+      "step": 1096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660008517771857e-05,
+      "loss": 2.0551,
+      "step": 1097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6593593447706865e-05,
+      "loss": 1.8956,
+      "step": 1098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6587096798032984e-05,
+      "loss": 1.9995,
+      "step": 1099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658059523354425e-05,
+      "loss": 2.0836,
+      "step": 1100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6574088759091664e-05,
+      "loss": 2.0789,
+      "step": 1101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567577379529884e-05,
+      "loss": 1.8307,
+      "step": 1102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561061099717235e-05,
+      "loss": 1.9807,
+      "step": 1103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6554539924515688e-05,
+      "loss": 2.0764,
+      "step": 1104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6548013858790873e-05,
+      "loss": 2.1244,
+      "step": 1105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6541482907412073e-05,
+      "loss": 2.0074,
+      "step": 1106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534947075252205e-05,
+      "loss": 2.0529,
+      "step": 1107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528406367187836e-05,
+      "loss": 2.1062,
+      "step": 1108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6521860788099165e-05,
+      "loss": 2.0786,
+      "step": 1109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515310342870038e-05,
+      "loss": 2.2249,
+      "step": 1110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650875503638792e-05,
+      "loss": 2.0701,
+      "step": 1111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6502194873543902e-05,
+      "loss": 2.0088,
+      "step": 1112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495629859232706e-05,
+      "loss": 2.0509,
+      "step": 1113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6489059998352668e-05,
+      "loss": 2.0652,
+      "step": 1114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6482485295805748e-05,
+      "loss": 2.0662,
+      "step": 1115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475905756497505e-05,
+      "loss": 1.841,
+      "step": 1116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469321385337123e-05,
+      "loss": 2.0366,
+      "step": 1117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6462732187237377e-05,
+      "loss": 2.0557,
+      "step": 1118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6456138167114658e-05,
+      "loss": 1.9588,
+      "step": 1119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449539329888934e-05,
+      "loss": 2.0547,
+      "step": 1120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644293568048379e-05,
+      "loss": 1.9544,
+      "step": 1121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643632722382639e-05,
+      "loss": 2.0879,
+      "step": 1122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429713964847483e-05,
+      "loss": 2.0551,
+      "step": 1123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6423095908481405e-05,
+      "loss": 2.043,
+      "step": 1124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641647305966607e-05,
+      "loss": 1.9611,
+      "step": 1125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409845423342967e-05,
+      "loss": 2.0304,
+      "step": 1126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403213004457163e-05,
+      "loss": 2.0919,
+      "step": 1127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6396575807957288e-05,
+      "loss": 2.0027,
+      "step": 1128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6389933838795532e-05,
+      "loss": 1.9958,
+      "step": 1129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638328710192766e-05,
+      "loss": 2.0816,
+      "step": 1130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6376635602312983e-05,
+      "loss": 2.1248,
+      "step": 1131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369979344914364e-05,
+      "loss": 2.1249,
+      "step": 1132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6363318334698224e-05,
+      "loss": 1.9988,
+      "step": 1133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635665257663453e-05,
+      "loss": 2.103,
+      "step": 1134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6349982075696778e-05,
+      "loss": 2.0667,
+      "step": 1135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343306836862026e-05,
+      "loss": 1.9818,
+      "step": 1136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336626865110846e-05,
+      "loss": 1.9757,
+      "step": 1137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632994216542735e-05,
+      "loss": 2.1121,
+      "step": 1138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6323252742799182e-05,
+      "loss": 2.1197,
+      "step": 1139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63165586022175e-05,
+      "loss": 2.0426,
+      "step": 1140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6309859748676985e-05,
+      "loss": 2.046,
+      "step": 1141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6303156187175843e-05,
+      "loss": 2.2339,
+      "step": 1142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6296447922715782e-05,
+      "loss": 2.1643,
+      "step": 1143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6289734960302026e-05,
+      "loss": 1.993,
+      "step": 1144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283017304943296e-05,
+      "loss": 1.9325,
+      "step": 1145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6276294961651832e-05,
+      "loss": 2.1446,
+      "step": 1146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626956793544335e-05,
+      "loss": 2.183,
+      "step": 1147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626283623133707e-05,
+      "loss": 2.0298,
+      "step": 1148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625609985435571e-05,
+      "loss": 2.081,
+      "step": 1149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6249358809525457e-05,
+      "loss": 2.0124,
+      "step": 1150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6242613101875998e-05,
+      "loss": 1.9765,
+      "step": 1151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6235862736440488e-05,
+      "loss": 1.9998,
+      "step": 1152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229107718255566e-05,
+      "loss": 2.155,
+      "step": 1153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622234805236133e-05,
+      "loss": 2.0082,
+      "step": 1154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621558374380136e-05,
+      "loss": 1.9278,
+      "step": 1155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208814797622695e-05,
+      "loss": 2.1352,
+      "step": 1156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202041218875825e-05,
+      "loss": 1.9633,
+      "step": 1157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195263012614705e-05,
+      "loss": 1.9558,
+      "step": 1158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618848018389675e-05,
+      "loss": 2.008,
+      "step": 1159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181692737782806e-05,
