diff --git "a/checkpoint-26010/trainer_state.json" "b/checkpoint-26010/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-26010/trainer_state.json"
@@ -0,0 +1,156085 @@
+{
+  "best_metric": 0.45247501134872437,
+  "best_model_checkpoint": "./medium-article-titles-engagement-all/checkpoint-26010",
+  "epoch": 1.0,
+  "global_step": 26010,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6875,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.6914,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.6886,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7051,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6932,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.6967,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.6975,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7076,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.6945,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.7013,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.6972,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6986,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.7057,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6878,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.6905,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6911,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.689,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 0.7046,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6864,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.6986,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 0.692,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.6996,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.7021,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6944,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6939,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.6931,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.6865,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.6789,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9e-05,
+      "loss": 0.6938,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 0.707,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1e-05,
+      "loss": 0.6978,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6878,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3e-05,
+      "loss": 0.7003,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.6971,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 0.6861,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 0.695,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7e-05,
+      "loss": 0.6992,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-05,
+      "loss": 0.6953,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.703,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 0.7045,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1e-05,
+      "loss": 0.6891,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-05,
+      "loss": 0.6778,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3e-05,
+      "loss": 0.6909,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.6907,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 0.7057,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.7291,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7e-05,
+      "loss": 0.6999,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 0.693,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9e-05,
+      "loss": 0.6911,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 0.6846,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999358809951274e-05,
+      "loss": 0.6862,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999871761990254e-05,
+      "loss": 0.6899,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999807642985381e-05,
+      "loss": 0.6806,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999743523980508e-05,
+      "loss": 0.6772,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9996794049756354e-05,
+      "loss": 0.6709,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999615285970762e-05,
+      "loss": 0.7035,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999551166965889e-05,
+      "loss": 0.7047,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9994870479610156e-05,
+      "loss": 0.6776,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999422928956143e-05,
+      "loss": 0.6891,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99935880995127e-05,
+      "loss": 0.7329,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9992946909463964e-05,
+      "loss": 0.6923,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9992305719415236e-05,
+      "loss": 0.6773,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999166452936651e-05,
+      "loss": 0.7002,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999102333931778e-05,
+      "loss": 0.677,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9990382149269044e-05,
+      "loss": 0.6776,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9989740959220316e-05,
+      "loss": 0.6733,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998909976917158e-05,
+      "loss": 0.6679,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998845857912285e-05,
+      "loss": 0.6719,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9987817389074124e-05,
+      "loss": 0.734,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9987176199025396e-05,
+      "loss": 0.753,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998653500897667e-05,
+      "loss": 0.6637,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998589381892793e-05,
+      "loss": 0.6924,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9985252628879204e-05,
+      "loss": 0.6542,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998461143883047e-05,
+      "loss": 0.6428,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998397024878174e-05,
+      "loss": 0.6298,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9983329058733006e-05,
+      "loss": 0.7027,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998268786868428e-05,
+      "loss": 0.7037,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998204667863555e-05,
+      "loss": 0.6772,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998140548858682e-05,
+      "loss": 0.6772,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998076429853809e-05,
+      "loss": 0.7103,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998012310848936e-05,
+      "loss": 0.7194,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997948191844063e-05,
+      "loss": 0.6671,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9978840728391895e-05,
+      "loss": 0.7094,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9978199538343166e-05,
+      "loss": 0.6957,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997755834829443e-05,
+      "loss": 0.701,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997691715824571e-05,
+      "loss": 0.7063,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997627596819698e-05,
+      "loss": 0.6921,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9975634778148246e-05,
+      "loss": 0.6906,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997499358809952e-05,
+      "loss": 0.6544,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997435239805078e-05,
+      "loss": 0.6991,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9973711208002055e-05,
+      "loss": 0.6843,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997307001795332e-05,
+      "loss": 0.7162,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997242882790459e-05,
+      "loss": 0.6903,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997178763785586e-05,
+      "loss": 0.6632,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9971146447807135e-05,
+      "loss": 0.6624,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9970505257758406e-05,
+      "loss": 0.6999,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996986406770967e-05,
+      "loss": 0.6853,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996922287766094e-05,
+      "loss": 0.6804,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996858168761221e-05,
+      "loss": 0.6837,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996794049756348e-05,
+      "loss": 0.6734,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9967299307514745e-05,
+      "loss": 0.6601,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9966658117466016e-05,
+      "loss": 0.69,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996601692741729e-05,
+      "loss": 0.6816,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996537573736856e-05,
+      "loss": 0.6952,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996473454731983e-05,
+      "loss": 0.6727,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9964093357271097e-05,
+      "loss": 0.6923,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996345216722237e-05,
+      "loss": 0.6931,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996281097717363e-05,
+      "loss": 0.6269,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9962169787124905e-05,
+      "loss": 0.6633,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996152859707618e-05,
+      "loss": 0.6427,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996088740702745e-05,
+      "loss": 0.7126,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996024621697871e-05,
+      "loss": 0.6922,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9959605026929985e-05,
+      "loss": 0.6797,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995896383688126e-05,
+      "loss": 0.6848,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995832264683252e-05,
+      "loss": 0.6809,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9957681456783793e-05,
+      "loss": 0.6594,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995704026673506e-05,
+      "loss": 0.669,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995639907668633e-05,
+      "loss": 0.6672,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99557578866376e-05,
+      "loss": 0.6758,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9955116696588873e-05,
+      "loss": 0.6773,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995447550654014e-05,
+      "loss": 0.6838,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995383431649141e-05,
+      "loss": 0.7066,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995319312644268e-05,
+      "loss": 0.6666,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995255193639395e-05,
+      "loss": 0.6994,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995191074634522e-05,
+      "loss": 0.6532,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995126955629649e-05,
+      "loss": 0.63,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995062836624776e-05,
+      "loss": 0.655,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994998717619903e-05,
+      "loss": 0.7004,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99493459861503e-05,
+      "loss": 0.7173,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9948704796101564e-05,
+      "loss": 0.6564,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9948063606052835e-05,
+      "loss": 0.6169,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994742241600411e-05,
+      "loss": 0.6415,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994678122595537e-05,
+      "loss": 0.669,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9946140035906644e-05,
+      "loss": 0.696,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9945498845857915e-05,
+      "loss": 0.7049,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994485765580919e-05,
+      "loss": 0.7374,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994421646576045e-05,
+      "loss": 0.7051,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9943575275711724e-05,
+      "loss": 0.6312,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994293408566299e-05,
+      "loss": 0.6345,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994229289561426e-05,
+      "loss": 0.6694,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994165170556553e-05,
+      "loss": 0.6714,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9941010515516804e-05,
+      "loss": 0.6624,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9940369325468076e-05,
+      "loss": 0.7267,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993972813541934e-05,
+      "loss": 0.6634,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993908694537061e-05,
+      "loss": 0.6911,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993844575532188e-05,
+      "loss": 0.693,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993780456527315e-05,
+      "loss": 0.6908,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993716337522442e-05,
+      "loss": 0.6699,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9936522185175686e-05,
+      "loss": 0.7021,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993588099512696e-05,