+      "loss": 2.0697,
+      "step": 1160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6174900679337185e-05,
+      "loss": 2.1206,
+      "step": 1161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168104013627618e-05,
+      "loss": 2.1269,
+      "step": 1162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161302745725292e-05,
+      "loss": 1.9201,
+      "step": 1163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615449688070482e-05,
+      "loss": 1.9636,
+      "step": 1164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6147686423644243e-05,
+      "loss": 1.9432,
+      "step": 1165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140871379625033e-05,
+      "loss": 2.2027,
+      "step": 1166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6134051753732083e-05,
+      "loss": 2.0445,
+      "step": 1167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6127227551053704e-05,
+      "loss": 2.2301,
+      "step": 1168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.612039877668162e-05,
+      "loss": 2.1108,
+      "step": 1169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6113565435710975e-05,
+      "loss": 1.9934,
+      "step": 1170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.61067275332403e-05,
+      "loss": 1.9785,
+      "step": 1171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609988507437156e-05,
+      "loss": 2.0852,
+      "step": 1172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609303806421009e-05,
+      "loss": 1.936,
+      "step": 1173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6086186507864635e-05,
+      "loss": 1.9845,
+      "step": 1174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6079330410447337e-05,
+      "loss": 1.8619,
+      "step": 1175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6072469777073712e-05,
+      "loss": 2.0956,
+      "step": 1176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065604612862674e-05,
+      "loss": 2.0737,
+      "step": 1177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6058734922936507e-05,
+      "loss": 2.0087,
+      "step": 1178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605186071242088e-05,
+      "loss": 2.0417,
+      "step": 1179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044981986444827e-05,
+      "loss": 1.9942,
+      "step": 1180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603809875014076e-05,
+      "loss": 2.0002,
+      "step": 1181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603121100864445e-05,
+      "loss": 1.8892,
+      "step": 1182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602431876709503e-05,
+      "loss": 2.1565,
+      "step": 1183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6017422030634993e-05,
+      "loss": 2.0255,
+      "step": 1184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6010520804410185e-05,
+      "loss": 2.1183,
+      "step": 1185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6003615093569803e-05,
+      "loss": 2.1372,
+      "step": 1186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5996704903266384e-05,
+      "loss": 2.078,
+      "step": 1187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5989790238655812e-05,
+      "loss": 2.0253,
+      "step": 1188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5982871104897315e-05,
+      "loss": 1.9555,
+      "step": 1189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597594750715344e-05,
+      "loss": 1.9375,
+      "step": 1190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5969019450590087e-05,
+      "loss": 2.2252,
+      "step": 1191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5962086940376465e-05,
+      "loss": 2.0209,
+      "step": 1192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5955149981685107e-05,
+      "loss": 2.1932,
+      "step": 1193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5948208579691876e-05,
+      "loss": 2.0681,
+      "step": 1194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5941262739575937e-05,
+      "loss": 2.1148,
+      "step": 1195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593431246651978e-05,
+      "loss": 2.0945,
+      "step": 1196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592735776570919e-05,
+      "loss": 1.9131,
+      "step": 1197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5920398642333265e-05,
+      "loss": 2.0219,
+      "step": 1198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59134351015844e-05,
+      "loss": 2.1576,
+      "step": 1199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590646714865828e-05,
+      "loss": 2.0444,
+      "step": 1200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589949478875389e-05,
+      "loss": 2.1156,
+      "step": 1201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5892518027073504e-05,
+      "loss": 1.8617,
+      "step": 1202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588553686882267e-05,
+      "loss": 1.9904,
+      "step": 1203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5878551319210228e-05,
+      "loss": 2.0382,
+      "step": 1204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5871561383448287e-05,
+      "loss": 2.0144,
+      "step": 1205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586456706675223e-05,
+      "loss": 1.9559,
+      "step": 1206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5857568374340713e-05,
+      "loss": 2.0797,
+      "step": 1207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5850565311435652e-05,
+      "loss": 2.1611,
+      "step": 1208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843557883262224e-05,
+      "loss": 2.1894,
+      "step": 1209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836546095048873e-05,
+      "loss": 2.1399,
+      "step": 1210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829529952027276e-05,
+      "loss": 2.0913,
+      "step": 1211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582250945943238e-05,
+      "loss": 2.1283,
+      "step": 1212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581548462250236e-05,
+      "loss": 1.8905,
+      "step": 1213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580845544647865e-05,
+      "loss": 2.036,
+      "step": 1214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5801421936605904e-05,
+      "loss": 2.0889,
+      "step": 1215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579438409813203e-05,
+      "loss": 2.1695,
+      "step": 1216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5787341936308135e-05,
+      "loss": 1.9981,
+      "step": 1217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5780295456388587e-05,
+      "loss": 2.167,
+      "step": 1218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5773244663630955e-05,
+      "loss": 2.008,
+      "step": 1219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5766189563296027e-05,
+      "loss": 1.9873,
+      "step": 1220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575913016064781e-05,
+      "loss": 2.1178,
+      "step": 1221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5752066460953522e-05,
+      "loss": 2.0543,
+      "step": 1222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5744998469483576e-05,
+      "loss": 2.1543,
+      "step": 1223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5737926191511607e-05,
+      "loss": 1.9831,
+      "step": 1224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5730849632314428e-05,
+      "loss": 2.0396,
+      "step": 1225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723768797172057e-05,
+      "loss": 2.0369,
+      "step": 1226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716683691367704e-05,
+      "loss": 2.0141,
+      "step": 1227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570959432018776e-05,
+      "loss": 1.8861,
+      "step": 1228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5702500688921804e-05,
+      "loss": 2.009,
+      "step": 1229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5695402802862586e-05,
+      "loss": 2.0958,
+      "step": 1230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5688300667306034e-05,
+      "loss": 1.964,
+      "step": 1231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568119428755125e-05,
+      "loss": 2.0791,
+      "step": 1232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56740836689005e-05,
+      "loss": 2.1173,
+      "step": 1233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5666968816659213e-05,
+      "loss": 1.8829,
+      "step": 1234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5659849736135978e-05,
+      "loss": 2.0971,
+      "step": 1235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652726432642533e-05,
+      "loss": 2.0022,
+      "step": 1236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645598911493777e-05,
+      "loss": 1.8265,
+      "step": 1237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5638467178007744e-05,
+      "loss": 2.0991,
+      "step": 1238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5631331237505625e-05,
+      "loss": 1.9832,
+      "step": 1239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5624191095311736e-05,
+      "loss": 1.9539,
+      "step": 1240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561704675675354e-05,
+      "loss": 2.1531,
+      "step": 1241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5609898227161618e-05,
+      "loss": 2.0157,
+      "step": 1242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602745511869693e-05,
+      "loss": 2.1204,
+      "step": 1243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5595588616214596e-05,
+      "loss": 1.9266,
+      "step": 1244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558842754553629e-05,
+      "loss": 2.0615,
+      "step": 1245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5581262305177848e-05,
+      "loss": 1.9658,
+      "step": 1246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557409290048545e-05,
+      "loss": 2.061,
+      "step": 1247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5566919336808388e-05,
+      "loss": 2.1229,
+      "step": 1248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555974161949906e-05,
+      "loss": 2.0,
+      "step": 1249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552559753912952e-05,
+      "loss": 1.966,
+      "step": 1250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554537374540866e-05,
+      "loss": 1.5199,
+      "step": 1251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5538183599347852e-05,
+      "loss": 1.4886,
+      "step": 1252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553098932109531e-05,
+      "loss": 1.4961,
+      "step": 1253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552379091601887e-05,
+      "loss": 1.4554,
+      "step": 1254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551658838948947e-05,
+      "loss": 1.5205,
+      "step": 1255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550938174688111e-05,
+      "loss": 1.3946,
+      "step": 1256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5502170993570863e-05,
+      "loss": 1.4881,
+      "step": 1257
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5494956134938877e-05,
+      "loss": 1.4212,
+      "step": 1258
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5487737176368352e-05,
+      "loss": 1.4514,
+      "step": 1259
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5480514123245557e-05,
+      "loss": 1.3637,
+      "step": 1260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5473286980959805e-05,
+      "loss": 1.5361,
+      "step": 1261
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5466055754903472e-05,
+      "loss": 1.4138,
+      "step": 1262
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5458820450471976e-05,
+      "loss": 1.4126,
+      "step": 1263
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5451581073063772e-05,
+      "loss": 1.3707,
+      "step": 1264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544433762808036e-05,
+      "loss": 1.4751,
+      "step": 1265
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437090120926284e-05,
+      "loss": 1.5566,
+      "step": 1266
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54298385570091e-05,
+      "loss": 1.5796,
+      "step": 1267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54225829417394e-05,
+      "loss": 1.4938,
+      "step": 1268
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5415323280530804e-05,
+      "loss": 1.4589,
+      "step": 1269
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540805957879994e-05,
+      "loss": 1.4819,
+      "step": 1270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5400791841966466e-05,
+      "loss": 1.5304,
+      "step": 1271
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393520075453026e-05,
+      "loss": 1.4266,
+      "step": 1272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386244284685302e-05,
+      "loss": 1.5542,
+      "step": 1273
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378964475091952e-05,
+      "loss": 1.3573,
+      "step": 1274
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371680652104644e-05,
+      "loss": 1.5648,
+      "step": 1275
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364392821158042e-05,
+      "loss": 1.2882,
+      "step": 1276
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357100987689798e-05,
+      "loss": 1.3606,
+      "step": 1277
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349805157140553e-05,
+      "loss": 1.2781,
+      "step": 1278
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5342505334953922e-05,
+      "loss": 1.4364,
+      "step": 1279
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5335201526576505e-05,
+      "loss": 1.5174,
+      "step": 1280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532789373745788e-05,
+      "loss": 1.4761,
+      "step": 1281
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5320581973050587e-05,
+      "loss": 1.4595,
+      "step": 1282
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313266238810133e-05,
+      "loss": 1.3712,
+      "step": 1283
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5305946540194992e-05,
+      "loss": 1.2939,
+      "step": 1284
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529862288266659e-05,
+      "loss": 1.2925,
+      "step": 1285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291295271689318e-05,
+      "loss": 1.4234,
+      "step": 1286