+      "loss": 0.6818,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993523980507823e-05,
+      "loss": 0.6696,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99345986150295e-05,
+      "loss": 0.6535,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9933957424980766e-05,
+      "loss": 0.6887,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993331623493204e-05,
+      "loss": 0.7054,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99326750448833e-05,
+      "loss": 0.6678,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9932033854834574e-05,
+      "loss": 0.658,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9931392664785846e-05,
+      "loss": 0.6759,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993075147473712e-05,
+      "loss": 0.664,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993011028468839e-05,
+      "loss": 0.6591,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9929469094639654e-05,
+      "loss": 0.6613,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9928827904590926e-05,
+      "loss": 0.6653,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992818671454219e-05,
+      "loss": 0.7277,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992754552449346e-05,
+      "loss": 0.6702,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992690433444473e-05,
+      "loss": 0.7418,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9926263144396e-05,
+      "loss": 0.6932,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992562195434727e-05,
+      "loss": 0.6749,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992498076429854e-05,
+      "loss": 0.7208,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9924339574249814e-05,
+      "loss": 0.6776,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992369838420108e-05,
+      "loss": 0.6869,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992305719415235e-05,
+      "loss": 0.6796,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9922416004103616e-05,
+      "loss": 0.6983,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992177481405489e-05,
+      "loss": 0.6474,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992113362400615e-05,
+      "loss": 0.6994,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9920492433957424e-05,
+      "loss": 0.6765,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99198512439087e-05,
+      "loss": 0.6874,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991921005385997e-05,
+      "loss": 0.6622,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991856886381124e-05,
+      "loss": 0.691,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9917927673762504e-05,
+      "loss": 0.6644,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9917286483713776e-05,
+      "loss": 0.698,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991664529366504e-05,
+      "loss": 0.7033,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991600410361631e-05,
+      "loss": 0.6996,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9915362913567584e-05,
+      "loss": 0.6781,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9914721723518856e-05,
+      "loss": 0.6521,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991408053347013e-05,
+      "loss": 0.6644,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991343934342139e-05,
+      "loss": 0.6924,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9912798153372664e-05,
+      "loss": 0.6455,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991215696332393e-05,
+      "loss": 0.6645,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99115157732752e-05,
+      "loss": 0.7119,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9910874583226466e-05,
+      "loss": 0.657,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991023339317774e-05,
+      "loss": 0.6966,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990959220312901e-05,
+      "loss": 0.6488,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990895101308028e-05,
+      "loss": 0.6564,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990830982303155e-05,
+      "loss": 0.6699,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990766863298282e-05,
+      "loss": 0.6909,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990702744293409e-05,
+      "loss": 0.6717,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9906386252885355e-05,
+      "loss": 0.6881,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9905745062836626e-05,
+      "loss": 0.6703,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99051038727879e-05,
+      "loss": 0.6634,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990446268273917e-05,
+      "loss": 0.6661,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9903821492690435e-05,
+      "loss": 0.6691,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9903180302641706e-05,
+      "loss": 0.6542,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990253911259298e-05,
+      "loss": 0.6716,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990189792254424e-05,
+      "loss": 0.6792,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9901256732495515e-05,
+      "loss": 0.6455,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990061554244678e-05,
+      "loss": 0.6509,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989997435239805e-05,
+      "loss": 0.6808,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989933316234932e-05,
+      "loss": 0.6797,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9898691972300595e-05,
+      "loss": 0.6784,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989805078225186e-05,
+      "loss": 0.6672,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989740959220313e-05,
+      "loss": 0.7168,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98967684021544e-05,
+      "loss": 0.6476,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989612721210567e-05,
+      "loss": 0.6922,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989548602205694e-05,
+      "loss": 0.6435,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989484483200821e-05,
+      "loss": 0.6573,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989420364195948e-05,
+      "loss": 0.6504,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989356245191075e-05,
+      "loss": 0.6728,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989292126186202e-05,
+      "loss": 0.6659,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9892280071813285e-05,
+      "loss": 0.6871,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9891638881764557e-05,
+      "loss": 0.6709,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989099769171583e-05,
+      "loss": 0.6336,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989035650166709e-05,
+      "loss": 0.68,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9889715311618365e-05,
+      "loss": 0.6536,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988907412156964e-05,
+      "loss": 0.6244,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988843293152091e-05,
+      "loss": 0.6552,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988779174147217e-05,
+      "loss": 0.6667,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9887150551423445e-05,
+      "loss": 0.6739,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988650936137471e-05,
+      "loss": 0.6252,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988586817132598e-05,
+      "loss": 0.7028,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9885226981277253e-05,
+      "loss": 0.6613,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9884585791228525e-05,
+      "loss": 0.6663,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98839446011798e-05,
+      "loss": 0.6131,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988330341113106e-05,
+      "loss": 0.7338,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9882662221082334e-05,
+      "loss": 0.6186,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98820210310336e-05,
+      "loss": 0.6284,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988137984098487e-05,
+      "loss": 0.6667,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9880738650936135e-05,
+      "loss": 0.7164,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988009746088741e-05,
+      "loss": 0.6672,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987945627083868e-05,
+      "loss": 0.6465,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987881508078995e-05,
+      "loss": 0.6449,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987817389074122e-05,
+      "loss": 0.6182,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987753270069249e-05,
+      "loss": 0.6723,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987689151064376e-05,
+      "loss": 0.646,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9876250320595024e-05,
+      "loss": 0.6653,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9875609130546295e-05,
+      "loss": 0.6917,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987496794049756e-05,
+      "loss": 0.6323,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987432675044884e-05,
+      "loss": 0.7016,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987368556040011e-05,
+      "loss": 0.6315,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9873044370351375e-05,
+      "loss": 0.6481,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987240318030265e-05,
+      "loss": 0.6611,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987176199025391e-05,
+      "loss": 0.6377,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9871120800205184e-05,
+      "loss": 0.6525,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987047961015645e-05,
+      "loss": 0.6918,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986983842010772e-05,
+      "loss": 0.6251,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986919723005899e-05,
+      "loss": 0.6595,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9868556040010264e-05,
+      "loss": 0.6906,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9867914849961536e-05,
+      "loss": 0.6699,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98672736599128e-05,
+      "loss": 0.6785,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986663246986407e-05,
+      "loss": 0.6728,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986599127981534e-05,
+      "loss": 0.6465,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986535008976661e-05,
+      "loss": 0.6385,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9864708899717874e-05,
+      "loss": 0.6783,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9864067709669146e-05,
+      "loss": 0.6517,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986342651962042e-05,
+      "loss": 0.6056,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986278532957169e-05,
+      "loss": 0.6538,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986214413952296e-05,
+      "loss": 0.6287,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9861502949474226e-05,
+      "loss": 0.6738,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98608617594255e-05,
+      "loss": 0.6437,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986022056937676e-05,
+      "loss": 0.6096,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9859579379328034e-05,
+      "loss": 0.6127,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9858938189279306e-05,
+      "loss": 0.6439,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985829699923058e-05,
+      "loss": 0.693,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985765580918184e-05,
+      "loss": 0.6826,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9857014619133114e-05,
+      "loss": 0.6542,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9856373429084386e-05,
+      "loss": 0.6542,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985573223903565e-05,
+      "loss": 0.7074,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985509104898692e-05,
+      "loss": 0.6379,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985444985893819e-05,
+      "loss": 0.6805,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985380866888946e-05,
+      "loss": 0.6123,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985316747884073e-05,
+      "loss": 0.6137,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9852526288792e-05,
+      "loss": 0.6421,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985188509874327e-05,
+      "loss": 0.635,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985124390869454e-05,
+      "loss": 0.6754,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985060271864581e-05,
+      "loss": 0.6953,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9849961528597076e-05,
+      "loss": 0.6877,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984932033854835e-05,
+      "loss": 0.6194,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984867914849962e-05,
+      "loss": 0.67,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984803795845089e-05,
+      "loss": 0.6293,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9847396768402156e-05,
+      "loss": 0.6375,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984675557835343e-05,
+      "loss": 0.694,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984611438830469e-05,
+      "loss": 0.6346,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9845473198255964e-05,
+      "loss": 0.7256,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9844832008207236e-05,
+      "loss": 0.6914,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98441908181585e-05,
+      "loss": 0.6517,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984354962810977e-05,
+      "loss": 0.6647,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9842908438061044e-05,
+      "loss": 0.7181,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9842267248012316e-05,
+      "loss": 0.6915,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984162605796358e-05,
+      "loss": 0.6824,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984098486791485e-05,
+      "loss": 0.6154,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984034367786612e-05,
+      "loss": 0.6559,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983970248781739e-05,
+      "loss": 0.702,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983906129776866e-05,
+      "loss": 0.6856,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983842010771993e-05,
+      "loss": 0.683,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9837778917671205e-05,
+      "loss": 0.6675,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983713772762247e-05,
+      "loss": 0.6311,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983649653757374e-05,
+      "loss": 0.636,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9835855347525006e-05,
+      "loss": 0.6535,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983521415747628e-05,
+      "loss": 0.6404,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983457296742754e-05,
+      "loss": 0.6698,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9833931777378815e-05,
+      "loss": 0.6385,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9833290587330086e-05,
+      "loss": 0.6764,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983264939728136e-05,
+      "loss": 0.66,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983200820723263e-05,
+      "loss": 0.7051,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9831367017183895e-05,
+      "loss": 0.6494,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9830725827135166e-05,
+      "loss": 0.6507,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983008463708643e-05,
+      "loss": 0.6481,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98294434470377e-05,
+      "loss": 0.708,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982880225698897e-05,
+      "loss": 0.6506,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9828161066940246e-05,
+      "loss": 0.6046,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982751987689152e-05,
+      "loss": 0.619,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982687868684278e-05,
+      "loss": 0.6548,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9826237496794055e-05,
+      "loss": 0.6492,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982559630674532e-05,
+      "loss": 0.5974,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982495511669659e-05,
+      "loss": 0.6586,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9824313926647856e-05,
+      "loss": 0.6798,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982367273659913e-05,
+      "loss": 0.6038,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98230315465504e-05,
+      "loss": 0.6457,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982239035650167e-05,
+      "loss": 0.5946,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982174916645294e-05,
+      "loss": 0.7132,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982110797640421e-05,
+      "loss": 0.6233,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982046678635548e-05,
+      "loss": 0.5911,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9819825596306745e-05,
+      "loss": 0.5719,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981918440625802e-05,
+      "loss": 0.6379,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981854321620928e-05,
+      "loss": 0.6674,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981790202616056e-05,
+      "loss": 0.6875,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981726083611183e-05,
+      "loss": 0.5306,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98166196460631e-05,
+      "loss": 0.6616,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981597845601437e-05,
+      "loss": 0.6349,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981533726596563e-05,
+      "loss": 0.7384,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9814696075916905e-05,
+      "loss": 0.7376,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981405488586817e-05,
+      "loss": 0.6593,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981341369581944e-05,
+      "loss": 0.6127,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9812772505770713e-05,
+      "loss": 0.5999,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9812131315721985e-05,
+      "loss": 0.6769,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981149012567326e-05,
+      "loss": 0.6062,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981084893562452e-05,
+      "loss": 0.6903,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9810207745575794e-05,
+      "loss": 0.7044,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980956655552706e-05,
+      "loss": 0.667,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980892536547833e-05,
+      "loss": 0.6666,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9808284175429595e-05,
+      "loss": 0.6397,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980764298538087e-05,
+      "loss": 0.6529,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980700179533214e-05,
+      "loss": 0.6008,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980636060528341e-05,
+      "loss": 0.6296,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980571941523468e-05,
+      "loss": 0.6476,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980507822518595e-05,
+      "loss": 0.6591,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980443703513722e-05,
+      "loss": 0.666,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9803795845088484e-05,
+      "loss": 0.6782,
+      "step": 356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9803154655039755e-05,
+      "loss": 0.5872,
+      "step": 357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980251346499103e-05,
+      "loss": 0.71,
+      "step": 358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98018722749423e-05,
+      "loss": 0.6091,
+      "step": 359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9801231084893564e-05,
+      "loss": 0.7609,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9800589894844835e-05,
+      "loss": 0.6844,
+      "step": 361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979994870479611e-05,
+      "loss": 0.7514,
+      "step": 362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979930751474737e-05,
+      "loss": 0.7359,
+      "step": 363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9798666324698644e-05,
+      "loss": 0.6408,
+      "step": 364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979802513464991e-05,
+      "loss": 0.6557,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979738394460118e-05,
+      "loss": 0.6067,
+      "step": 366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979674275455245e-05,
+      "loss": 0.6569,
+      "step": 367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9796101564503724e-05,
+      "loss": 0.6196,
+      "step": 368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979546037445499e-05,
+      "loss": 0.6015,
+      "step": 369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979481918440626e-05,
+      "loss": 0.6689,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979417799435753e-05,
+      "loss": 0.6738,
+      "step": 371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97935368043088e-05,
+      "loss": 0.5636,
+      "step": 372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979289561426007e-05,
+      "loss": 0.6747,
+      "step": 373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979225442421134e-05,
+      "loss": 0.5829,
+      "step": 374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979161323416261e-05,
+      "loss": 0.6046,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979097204411388e-05,
+      "loss": 0.7024,
+      "step": 376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979033085406515e-05,
+      "loss": 0.6996,
+      "step": 377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9789689664016414e-05,
+      "loss": 0.6969,
+      "step": 378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9789048473967686e-05,
+      "loss": 0.6852,
+      "step": 379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978840728391896e-05,
+      "loss": 0.7081,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978776609387022e-05,
+      "loss": 0.5689,
+      "step": 381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9787124903821494e-05,
+      "loss": 0.634,
+      "step": 382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9786483713772766e-05,
+      "loss": 0.6174,
+      "step": 383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978584252372404e-05,
+      "loss": 0.7074,
+      "step": 384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97852013336753e-05,
+      "loss": 0.6342,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9784560143626574e-05,
+      "loss": 0.6801,
+      "step": 386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978391895357784e-05,
+      "loss": 0.5336,
+      "step": 387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978327776352911e-05,
+      "loss": 0.6806,
+      "step": 388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978263657348038e-05,
+      "loss": 0.7306,
+      "step": 389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9781995383431654e-05,
+      "loss": 0.6278,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9781354193382926e-05,
+      "loss": 0.6593,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.978071300333419e-05,
+      "loss": 0.6317,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.978007181328546e-05,
+      "loss": 0.6302,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977943062323673e-05,
+      "loss": 0.7447,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9778789433188e-05,
+      "loss": 0.677,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9778148243139264e-05,
+      "loss": 0.6667,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9777507053090536e-05,
+      "loss": 0.6245,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977686586304181e-05,
+      "loss": 0.6026,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977622467299308e-05,
+      "loss": 0.676,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977558348294435e-05,
+      "loss": 0.6143,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9774942292895616e-05,
+      "loss": 0.657,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977430110284689e-05,
+      "loss": 0.6636,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977365991279815e-05,
+      "loss": 0.6205,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9773018722749424e-05,
+      "loss": 0.599,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977237753270069e-05,
+      "loss": 0.7161,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977173634265197e-05,
+      "loss": 0.6414,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977109515260324e-05,
+      "loss": 0.6789,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9770453962554504e-05,
+      "loss": 0.6112,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9769812772505776e-05,
+      "loss": 0.698,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976917158245704e-05,
+      "loss": 0.6958,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976853039240831e-05,
+      "loss": 0.6233,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976788920235958e-05,
+      "loss": 0.6292,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976724801231085e-05,
+      "loss": 0.6567,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976660682226212e-05,
+      "loss": 0.8298,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976596563221339e-05,
+      "loss": 0.7153,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9765324442164665e-05,