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52839637127305e-05,
+      "loss": 1.4797,
+      "step": 1287
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5276628211260424e-05,
+      "loss": 1.3987,
+      "step": 1288
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5269288772752298e-05,
+      "loss": 1.461,
+      "step": 1289
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5261945402682292e-05,
+      "loss": 1.3517,
+      "step": 1290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525459810652949e-05,
+      "loss": 1.5289,
+      "step": 1291
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5247246889775915e-05,
+      "loss": 1.3701,
+      "step": 1292
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5239891757906509e-05,
+      "loss": 1.4686,
+      "step": 1293
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5232532716409148e-05,
+      "loss": 1.3453,
+      "step": 1294
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225169770774605e-05,
+      "loss": 1.4581,
+      "step": 1295
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5217802926496585e-05,
+      "loss": 1.4185,
+      "step": 1296
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210432189071691e-05,
+      "loss": 1.5142,
+      "step": 1297
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5203057563999437e-05,
+      "loss": 1.3549,
+      "step": 1298
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519567905678223e-05,
+      "loss": 1.4931,
+      "step": 1299
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5188296672925378e-05,
+      "loss": 1.2532,
+      "step": 1300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180910417937084e-05,
+      "loss": 1.3916,
+      "step": 1301
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173520297328437e-05,
+      "loss": 1.2732,
+      "step": 1302
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5166126316613409e-05,
+      "loss": 1.3864,
+      "step": 1303
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158728481308853e-05,
+      "loss": 1.51,
+      "step": 1304
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151326796934499e-05,
+      "loss": 1.4698,
+      "step": 1305
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514392126901295e-05,
+      "loss": 1.4337,
+      "step": 1306
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513651190306967e-05,
+      "loss": 1.3157,
+      "step": 1307
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129098704632994e-05,
+      "loss": 1.423,
+      "step": 1308
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121681679234112e-05,
+      "loss": 1.4064,
+      "step": 1309
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511426083240708e-05,
+      "loss": 1.4638,
+      "step": 1310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5106836169688788e-05,
+      "loss": 1.4592,
+      "step": 1311
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5099407696618983e-05,
+      "loss": 1.5686,
+      "step": 1312
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091975418740257e-05,
+      "loss": 1.3284,
+      "step": 1313
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084539341598036e-05,
+      "loss": 1.3091,
+      "step": 1314
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5077099470740582e-05,
+      "loss": 1.3538,
+      "step": 1315
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5069655811718988e-05,
+      "loss": 1.2863,
+      "step": 1316
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5062208370087178e-05,
+      "loss": 1.5392,
+      "step": 1317
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505475715140189e-05,
+      "loss": 1.3982,
+      "step": 1318
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5047302161222684e-05,
+      "loss": 1.3306,
+      "step": 1319
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503984340511193e-05,
+      "loss": 1.4184,
+      "step": 1320
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503238088863482e-05,
+      "loss": 1.4854,
+      "step": 1321
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5024914617359341e-05,
+      "loss": 1.4061,
+      "step": 1322
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017444596856283e-05,
+      "loss": 1.4181,
+      "step": 1323
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5009970832699234e-05,
+      "loss": 1.4264,
+      "step": 1324
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500249333046458e-05,
+      "loss": 1.3223,
+      "step": 1325
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995012095731487e-05,
+      "loss": 1.4622,
+      "step": 1326
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498752713408191e-05,
+      "loss": 1.5162,
+      "step": 1327
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980038451100589e-05,
+      "loss": 1.4249,
+      "step": 1328
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497254605237504e-05,
+      "loss": 1.4347,
+      "step": 1329
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.496504994349554e-05,
+      "loss": 1.3395,
+      "step": 1330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495755013005515e-05,
+      "loss": 1.504,
+      "step": 1331
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4950046617649684e-05,
+      "loss": 1.3939,
+      "step": 1332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4942539411877727e-05,
+      "loss": 1.2609,
+      "step": 1333
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4935028518340604e-05,
+      "loss": 1.2513,
+      "step": 1334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4927513942642402e-05,
+      "loss": 1.4252,
+      "step": 1335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4919995690389958e-05,
+      "loss": 1.5399,
+      "step": 1336
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912473767192842e-05,
+      "loss": 1.4675,
+      "step": 1337
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4904948178663374e-05,
+      "loss": 1.3371,
+      "step": 1338
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4897418930416598e-05,
+      "loss": 1.3351,
+      "step": 1339
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4889886028070295e-05,
+      "loss": 1.3205,
+      "step": 1340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882349477244975e-05,
+      "loss": 1.4257,
+      "step": 1341
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4874809283563867e-05,
+      "loss": 1.3122,
+      "step": 1342
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867265452652913e-05,
+      "loss": 1.5769,
+      "step": 1343
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4859717990140776e-05,
+      "loss": 1.376,
+      "step": 1344
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485216690165883e-05,
+      "loss": 1.4463,
+      "step": 1345
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4844612192841143e-05,
+      "loss": 1.3795,
+      "step": 1346
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4837053869324498e-05,
+      "loss": 1.3144,
+      "step": 1347
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829491936748368e-05,
+      "loss": 1.5569,
+      "step": 1348
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4821926400754915e-05,
+      "loss": 1.4278,
+      "step": 1349