+      "loss": 0.6625,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976468325211593e-05,
+      "loss": 0.6642,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97640420620672e-05,
+      "loss": 0.6737,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9763400872018466e-05,
+      "loss": 0.6366,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976275968196974e-05,
+      "loss": 0.6784,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9762118491921e-05,
+      "loss": 0.6855,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9761477301872275e-05,
+      "loss": 0.6458,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9760836111823546e-05,
+      "loss": 0.6852,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976019492177482e-05,
+      "loss": 0.6874,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975955373172609e-05,
+      "loss": 0.6537,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9758912541677355e-05,
+      "loss": 0.6629,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9758271351628626e-05,
+      "loss": 0.641,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975763016157989e-05,
+      "loss": 0.666,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975698897153116e-05,
+      "loss": 0.6118,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9756347781482435e-05,
+      "loss": 0.638,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9755706591433707e-05,
+      "loss": 0.603,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975506540138497e-05,
+      "loss": 0.6276,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975442421133624e-05,
+      "loss": 0.6184,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9753783021287515e-05,
+      "loss": 0.708,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975314183123878e-05,
+      "loss": 0.7014,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975250064119005e-05,
+      "loss": 0.7006,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9751859451141316e-05,
+      "loss": 0.6508,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975121826109259e-05,
+      "loss": 0.5984,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975057707104386e-05,
+      "loss": 0.573,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974993588099513e-05,
+      "loss": 0.6654,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9749294690946397e-05,
+      "loss": 0.632,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974865350089767e-05,
+      "loss": 0.6371,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974801231084894e-05,
+      "loss": 0.7639,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9747371120800205e-05,
+      "loss": 0.6173,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974672993075148e-05,
+      "loss": 0.742,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974608874070275e-05,
+      "loss": 0.6789,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974544755065402e-05,
+      "loss": 0.6756,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9744806360605285e-05,
+      "loss": 0.6618,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974416517055656e-05,
+      "loss": 0.6753,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974352398050782e-05,
+      "loss": 0.6843,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9742882790459093e-05,
+      "loss": 0.6798,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9742241600410365e-05,
+      "loss": 0.6273,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974160041036163e-05,
+      "loss": 0.6631,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97409592203129e-05,
+      "loss": 0.5897,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9740318030264173e-05,
+      "loss": 0.629,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9739676840215445e-05,
+      "loss": 0.6795,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973903565016671e-05,
+      "loss": 0.6424,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973839446011798e-05,
+      "loss": 0.6624,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973775327006925e-05,
+      "loss": 0.6148,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973711208002052e-05,
+      "loss": 0.6378,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973647088997179e-05,
+      "loss": 0.6271,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973582969992306e-05,
+      "loss": 0.6655,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9735188509874334e-05,
+      "loss": 0.6797,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97345473198256e-05,
+      "loss": 0.635,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973390612977687e-05,
+      "loss": 0.6964,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9733264939728135e-05,
+      "loss": 0.6723,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973262374967941e-05,
+      "loss": 0.6123,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973198255963067e-05,
+      "loss": 0.7147,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9731341369581944e-05,
+      "loss": 0.6126,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9730700179533215e-05,
+      "loss": 0.6339,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973005898948449e-05,
+      "loss": 0.6323,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972941779943576e-05,
+      "loss": 0.5681,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9728776609387024e-05,
+      "loss": 0.6254,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9728135419338295e-05,
+      "loss": 0.6086,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972749422928956e-05,
+      "loss": 0.6252,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972685303924083e-05,
+      "loss": 0.6609,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97262118491921e-05,
+      "loss": 0.6388,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9725570659143376e-05,
+      "loss": 0.6624,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972492946909465e-05,
+      "loss": 0.6756,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972428827904591e-05,
+      "loss": 0.7042,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9723647088997184e-05,
+      "loss": 0.6019,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972300589894845e-05,
+      "loss": 0.6086,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972236470889972e-05,
+      "loss": 0.6523,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9721723518850986e-05,
+      "loss": 0.6145,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972108232880226e-05,
+      "loss": 0.7168,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972044113875353e-05,
+      "loss": 0.5385,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97197999487048e-05,
+      "loss": 0.6165,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971915875865607e-05,
+      "loss": 0.6073,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971851756860734e-05,
+      "loss": 0.6785,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971787637855861e-05,
+      "loss": 0.6888,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9717235188509874e-05,
+      "loss": 0.6306,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9716593998461146e-05,
+      "loss": 0.5827,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971595280841241e-05,
+      "loss": 0.6549,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971531161836369e-05,
+      "loss": 0.7205,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9714670428314954e-05,
+      "loss": 0.5896,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9714029238266226e-05,
+      "loss": 0.6253,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97133880482175e-05,
+      "loss": 0.6638,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971274685816876e-05,
+      "loss": 0.6144,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9712105668120034e-05,
+      "loss": 0.6595,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97114644780713e-05,
+      "loss": 0.6494,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971082328802257e-05,
+      "loss": 0.714,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971018209797384e-05,
+      "loss": 0.6523,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9709540907925114e-05,
+      "loss": 0.6692,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970889971787638e-05,
+      "loss": 0.6682,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970825852782765e-05,
+      "loss": 0.6382,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970761733777892e-05,
+      "loss": 0.617,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970697614773019e-05,
+      "loss": 0.6718,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970633495768146e-05,
+      "loss": 0.6417,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9705693767632724e-05,
+      "loss": 0.649,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9705052577583996e-05,
+      "loss": 0.6783,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970441138753527e-05,
+      "loss": 0.6132,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970377019748654e-05,
+      "loss": 0.6431,
+      "step": 512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9703129007437804e-05,
+      "loss": 0.6062,
+      "step": 513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9702487817389076e-05,
+      "loss": 0.647,
+      "step": 514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970184662734035e-05,
+      "loss": 0.597,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970120543729161e-05,
+      "loss": 0.6179,
+      "step": 516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9700564247242884e-05,
+      "loss": 0.6671,
+      "step": 517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9699923057194156e-05,
+      "loss": 0.6752,
+      "step": 518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969928186714543e-05,
+      "loss": 0.597,
+      "step": 519
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969864067709669e-05,
+      "loss": 0.5915,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9697999487047964e-05,
+      "loss": 0.6389,
+      "step": 521
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969735829699923e-05,
+      "loss": 0.6883,
+      "step": 522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96967171069505e-05,
+      "loss": 0.656,
+      "step": 523
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969607591690177e-05,
+      "loss": 0.5763,
+      "step": 524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969543472685304e-05,
+      "loss": 0.5899,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969479353680431e-05,
+      "loss": 0.6352,
+      "step": 526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969415234675558e-05,
+      "loss": 0.6731,
+      "step": 527
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969351115670685e-05,
+      "loss": 0.6156,
+      "step": 528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969286996665812e-05,
+      "loss": 0.6267,
+      "step": 529
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969222877660939e-05,
+      "loss": 0.6753,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969158758656066e-05,
+      "loss": 0.7151,
+      "step": 531
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9690946396511926e-05,
+      "loss": 0.5535,
+      "step": 532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96903052064632e-05,
+      "loss": 0.6686,
+      "step": 533
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968966401641447e-05,
+      "loss": 0.5681,
+      "step": 534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968902282636574e-05,
+      "loss": 0.6417,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9688381636317006e-05,
+      "loss": 0.6123,
+      "step": 536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968774044626828e-05,
+      "loss": 0.5953,
+      "step": 537
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968709925621954e-05,
+      "loss": 0.6945,
+      "step": 538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9686458066170815e-05,
+      "loss": 0.6493,
+      "step": 539
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9685816876122086e-05,
+      "loss": 0.5992,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968517568607335e-05,
+      "loss": 0.5946,
+      "step": 541
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968453449602462e-05,
+      "loss": 0.7016,
+      "step": 542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9683893305975895e-05,
+      "loss": 0.6745,
+      "step": 543
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9683252115927167e-05,
+      "loss": 0.6934,
+      "step": 544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968261092587843e-05,
+      "loss": 0.5999,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96819697358297e-05,
+      "loss": 0.6332,
+      "step": 546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968132854578097e-05,
+      "loss": 0.6845,
+      "step": 547
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968068735573224e-05,
+      "loss": 0.6592,
+      "step": 548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968004616568351e-05,
+      "loss": 0.7044,
+      "step": 549
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967940497563478e-05,
+      "loss": 0.6782,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9678763785586055e-05,
+      "loss": 0.6153,
+      "step": 551
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967812259553732e-05,
+      "loss": 0.6497,
+      "step": 552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967748140548859e-05,
+      "loss": 0.64,
+      "step": 553
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9676840215439857e-05,
+      "loss": 0.5976,
+      "step": 554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967619902539113e-05,
+      "loss": 0.6447,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967555783534239e-05,
+      "loss": 0.679,
+      "step": 556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9674916645293665e-05,
+      "loss": 0.7191,
+      "step": 557
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967427545524494e-05,
+      "loss": 0.5963,
+      "step": 558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967363426519621e-05,
+      "loss": 0.6973,
+      "step": 559
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967299307514748e-05,
+      "loss": 0.5447,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9672351885098745e-05,
+      "loss": 0.5777,
+      "step": 561
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967171069505002e-05,
+      "loss": 0.6521,
+      "step": 562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967106950500128e-05,
+      "loss": 0.6685,
+      "step": 563
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9670428314952553e-05,
+      "loss": 0.6211,
+      "step": 564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966978712490382e-05,
+      "loss": 0.6365,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96691459348551e-05,
+      "loss": 0.6297,
+      "step": 566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966850474480637e-05,
+      "loss": 0.6609,
+      "step": 567
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9667863554757634e-05,
+      "loss": 0.6742,
+      "step": 568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9667222364708905e-05,
+      "loss": 0.6379,
+      "step": 569
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966658117466017e-05,
+      "loss": 0.6345,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966593998461144e-05,
+      "loss": 0.6507,
+      "step": 571
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966529879456271e-05,
+      "loss": 0.6817,
+      "step": 572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966465760451398e-05,
+      "loss": 0.6335,
+      "step": 573
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966401641446525e-05,
+      "loss": 0.6243,
+      "step": 574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966337522441652e-05,
+      "loss": 0.669,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9662734034367794e-05,
+      "loss": 0.7154,
+      "step": 576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966209284431906e-05,
+      "loss": 0.5784,
+      "step": 577
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966145165427033e-05,
+      "loss": 0.5828,
+      "step": 578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9660810464221595e-05,
+      "loss": 0.6705,
+      "step": 579
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966016927417287e-05,
+      "loss": 0.5542,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965952808412413e-05,
+      "loss": 0.6549,
+      "step": 581
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9658886894075404e-05,
+      "loss": 0.6165,
+      "step": 582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9658245704026675e-05,
+      "loss": 0.6875,
+      "step": 583
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965760451397795e-05,
+      "loss": 0.6797,
+      "step": 584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965696332392922e-05,
+      "loss": 0.6613,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9656322133880484e-05,
+      "loss": 0.6784,
+      "step": 586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9655680943831755e-05,
+      "loss": 0.6743,
+      "step": 587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965503975378302e-05,
+      "loss": 0.5665,
+      "step": 588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965439856373429e-05,
+      "loss": 0.6927,
+      "step": 589
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9653757373685564e-05,
+      "loss": 0.6604,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9653116183636836e-05,
+      "loss": 0.5986,
+      "step": 591
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96524749935881e-05,
+      "loss": 0.5905,
+      "step": 592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965183380353937e-05,
+      "loss": 0.6559,
+      "step": 593
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9651192613490644e-05,
+      "loss": 0.6169,
+      "step": 594
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965055142344191e-05,
+      "loss": 0.6748,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964991023339318e-05,
+      "loss": 0.6785,
+      "step": 596
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9649269043344446e-05,
+      "loss": 0.7179,
+      "step": 597
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964862785329572e-05,
+      "loss": 0.6151,
+      "step": 598
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964798666324699e-05,
+      "loss": 0.626,
+      "step": 599
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964734547319826e-05,
+      "loss": 0.5889,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9646704283149526e-05,
+      "loss": 0.6531,
+      "step": 601
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96460630931008e-05,
+      "loss": 0.6651,
+      "step": 602
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964542190305207e-05,
+      "loss": 0.643,
+      "step": 603
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9644780713003334e-05,
+      "loss": 0.6693,
+      "step": 604
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9644139522954606e-05,
+      "loss": 0.636,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964349833290588e-05,
+      "loss": 0.5811,
+      "step": 606
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964285714285715e-05,
+      "loss": 0.7276,
+      "step": 607
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9642215952808414e-05,
+      "loss": 0.6619,
+      "step": 608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9641574762759686e-05,
+      "loss": 0.5992,
+      "step": 609
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964093357271095e-05,
+      "loss": 0.572,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964029238266222e-05,
+      "loss": 0.4896,
+      "step": 611
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9639651192613494e-05,
+      "loss": 0.5452,
+      "step": 612
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963901000256476e-05,
+      "loss": 0.6812,
+      "step": 613
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963836881251603e-05,
+      "loss": 0.6382,
+      "step": 614
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96377276224673e-05,
+      "loss": 0.6762,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9637086432418574e-05,
+      "loss": 0.6938,
+      "step": 616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963644524236984e-05,
+      "loss": 0.6374,
+      "step": 617
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963580405232111e-05,
+      "loss": 0.6314,
+      "step": 618
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9635162862272376e-05,
+      "loss": 0.643,
+      "step": 619
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963452167222365e-05,
+      "loss": 0.6124,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963388048217492e-05,
+      "loss": 0.7339,
+      "step": 621
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963323929212619e-05,
+      "loss": 0.5743,
+      "step": 622
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963259810207746e-05,
+      "loss": 0.6549,
+      "step": 623
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963195691202873e-05,
+      "loss": 0.5656,
+      "step": 624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963131572198e-05,
+      "loss": 0.637,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9630674531931264e-05,
+      "loss": 0.6799,
+      "step": 626
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9630033341882536e-05,
+      "loss": 0.6278,
+      "step": 627
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96293921518338e-05,
+      "loss": 0.6471,
+      "step": 628
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962875096178507e-05,
+      "loss": 0.6452,
+      "step": 629
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9628109771736344e-05,
+      "loss": 0.5867,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9627468581687616e-05,
+      "loss": 0.6439,
+      "step": 631
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962682739163889e-05,
+      "loss": 0.6095,
+      "step": 632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962618620159015e-05,
+      "loss": 0.6533,
+      "step": 633
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9625545011541425e-05,
+      "loss": 0.6396,
+      "step": 634
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962490382149269e-05,
+      "loss": 0.6318,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962426263144396e-05,
+      "loss": 0.63,
+      "step": 636
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9623621441395226e-05,
+      "loss": 0.6757,
+      "step": 637
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9622980251346505e-05,
+      "loss": 0.6398,
+      "step": 638
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9622339061297776e-05,
+      "loss": 0.6507,
+      "step": 639
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962169787124904e-05,
+      "loss": 0.5963,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962105668120031e-05,
+      "loss": 0.6532,
+      "step": 641
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962041549115158e-05,
+      "loss": 0.7162,
+      "step": 642
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.961977430110285e-05,
+      "loss": 0.7003,
+      "step": 643
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9619133111054115e-05,
+      "loss": 0.6422,
+      "step": 644
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9618491921005386e-05,
+      "loss": 0.6904,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.961785073095666e-05,
+      "loss": 0.6163,
+      "step": 646
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.961720954090793e-05,
+      "loss": 0.667,
+      "step": 647
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96165683508592e-05,
+      "loss": 0.6777,
+      "step": 648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9615927160810466e-05,
+      "loss": 0.5493,
+      "step": 649
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.961528597076174e-05,
+      "loss": 0.6507,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9614644780713e-05,
+      "loss": 0.7253,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9614003590664275e-05,
+      "loss": 0.6924,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961336240061554e-05,
+      "loss": 0.6597,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961272121056682e-05,
+      "loss": 0.5915,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961208002051808e-05,
+      "loss": 0.6417,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9611438830469355e-05,
+      "loss": 0.6283,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9610797640420627e-05,
+      "loss": 0.6488,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961015645037189e-05,
+      "loss": 0.6755,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960951526032316e-05,
+      "loss": 0.6043,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960887407027443e-05,
+      "loss": 0.6193,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.96082328802257e-05,
+      "loss": 0.6466,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960759169017697e-05,
+      "loss": 0.6463,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960695050012824e-05,
+      "loss": 0.6494,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960630931007951e-05,
+      "loss": 0.6123,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960566812003078e-05,
+      "loss": 0.6113,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960502692998205e-05,