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814357266989e-05,
+      "loss": 1.3739,
+      "step": 1350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806784541098162e-05,
+      "loss": 1.5259,
+      "step": 1351
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479920822873262e-05,
+      "loss": 1.3453,
+      "step": 1352
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791628335545269e-05,
+      "loss": 1.3719,
+      "step": 1353
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4784044867191676e-05,
+      "loss": 1.4023,
+      "step": 1354
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4776457829330078e-05,
+      "loss": 1.3358,
+      "step": 1355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768867227621375e-05,
+      "loss": 1.3767,
+      "step": 1356
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476127306772912e-05,
+      "loss": 1.7132,
+      "step": 1357
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753675355319527e-05,
+      "loss": 1.5239,
+      "step": 1358
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4746074096061463e-05,
+      "loss": 1.3091,
+      "step": 1359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738469295626435e-05,
+      "loss": 1.5133,
+      "step": 1360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473086095968859e-05,
+      "loss": 1.3811,
+      "step": 1361
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4723249093924725e-05,
+      "loss": 1.3599,
+      "step": 1362
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471563370401426e-05,
+      "loss": 1.4786,
+      "step": 1363
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4708014795639249e-05,
+      "loss": 1.4444,
+      "step": 1364
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470039237448437e-05,
+      "loss": 1.4369,
+      "step": 1365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692766446236914e-05,
+      "loss": 1.4035,
+      "step": 1366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4685137016586808e-05,
+      "loss": 1.4515,
+      "step": 1367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4677504091226576e-05,
+      "loss": 1.3604,
+      "step": 1368
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466986767585135e-05,
+      "loss": 1.4794,
+      "step": 1369
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662227776158877e-05,
+      "loss": 1.2529,
+      "step": 1370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4654584397849496e-05,
+      "loss": 1.3584,
+      "step": 1371
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646937546626141e-05,
+      "loss": 1.3986,
+      "step": 1372
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639287228194339e-05,
+      "loss": 1.3621,
+      "step": 1373
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4631633448262202e-05,
+      "loss": 1.4718,
+      "step": 1374
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462397621254043e-05,
+      "loss": 1.4467,
+      "step": 1375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4616315526742296e-05,
+      "loss": 1.4251,
+      "step": 1376
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4608651396583649e-05,
+      "loss": 1.3707,
+      "step": 1377
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460098382778291e-05,
+      "loss": 1.4508,
+      "step": 1378
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4593312826061063e-05,
+      "loss": 1.2687,
+      "step": 1379
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4585638397141657e-05,
+      "loss": 1.4205,
+      "step": 1380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457796054675079e-05,
+      "loss": 1.4804,
+      "step": 1381
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4570279280617119e-05,
+      "loss": 1.6334,
+      "step": 1382
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562594604471851e-05,
+      "loss": 1.3818,
+      "step": 1383
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455490652404874e-05,
+      "loss": 1.3645,
+      "step": 1384
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547215045084065e-05,
+      "loss": 1.3384,
+      "step": 1385
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539520173316654e-05,
+      "loss": 1.3147,
+      "step": 1386
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4531821914487869e-05,
+      "loss": 1.6073,
+      "step": 1387
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524120274341588e-05,
+      "loss": 1.4234,
+      "step": 1388
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451641525862422e-05,
+      "loss": 1.5198,
+      "step": 1389
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4508706873084691e-05,
+      "loss": 1.4815,
+      "step": 1390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4500995123474436e-05,
+      "loss": 1.5514,
+      "step": 1391
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4493280015547406e-05,
+      "loss": 1.3883,
+      "step": 1392
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485561555060059e-05,
+      "loss": 1.4709,
+      "step": 1393
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4477839747771347e-05,
+      "loss": 1.4957,
+      "step": 1394
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4470114599442729e-05,
+      "loss": 1.4036,
+      "step": 1395
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4462386115838145e-05,
+      "loss": 1.4091,
+      "step": 1396
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4454654302724036e-05,
+      "loss": 1.3676,
+      "step": 1397
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444691916586932e-05,
+      "loss": 1.4983,
+      "step": 1398
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439180711045395e-05,
+      "loss": 1.4933,
+      "step": 1399
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4431438944026135e-05,
+      "loss": 1.4883,
+      "step": 1400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423693870587886e-05,
+      "loss": 1.5289,
+      "step": 1401
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4415945496509465e-05,
+      "loss": 1.297,
+      "step": 1402
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4408193827572143e-05,
+      "loss": 1.2833,
+      "step": 1403
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400438869559659e-05,
+      "loss": 1.4018,
+      "step": 1404
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.43926806282582e-05,
+      "loss": 1.4689,
+      "step": 1405
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4384919109456402e-05,
+      "loss": 1.356,
+      "step": 1406
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437715431894535e-05,
+      "loss": 1.4919,
+      "step": 1407
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4369386262518568e-05,
+      "loss": 1.3713,
+      "step": 1408
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361614945972018e-05,
+      "loss": 1.3646,
+      "step": 1409
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4353840375104093e-05,
+      "loss": 1.2597,
+      "step": 1410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4346062555715619e-05,
+      "loss": 1.2797,
+      "step": 1411
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4338281493609835e-05,
+      "loss": 1.4289,
+      "step": 1412