+      "loss": 0.6986,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960438573993332e-05,
+      "loss": 0.6115,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960374454988459e-05,
+      "loss": 0.6127,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960310335983585e-05,
+      "loss": 0.6837,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9602462169787125e-05,
+      "loss": 0.6183,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.96018209797384e-05,
+      "loss": 0.7084,
+      "step": 671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960117978968967e-05,
+      "loss": 0.6384,
+      "step": 672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960053859964093e-05,
+      "loss": 0.6168,
+      "step": 673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9599897409592205e-05,
+      "loss": 0.6042,
+      "step": 674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959925621954348e-05,
+      "loss": 0.6329,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959861502949474e-05,
+      "loss": 0.5845,
+      "step": 676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9597973839446013e-05,
+      "loss": 0.6863,
+      "step": 677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9597332649397285e-05,
+      "loss": 0.6907,
+      "step": 678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959669145934856e-05,
+      "loss": 0.6315,
+      "step": 679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959605026929982e-05,
+      "loss": 0.6329,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9595409079251094e-05,
+      "loss": 0.6551,
+      "step": 681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959476788920236e-05,
+      "loss": 0.5753,
+      "step": 682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959412669915363e-05,
+      "loss": 0.6599,
+      "step": 683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95934855091049e-05,
+      "loss": 0.6489,
+      "step": 684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959284431905617e-05,
+      "loss": 0.6618,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959220312900744e-05,
+      "loss": 0.6326,
+      "step": 686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959156193895871e-05,
+      "loss": 0.6609,
+      "step": 687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959092074890998e-05,
+      "loss": 0.6211,
+      "step": 688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959027955886125e-05,
+      "loss": 0.6377,
+      "step": 689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958963836881252e-05,
+      "loss": 0.6092,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9588997178763784e-05,
+      "loss": 0.6439,
+      "step": 691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9588355988715055e-05,
+      "loss": 0.6341,
+      "step": 692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958771479866633e-05,
+      "loss": 0.6453,
+      "step": 693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95870736086176e-05,
+      "loss": 0.6493,
+      "step": 694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958643241856887e-05,
+      "loss": 0.5392,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9585791228520135e-05,
+      "loss": 0.6367,
+      "step": 696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958515003847141e-05,
+      "loss": 0.7526,
+      "step": 697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958450884842267e-05,
+      "loss": 0.7398,
+      "step": 698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9583867658373944e-05,
+      "loss": 0.6471,
+      "step": 699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958322646832521e-05,
+      "loss": 0.6762,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958258527827648e-05,
+      "loss": 0.6052,
+      "step": 701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958194408822775e-05,
+      "loss": 0.5991,
+      "step": 702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9581302898179024e-05,
+      "loss": 0.6992,
+      "step": 703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9580661708130296e-05,
+      "loss": 0.5962,
+      "step": 704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958002051808156e-05,
+      "loss": 0.6302,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957937932803283e-05,
+      "loss": 0.6557,
+      "step": 706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95787381379841e-05,
+      "loss": 0.7107,
+      "step": 707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957809694793537e-05,
+      "loss": 0.7375,
+      "step": 708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9577455757886634e-05,
+      "loss": 0.5859,
+      "step": 709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957681456783791e-05,
+      "loss": 0.7068,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9576173377789184e-05,
+      "loss": 0.6699,
+      "step": 711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957553218774045e-05,
+      "loss": 0.6456,
+      "step": 712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957489099769172e-05,
+      "loss": 0.6466,
+      "step": 713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9574249807642986e-05,
+      "loss": 0.7084,
+      "step": 714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957360861759426e-05,
+      "loss": 0.6147,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957296742754552e-05,
+      "loss": 0.7033,
+      "step": 716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9572326237496794e-05,
+      "loss": 0.6553,
+      "step": 717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9571685047448066e-05,
+      "loss": 0.6992,
+      "step": 718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957104385739934e-05,
+      "loss": 0.6454,
+      "step": 719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957040266735061e-05,
+      "loss": 0.5606,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9569761477301874e-05,
+      "loss": 0.6489,
+      "step": 721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9569120287253146e-05,
+      "loss": 0.621,
+      "step": 722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956847909720441e-05,
+      "loss": 0.6198,
+      "step": 723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956783790715568e-05,
+      "loss": 0.5835,
+      "step": 724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956719671710695e-05,
+      "loss": 0.6816,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9566555527058226e-05,
+      "loss": 0.6091,
+      "step": 726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95659143370095e-05,
+      "loss": 0.611,
+      "step": 727
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956527314696076e-05,
+      "loss": 0.6053,
+      "step": 728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9564631956912034e-05,
+      "loss": 0.689,
+      "step": 729
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95639907668633e-05,
+      "loss": 0.6261,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956334957681457e-05,
+      "loss": 0.7889,
+      "step": 731
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9562708386765836e-05,
+      "loss": 0.6633,
+      "step": 732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956206719671711e-05,
+      "loss": 0.5869,
+      "step": 733
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956142600666838e-05,
+      "loss": 0.6152,
+      "step": 734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956078481661965e-05,
+      "loss": 0.6178,
+      "step": 735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956014362657092e-05,
+      "loss": 0.6232,
+      "step": 736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955950243652219e-05,
+      "loss": 0.6795,
+      "step": 737
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955886124647346e-05,
+      "loss": 0.5793,
+      "step": 738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9558220056424724e-05,
+      "loss": 0.6005,
+      "step": 739
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9557578866375996e-05,
+      "loss": 0.6802,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955693767632726e-05,
+      "loss": 0.6088,
+      "step": 741
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955629648627854e-05,
+      "loss": 0.6884,
+      "step": 742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9555655296229804e-05,
+      "loss": 0.6526,
+      "step": 743
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9555014106181076e-05,
+      "loss": 0.5922,
+      "step": 744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955437291613235e-05,
+      "loss": 0.6442,
+      "step": 745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955373172608361e-05,
+      "loss": 0.6115,
+      "step": 746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9553090536034885e-05,
+      "loss": 0.635,
+      "step": 747
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955244934598615e-05,
+      "loss": 0.6548,
+      "step": 748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955180815593742e-05,
+      "loss": 0.6193,
+      "step": 749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955116696588869e-05,
+      "loss": 0.6401,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9550525775839965e-05,
+      "loss": 0.6261,
+      "step": 751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954988458579123e-05,
+      "loss": 0.6344,
+      "step": 752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95492433957425e-05,
+      "loss": 0.6219,
+      "step": 753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954860220569377e-05,
+      "loss": 0.6517,
+      "step": 754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954796101564504e-05,
+      "loss": 0.5861,
+      "step": 755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954731982559631e-05,
+      "loss": 0.6639,
+      "step": 756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9546678635547575e-05,
+      "loss": 0.684,
+      "step": 757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9546037445498846e-05,
+      "loss": 0.6374,
+      "step": 758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954539625545012e-05,
+      "loss": 0.6116,
+      "step": 759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954475506540139e-05,
+      "loss": 0.7166,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9544113875352655e-05,
+      "loss": 0.6642,
+      "step": 761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9543472685303926e-05,
+      "loss": 0.605,
+      "step": 762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95428314952552e-05,
+      "loss": 0.6058,
+      "step": 763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954219030520646e-05,
+      "loss": 0.6922,
+      "step": 764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9541549115157735e-05,
+      "loss": 0.6777,
+      "step": 765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9540907925109007e-05,
+      "loss": 0.6443,
+      "step": 766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954026673506028e-05,
+      "loss": 0.6693,
+      "step": 767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953962554501154e-05,
+      "loss": 0.6427,
+      "step": 768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9538984354962815e-05,
+      "loss": 0.6772,
+      "step": 769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953834316491408e-05,
+      "loss": 0.6608,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953770197486535e-05,
+      "loss": 0.6393,
+      "step": 771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953706078481662e-05,
+      "loss": 0.6128,
+      "step": 772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953641959476789e-05,
+      "loss": 0.6396,
+      "step": 773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953577840471916e-05,
+      "loss": 0.6322,
+      "step": 774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953513721467043e-05,
+      "loss": 0.6788,
+      "step": 775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95344960246217e-05,
+      "loss": 0.6351,
+      "step": 776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953385483457297e-05,
+      "loss": 0.5719,
+      "step": 777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953321364452424e-05,
+      "loss": 0.7127,
+      "step": 778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9532572454475505e-05,
+      "loss": 0.664,
+      "step": 779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953193126442678e-05,
+      "loss": 0.5958,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953129007437805e-05,
+      "loss": 0.709,
+      "step": 781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953064888432932e-05,
+      "loss": 0.6559,
+      "step": 782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953000769428059e-05,
+      "loss": 0.6339,