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330497194592407e-05,
+      "loss": 1.5295,
+      "step": 1413
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4322709664471423e-05,
+      "loss": 1.5427,
+      "step": 1414
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4314918909057372e-05,
+      "loss": 1.5053,
+      "step": 1415
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4307124934163149e-05,
+      "loss": 1.5216,
+      "step": 1416
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429932774560405e-05,
+      "loss": 1.3527,
+      "step": 1417
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429152734919778e-05,
+      "loss": 1.2904,
+      "step": 1418
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4283723750764429e-05,
+      "loss": 1.5267,
+      "step": 1419
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4275916956126475e-05,
+      "loss": 1.4632,
+      "step": 1420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268106971108781e-05,
+      "loss": 1.3546,
+      "step": 1421
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4260293801538598e-05,
+      "loss": 1.4501,
+      "step": 1422
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252477453245546e-05,
+      "loss": 1.3872,
+      "step": 1423
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244657932061614e-05,
+      "loss": 1.2606,
+      "step": 1424
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4236835243821168e-05,
+      "loss": 1.4879,
+      "step": 1425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229009394360929e-05,
+      "loss": 1.3569,
+      "step": 1426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4221180389519984e-05,
+      "loss": 1.479,
+      "step": 1427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213348235139761e-05,
+      "loss": 1.4363,
+      "step": 1428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4205512937064055e-05,
+      "loss": 1.4399,
+      "step": 1429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4197674501138993e-05,
+      "loss": 1.467,
+      "step": 1430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418983293321305e-05,
+      "loss": 1.3479,
+      "step": 1431
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181988239137038e-05,
+      "loss": 1.4517,
+      "step": 1432
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41741404247641e-05,
+      "loss": 1.3898,
+      "step": 1433
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166289495949705e-05,
+      "loss": 1.4365,
+      "step": 1434
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415843545855165e-05,
+      "loss": 1.4992,
+      "step": 1435
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4150578318430044e-05,
+      "loss": 1.5308,
+      "step": 1436
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4142718081447324e-05,
+      "loss": 1.54,
+      "step": 1437
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134854753468226e-05,
+      "loss": 1.3933,
+      "step": 1438
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126988340359796e-05,
+      "loss": 1.49,
+      "step": 1439
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411911884799138e-05,
+      "loss": 1.4395,
+      "step": 1440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4111246282234626e-05,
+      "loss": 1.5619,
+      "step": 1441
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4103370648963475e-05,
+      "loss": 1.5005,
+      "step": 1442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.409549195405415e-05,
+      "loss": 1.4743,
+      "step": 1443
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4087610203385166e-05,
+      "loss": 1.517,
+      "step": 1444
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4079725402837315e-05,
+      "loss": 1.5285,
+      "step": 1445
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071837558293664e-05,
+      "loss": 1.4585,
+      "step": 1446
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4063946675639549e-05,
+      "loss": 1.4732,
+      "step": 1447
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4056052760762577e-05,
+      "loss": 1.3741,
+      "step": 1448
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4048155819552617e-05,
+      "loss": 1.3908,
+      "step": 1449
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040255857901797e-05,
+      "loss": 1.3973,
+      "step": 1450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403235288170449e-05,
+      "loss": 1.3177,
+      "step": 1451
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4024446896857331e-05,
+      "loss": 1.2593,
+      "step": 1452
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016537909259191e-05,
+      "loss": 1.4716,
+      "step": 1453
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4008625924811185e-05,
+      "loss": 1.4817,
+      "step": 1454
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4000710949416663e-05,
+      "loss": 1.4242,
+      "step": 1455
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3992792988981206e-05,
+      "loss": 1.3364,
+      "step": 1456
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3984872049412623e-05,
+      "loss": 1.5831,
+      "step": 1457
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976948136620947e-05,
+      "loss": 1.4545,
+      "step": 1458
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3969021256518424e-05,
+      "loss": 1.5038,
+      "step": 1459
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3961091415019525e-05,
+      "loss": 1.3621,
+      "step": 1460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953158618040918e-05,
+      "loss": 1.394,
+      "step": 1461
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3945222871501486e-05,
+      "loss": 1.3988,
+      "step": 1462
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937284181322308e-05,
+      "loss": 1.412,
+      "step": 1463
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3929342553426658e-05,
+      "loss": 1.5237,
+      "step": 1464
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921397993740003e-05,
+      "loss": 1.4397,
+      "step": 1465
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391345050819e-05,
+      "loss": 1.3342,
+      "step": 1466
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905500102706491e-05,
+      "loss": 1.5929,
+      "step": 1467
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3897546783221484e-05,
+      "loss": 1.3694,
+      "step": 1468
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.388959055566918e-05,
+      "loss": 1.4133,
+      "step": 1469
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3881631425985935e-05,
+      "loss": 1.5339,
+      "step": 1470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873669400110278e-05,
+      "loss": 1.5302,
+      "step": 1471
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3865704483982894e-05,
+      "loss": 1.2849,
+      "step": 1472
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385773668354663e-05,
+      "loss": 1.3443,
+      "step": 1473
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3849766004746477e-05,
+      "loss": 1.354,
+      "step": 1474
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841792453529582e-05,
+      "loss": 1.439,
+      "step": 1475
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3833816035845233e-05,
+      "loss": 1.3799,
+      "step": 1476
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3825836757644853e-05,
+      "loss": 1.3255,
+      "step": 1477
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3817854624882002e-05,
+      "loss": 1.3855,
+      "step": 1478
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809869643512368e-05,
+      "loss": 1.4254,
+      "step": 1479
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3801881819493772e-05,
+      "loss": 1.3832,
+      "step": 1480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3793891158786149e-05,
+      "loss": 1.448,
+      "step": 1481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3785897667351545e-05,
+      "loss": 1.2874,
+      "step": 1482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377790135115413e-05,
+      "loss": 1.369,
+      "step": 1483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3769902216160176e-05,
+      "loss": 1.5569,
+      "step": 1484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376190026833806e-05,
+      "loss": 1.434,
+      "step": 1485
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753895513658249e-05,
+      "loss": 1.4307,
+      "step": 1486
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374588795809332e-05,
+      "loss": 1.5067,
+      "step": 1487
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3737877607617926e-05,
+      "loss": 1.4275,
+      "step": 1488
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3729864468208817e-05,
+      "loss": 1.4282,
+      "step": 1489
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721848545844812e-05,
+      "loss": 1.5925,
+      "step": 1490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3713829846506813e-05,
+      "loss": 1.3954,
+      "step": 1491
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3705808376177791e-05,
+      "loss": 1.4708,
+      "step": 1492
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3697784140842796e-05,
+      "loss": 1.528,
+      "step": 1493
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3689757146488918e-05,
+      "loss": 1.3875,
+      "step": 1494
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3681727399105328e-05,
+      "loss": 1.3771,
+      "step": 1495
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367369490468324e-05,
+      "loss": 1.4352,
+      "step": 1496
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366565966921592e-05,
+      "loss": 1.3498,
+      "step": 1497
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365762169869868e-05,
+      "loss": 1.3987,
+      "step": 1498
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3649580999128871e-05,
+      "loss": 1.4374,
+      "step": 1499
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3641537576505882e-05,
+      "loss": 1.3938,
+      "step": 1500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3633491436831131e-05,
+      "loss": 1.4352,
+      "step": 1501
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3625442586108066e-05,
+      "loss": 1.4606,
+      "step": 1502
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3617391030342158e-05,
+      "loss": 1.4801,
+      "step": 1503
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609336775540892e-05,
+      "loss": 1.5535,
+      "step": 1504
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3601279827713771e-05,
+      "loss": 1.3629,
+      "step": 1505
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3593220192872308e-05,
+      "loss": 1.3985,
+      "step": 1506
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358515787703002e-05,
+      "loss": 1.4363,
+      "step": 1507
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.357709288620242e-05,
+      "loss": 1.3856,
+      "step": 1508
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3569025226407025e-05,
+      "loss": 1.4375,
+      "step": 1509
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3560954903663333e-05,
+      "loss": 1.4,
+      "step": 1510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3552881923992838e-05,
+      "loss": 1.5308,
+      "step": 1511
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3544806293419016e-05,
+      "loss": 1.537,
+      "step": 1512
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353672801796731e-05,
+      "loss": 1.4314,
+      "step": 1513
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3528647103665149e-05,
+      "loss": 1.3809,
+      "step": 1514
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352056355654193e-05,
+      "loss": 1.5236,
+      "step": 1515
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512477382629008e-05,
+      "loss": 1.4296,
+      "step": 1516
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504388587959695e-05,
+      "loss": 1.3762,
+      "step": 1517
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3496297178569275e-05,
+      "loss": 1.5083,
+      "step": 1518
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3488203160494963e-05,
+      "loss": 1.3264,
+      "step": 1519
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3480106539775935e-05,
+      "loss": 1.426,
+      "step": 1520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3472007322453298e-05,
+      "loss": 1.471,
+      "step": 1521
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463905514570105e-05,
+      "loss": 1.4312,
+      "step": 1522
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345580112217134e-05,
+      "loss": 1.4058,
+      "step": 1523
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344769415130391e-05,
+      "loss": 1.3457,
+      "step": 1524
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3439584608016654e-05,
+      "loss": 1.4246,
+      "step": 1525
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3431472498360325e-05,
+      "loss": 1.3896,
+      "step": 1526
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3423357828387589e-05,
+      "loss": 1.381,
+      "step": 1527
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341524060415303e-05,
+      "loss": 1.3823,
+      "step": 1528
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.340712083171313e-05,
+      "loss": 1.2307,
+      "step": 1529
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3398998517126275e-05,
+      "loss": 1.5409,
+      "step": 1530
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3390873666452753e-05,
+      "loss": 1.5202,
+      "step": 1531
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3382746285754736e-05,
+      "loss": 1.3458,
+      "step": 1532
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3374616381096285e-05,
+      "loss": 1.4975,
+      "step": 1533
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336648395854335e-05,
+      "loss": 1.5972,
+      "step": 1534
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3358349024163754e-05,
+      "loss": 1.419,
+      "step": 1535
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33502115840272e-05,
+      "loss": 1.4685,
+      "step": 1536
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342071644205253e-05,
+      "loss": 1.5517,
+      "step": 1537
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3333929210771346e-05,
+      "loss": 1.5507,
+      "step": 1538
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3325784289800776e-05,