+      "step": 783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952936650423186e-05,
+      "loss": 0.6798,
+      "step": 784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952872531418313e-05,
+      "loss": 0.6199,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9528084124134393e-05,
+      "loss": 0.6618,
+      "step": 786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9527442934085665e-05,
+      "loss": 0.6302,
+      "step": 787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952680174403693e-05,
+      "loss": 0.6198,
+      "step": 788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95261605539882e-05,
+      "loss": 0.6493,
+      "step": 789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9525519363939473e-05,
+      "loss": 0.7005,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9524878173890745e-05,
+      "loss": 0.6302,
+      "step": 791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952423698384202e-05,
+      "loss": 0.6607,
+      "step": 792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952359579379328e-05,
+      "loss": 0.5616,
+      "step": 793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9522954603744554e-05,
+      "loss": 0.6194,
+      "step": 794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952231341369582e-05,
+      "loss": 0.6627,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952167222364709e-05,
+      "loss": 0.6716,
+      "step": 796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9521031033598355e-05,
+      "loss": 0.6097,
+      "step": 797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9520389843549634e-05,
+      "loss": 0.6457,
+      "step": 798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9519748653500905e-05,
+      "loss": 0.603,
+      "step": 799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951910746345217e-05,
+      "loss": 0.6673,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951846627340344e-05,
+      "loss": 0.5897,
+      "step": 801
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951782508335471e-05,
+      "loss": 0.7311,
+      "step": 802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951718389330598e-05,
+      "loss": 0.6246,
+      "step": 803
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9516542703257244e-05,
+      "loss": 0.6788,
+      "step": 804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9515901513208515e-05,
+      "loss": 0.6693,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951526032315979e-05,
+      "loss": 0.6324,
+      "step": 806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951461913311106e-05,
+      "loss": 0.5892,
+      "step": 807
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951397794306233e-05,
+      "loss": 0.6213,
+      "step": 808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9513336753013595e-05,
+      "loss": 0.6206,
+      "step": 809
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951269556296487e-05,
+      "loss": 0.6968,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951205437291613e-05,
+      "loss": 0.6917,
+      "step": 811
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9511413182867404e-05,
+      "loss": 0.5594,
+      "step": 812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951077199281867e-05,
+      "loss": 0.5473,
+      "step": 813
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951013080276995e-05,
+      "loss": 0.6289,
+      "step": 814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950948961272121e-05,
+      "loss": 0.6278,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9508848422672484e-05,
+      "loss": 0.6463,
+      "step": 816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9508207232623756e-05,
+      "loss": 0.5428,
+      "step": 817
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950756604257502e-05,
+      "loss": 0.6682,
+      "step": 818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950692485252629e-05,
+      "loss": 0.6326,
+      "step": 819
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950628366247756e-05,
+      "loss": 0.6056,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950564247242883e-05,
+      "loss": 0.6131,
+      "step": 821
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95050012823801e-05,
+      "loss": 0.6319,
+      "step": 822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950436009233137e-05,
+      "loss": 0.6517,
+      "step": 823
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950371890228264e-05,
+      "loss": 0.6375,
+      "step": 824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950307771223391e-05,
+      "loss": 0.5787,
+      "step": 825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950243652218518e-05,
+      "loss": 0.7115,
+      "step": 826
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9501795332136446e-05,
+      "loss": 0.6594,
+      "step": 827
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950115414208772e-05,
+      "loss": 0.657,
+      "step": 828
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950051295203898e-05,
+      "loss": 0.6241,
+      "step": 829
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9499871761990254e-05,
+      "loss": 0.6049,
+      "step": 830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9499230571941526e-05,
+      "loss": 0.6005,
+      "step": 831
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94985893818928e-05,
+      "loss": 0.5644,
+      "step": 832
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949794819184406e-05,
+      "loss": 0.6173,
+      "step": 833
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9497307001795334e-05,
+      "loss": 0.664,
+      "step": 834
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9496665811746606e-05,
+      "loss": 0.6406,
+      "step": 835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949602462169787e-05,
+      "loss": 0.6089,
+      "step": 836
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949538343164914e-05,
+      "loss": 0.6305,
+      "step": 837
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9494742241600414e-05,
+      "loss": 0.5968,
+      "step": 838
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9494101051551686e-05,
+      "loss": 0.6085,
+      "step": 839
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949345986150295e-05,
+      "loss": 0.6612,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949281867145422e-05,
+      "loss": 0.6401,
+      "step": 841
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949217748140549e-05,
+      "loss": 0.6339,
+      "step": 842
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949153629135676e-05,
+      "loss": 0.6294,
+      "step": 843
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949089510130803e-05,
+      "loss": 0.6112,
+      "step": 844
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9490253911259296e-05,
+      "loss": 0.5932,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948961272121057e-05,
+      "loss": 0.5756,
+      "step": 846
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948897153116184e-05,
+      "loss": 0.5911,
+      "step": 847
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948833034111311e-05,
+      "loss": 0.638,
+      "step": 848
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9487689151064376e-05,
+      "loss": 0.6916,
+      "step": 849
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948704796101565e-05,
+      "loss": 0.7131,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948640677096691e-05,
+      "loss": 0.6246,
+      "step": 851
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9485765580918184e-05,
+      "loss": 0.6692,
+      "step": 852
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9485124390869456e-05,
+      "loss": 0.7067,
+      "step": 853
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948448320082073e-05,
+      "loss": 0.633,
+      "step": 854
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9483842010772e-05,
+      "loss": 0.7198,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9483200820723264e-05,
+      "loss": 0.645,
+      "step": 856
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9482559630674536e-05,
+      "loss": 0.6665,
+      "step": 857
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94819184406258e-05,
+      "loss": 0.6515,
+      "step": 858
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948127725057707e-05,
+      "loss": 0.6729,
+      "step": 859
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948063606052834e-05,
+      "loss": 0.64,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947999487047961e-05,
+      "loss": 0.6764,
+      "step": 861
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947935368043088e-05,
+      "loss": 0.6701,
+      "step": 862
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947871249038215e-05,
+      "loss": 0.6919,
+      "step": 863
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9478071300333425e-05,
+      "loss": 0.6646,
+      "step": 864
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947743011028469e-05,
+      "loss": 0.6714,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947678892023596e-05,
+      "loss": 0.6658,
+      "step": 866
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9476147730187226e-05,
+      "loss": 0.6834,
+      "step": 867
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94755065401385e-05,
+      "loss": 0.6594,
+      "step": 868
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947486535008976e-05,
+      "loss": 0.6667,
+      "step": 869
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947422416004104e-05,
+      "loss": 0.6677,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947358296999231e-05,
+      "loss": 0.629,
+      "step": 871
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947294177994358e-05,
+      "loss": 0.6192,
+      "step": 872
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947230058989485e-05,
+      "loss": 0.5925,
+      "step": 873
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9471659399846115e-05,
+      "loss": 0.6705,
+      "step": 874
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9471018209797386e-05,
+      "loss": 0.5997,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947037701974865e-05,
+      "loss": 0.6135,
+      "step": 876
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946973582969992e-05,
+      "loss": 0.6076,
+      "step": 877
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9469094639651195e-05,
+      "loss": 0.5946,
+      "step": 878
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9468453449602467e-05,
+      "loss": 0.6681,
+      "step": 879
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946781225955374e-05,
+      "loss": 0.6808,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9467171069505e-05,
+      "loss": 0.6558,
+      "step": 881
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9466529879456275e-05,
+      "loss": 0.6518,
+      "step": 882
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946588868940754e-05,
+      "loss": 0.6226,
+      "step": 883
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946524749935881e-05,
+      "loss": 0.6351,
+      "step": 884
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9464606309310077e-05,
+      "loss": 0.6952,
+      "step": 885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9463965119261355e-05,
+      "loss": 0.583,
+      "step": 886
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946332392921262e-05,
+      "loss": 0.5459,
+      "step": 887
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946268273916389e-05,
+      "loss": 0.6808,
+      "step": 888
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946204154911516e-05,
+      "loss": 0.6128,
+      "step": 889
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946140035906643e-05,
+      "loss": 0.6721,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94607591690177e-05,
+      "loss": 0.6236,
+      "step": 891
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9460117978968965e-05,
+      "loss": 0.6064,
+      "step": 892
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945947678892024e-05,
+      "loss": 0.6573,
+      "step": 893
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945883559887151e-05,
+      "loss": 0.7008,
+      "step": 894
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945819440882278e-05,
+      "loss": 0.6129,
+      "step": 895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9457553218774045e-05,
+      "loss": 0.5817,
+      "step": 896
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945691202872532e-05,
+      "loss": 0.632,
+      "step": 897
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945627083867659e-05,
+      "loss": 0.6305,
+      "step": 898
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9455629648627853e-05,
+      "loss": 0.6501,