+      "loss": 1.4057,
+      "step": 1539
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3317636887370695e-05,
+      "loss": 1.4409,
+      "step": 1540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33094870095601e-05,
+      "loss": 1.494,
+      "step": 1541
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330133466244984e-05,
+      "loss": 1.4179,
+      "step": 1542
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3293179852122613e-05,
+      "loss": 1.2741,
+      "step": 1543
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3285022584662948e-05,
+      "loss": 1.3995,
+      "step": 1544
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3276862866157199e-05,
+      "loss": 1.5713,
+      "step": 1545
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3268700702693561e-05,
+      "loss": 1.4212,
+      "step": 1546
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3260536100362055e-05,
+      "loss": 1.3175,
+      "step": 1547
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.325236906525451e-05,
+      "loss": 1.4957,
+      "step": 1548
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244199603464581e-05,
+      "loss": 1.192,
+      "step": 1549
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236027721087724e-05,
+      "loss": 1.3951,
+      "step": 1550
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3227853424221206e-05,
+      "loss": 1.5306,
+      "step": 1551
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219676718964103e-05,
+      "loss": 1.6042,
+      "step": 1552
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211497611417273e-05,
+      "loss": 1.5467,
+      "step": 1553
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203316107683376e-05,
+      "loss": 1.3002,
+      "step": 1554
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3195132213866865e-05,
+      "loss": 1.5206,
+      "step": 1555
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3186945936073961e-05,
+      "loss": 1.5053,
+      "step": 1556
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317875728041267e-05,
+      "loss": 1.4805,
+      "step": 1557
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3170566252992782e-05,
+      "loss": 1.5254,
+      "step": 1558
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162372859925845e-05,
+      "loss": 1.4268,
+      "step": 1559
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3154177107325174e-05,
+      "loss": 1.4437,
+      "step": 1560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145979001305849e-05,
+      "loss": 1.4107,
+      "step": 1561
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3137778547984704e-05,
+      "loss": 1.4194,
+      "step": 1562
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3129575753480322e-05,
+      "loss": 1.3659,
+      "step": 1563
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121370623913033e-05,
+      "loss": 1.5579,
+      "step": 1564
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113163165404913e-05,
+      "loss": 1.3553,
+      "step": 1565
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3104953384079772e-05,
+      "loss": 1.4931,
+      "step": 1566
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3096741286063162e-05,
+      "loss": 1.4716,
+      "step": 1567
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088526877482343e-05,
+      "loss": 1.5263,
+      "step": 1568
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.308031016446632e-05,
+      "loss": 1.3165,
+      "step": 1569
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.307209115314581e-05,
+      "loss": 1.3682,
+      "step": 1570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3063869849653244e-05,
+      "loss": 1.3977,
+      "step": 1571
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055646260122763e-05,
+      "loss": 1.3818,
+      "step": 1572
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3047420390690211e-05,
+      "loss": 1.5744,
+      "step": 1573
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.303919224749314e-05,
+      "loss": 1.4652,
+      "step": 1574
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3030961836670794e-05,
+      "loss": 1.4566,
+      "step": 1575
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302272916436411e-05,
+      "loss": 1.3942,
+      "step": 1576
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3014494236715711e-05,
+      "loss": 1.4514,
+      "step": 1577
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3006257059869907e-05,
+      "loss": 1.3885,
+      "step": 1578
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2998017639972678e-05,
+      "loss": 1.4102,
+      "step": 1579
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2989775983171688e-05,
+      "loss": 1.4057,
+      "step": 1580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981532095616262e-05,
+      "loss": 1.3876,
+      "step": 1581
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973285983457394e-05,
+      "loss": 1.4518,
+      "step": 1582
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2965037652847734e-05,
+      "loss": 1.4553,
+      "step": 1583
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295678710994159e-05,
+      "loss": 1.4334,
+      "step": 1584
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294853436089492e-05,
+      "loss": 1.4896,
+      "step": 1585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940279411865328e-05,
+      "loss": 1.4025,
+      "step": 1586
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293202226901206e-05,
+      "loss": 1.4184,
+      "step": 1587
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2923762938495996e-05,
+      "loss": 1.3865,
+      "step": 1588
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2915501426479655e-05,
+      "loss": 1.4382,
+      "step": 1589
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2907237739127174e-05,
+      "loss": 1.3508,
+      "step": 1590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2898971882604325e-05,
+      "loss": 1.3573,
+      "step": 1591
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2890703863078487e-05,
+      "loss": 1.4157,
+      "step": 1592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2882433686718656e-05,
+      "loss": 1.4982,
+      "step": 1593
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2874161359695446e-05,
+      "loss": 1.5905,
+      "step": 1594
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2865886888181059e-05,
+      "loss": 1.2865,
+      "step": 1595
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2857610278349315e-05,
+      "loss": 1.4524,
+      "step": 1596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2849331536375615e-05,
+      "loss": 1.4695,
+      "step": 1597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2841050668436965e-05,
+      "loss": 1.2064,
+      "step": 1598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2832767680711941e-05,
+      "loss": 1.3949,
+      "step": 1599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824482579380715e-05,
+      "loss": 1.5902,
+      "step": 1600
+    }
+  ],
+  "max_steps": 3750,
+  "num_train_epochs": 3,
+  "total_flos": 1.62836665667158e+17,
+  "trial_name": null,
+  "trial_params": null
+}