+      "step": 899
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9454988458579125e-05,
+      "loss": 0.6227,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945434726853039e-05,
+      "loss": 0.5778,
+      "step": 901
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945370607848167e-05,
+      "loss": 0.6382,
+      "step": 902
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9453064888432934e-05,
+      "loss": 0.6131,
+      "step": 903
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9452423698384205e-05,
+      "loss": 0.7437,
+      "step": 904
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945178250833548e-05,
+      "loss": 0.7296,
+      "step": 905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945114131828674e-05,
+      "loss": 0.6221,
+      "step": 906
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9450500128238014e-05,
+      "loss": 0.6076,
+      "step": 907
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.944985893818928e-05,
+      "loss": 0.628,
+      "step": 908
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.944921774814055e-05,
+      "loss": 0.6738,
+      "step": 909
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.944857655809182e-05,
+      "loss": 0.6696,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9447935368043094e-05,
+      "loss": 0.5837,
+      "step": 911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944729417799436e-05,
+      "loss": 0.6761,
+      "step": 912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944665298794563e-05,
+      "loss": 0.6168,
+      "step": 913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94460117978969e-05,
+      "loss": 0.683,
+      "step": 914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944537060784817e-05,
+      "loss": 0.5655,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944472941779944e-05,
+      "loss": 0.5927,
+      "step": 916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9444088227750704e-05,
+      "loss": 0.649,
+      "step": 917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9443447037701975e-05,
+      "loss": 0.648,
+      "step": 918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944280584765325e-05,
+      "loss": 0.5429,
+      "step": 919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944216465760452e-05,
+      "loss": 0.7035,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9441523467555784e-05,
+      "loss": 0.6246,
+      "step": 921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9440882277507055e-05,
+      "loss": 0.6124,
+      "step": 922
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944024108745833e-05,
+      "loss": 0.6932,
+      "step": 923
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943959989740959e-05,
+      "loss": 0.6222,
+      "step": 924
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9438958707360864e-05,
+      "loss": 0.6933,
+      "step": 925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9438317517312136e-05,
+      "loss": 0.597,
+      "step": 926
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943767632726341e-05,
+      "loss": 0.6565,
+      "step": 927
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943703513721467e-05,
+      "loss": 0.6734,
+      "step": 928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9436393947165944e-05,
+      "loss": 0.628,
+      "step": 929
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943575275711721e-05,
+      "loss": 0.686,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943511156706848e-05,
+      "loss": 0.6623,
+      "step": 931
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943447037701975e-05,
+      "loss": 0.6719,
+      "step": 932
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943382918697102e-05,
+      "loss": 0.6355,
+      "step": 933
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943318799692229e-05,
+      "loss": 0.612,
+      "step": 934
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943254680687356e-05,
+      "loss": 0.6643,
+      "step": 935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943190561682483e-05,
+      "loss": 0.5757,
+      "step": 936
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94312644267761e-05,
+      "loss": 0.6696,
+      "step": 937
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943062323672737e-05,
+      "loss": 0.7139,
+      "step": 938
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9429982046678634e-05,
+      "loss": 0.6178,
+      "step": 939
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9429340856629906e-05,
+      "loss": 0.6183,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942869966658118e-05,
+      "loss": 0.6509,
+      "step": 941
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942805847653245e-05,
+      "loss": 0.5784,
+      "step": 942
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942741728648372e-05,
+      "loss": 0.5949,
+      "step": 943
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9426776096434986e-05,
+      "loss": 0.685,
+      "step": 944
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942613490638626e-05,
+      "loss": 0.6262,
+      "step": 945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942549371633752e-05,
+      "loss": 0.6199,
+      "step": 946
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9424852526288794e-05,
+      "loss": 0.6449,
+      "step": 947
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942421133624006e-05,
+      "loss": 0.6261,
+      "step": 948
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942357014619133e-05,
+      "loss": 0.5987,
+      "step": 949
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94229289561426e-05,
+      "loss": 0.6173,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9422287766093874e-05,
+      "loss": 0.6068,
+      "step": 951
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9421646576045146e-05,
+      "loss": 0.6703,
+      "step": 952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942100538599641e-05,
+      "loss": 0.6945,
+      "step": 953
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942036419594768e-05,
+      "loss": 0.5756,
+      "step": 954
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941972300589895e-05,
+      "loss": 0.6707,
+      "step": 955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941908181585022e-05,
+      "loss": 0.6794,
+      "step": 956
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9418440625801484e-05,
+      "loss": 0.6802,
+      "step": 957
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941779943575276e-05,
+      "loss": 0.6641,
+      "step": 958
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9417158245704034e-05,
+      "loss": 0.5789,
+      "step": 959
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94165170556553e-05,
+      "loss": 0.6074,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941587586560657e-05,
+      "loss": 0.6838,
+      "step": 961
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9415234675557836e-05,
+      "loss": 0.6742,
+      "step": 962
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941459348550911e-05,
+      "loss": 0.6201,
+      "step": 963
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941395229546037e-05,
+      "loss": 0.6545,
+      "step": 964
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9413311105411644e-05,
+      "loss": 0.6457,
+      "step": 965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9412669915362916e-05,
+      "loss": 0.6683,
+      "step": 966
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941202872531419e-05,
+      "loss": 0.5826,
+      "step": 967
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941138753526546e-05,
+      "loss": 0.6712,
+      "step": 968
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9410746345216725e-05,
+      "loss": 0.6487,
+      "step": 969
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9410105155167996e-05,
+      "loss": 0.6526,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940946396511926e-05,
+      "loss": 0.6315,
+      "step": 971
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940882277507053e-05,
+      "loss": 0.6124,
+      "step": 972
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94081815850218e-05,
+      "loss": 0.5969,
+      "step": 973
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9407540394973076e-05,
+      "loss": 0.7066,
+      "step": 974
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940689920492434e-05,
+      "loss": 0.5995,
+      "step": 975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940625801487561e-05,
+      "loss": 0.6133,
+      "step": 976
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9405616824826885e-05,
+      "loss": 0.5847,
+      "step": 977
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940497563477815e-05,
+      "loss": 0.6204,
+      "step": 978
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940433444472942e-05,
+      "loss": 0.6569,
+      "step": 979
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9403693254680686e-05,
+      "loss": 0.5952,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940305206463196e-05,
+      "loss": 0.6449,
+      "step": 981
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940241087458323e-05,
+      "loss": 0.6288,
+      "step": 982
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94017696845345e-05,
+      "loss": 0.7162,
+      "step": 983
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9401128494485766e-05,
+      "loss": 0.6493,
+      "step": 984
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940048730443704e-05,
+      "loss": 0.7039,
+      "step": 985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939984611438831e-05,
+      "loss": 0.6812,
+      "step": 986
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9399204924339575e-05,
+      "loss": 0.6324,
+      "step": 987
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9398563734290846e-05,
+      "loss": 0.6342,
+      "step": 988
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939792254424211e-05,
+      "loss": 0.5952,
+      "step": 989
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939728135419339e-05,
+      "loss": 0.6421,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9396640164144655e-05,
+      "loss": 0.6702,
+      "step": 991
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9395998974095927e-05,
+      "loss": 0.584,
+      "step": 992
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939535778404719e-05,
+      "loss": 0.5872,
+      "step": 993
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939471659399846e-05,
+      "loss": 0.7133,
+      "step": 994
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9394075403949735e-05,
+      "loss": 0.625,
+      "step": 995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9393434213901e-05,
+      "loss": 0.6517,
+      "step": 996
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939279302385227e-05,
+      "loss": 0.631,
+      "step": 997
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939215183380354e-05,
+      "loss": 0.6681,
+      "step": 998
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9391510643754815e-05,
+      "loss": 0.5949,
+      "step": 999
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939086945370608e-05,
+      "loss": 0.6614,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939022826365735e-05,
+      "loss": 0.6788,
+      "step": 1001
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938958707360862e-05,
+      "loss": 0.6285,
+      "step": 1002
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938894588355989e-05,
+      "loss": 0.5653,
+      "step": 1003
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938830469351116e-05,
+      "loss": 0.5902,
+      "step": 1004
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9387663503462425e-05,
+      "loss": 0.6269,
+      "step": 1005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93870223134137e-05,
+      "loss": 0.6283,
+      "step": 1006
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938638112336497e-05,
+      "loss": 0.678,
+      "step": 1007
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938573993331624e-05,
+      "loss": 0.6407,
+      "step": 1008
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9385098743267505e-05,
+      "loss": 0.6411,
+      "step": 1009
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938445755321878e-05,
+      "loss": 0.6333,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938381636317004e-05,
+      "loss": 0.6142,
+      "step": 1011
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9383175173121313e-05,
+      "loss": 0.7537,
+      "step": 1012
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9382533983072585e-05,
+      "loss": 0.6601,
+      "step": 1013
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938189279302386e-05,
+      "loss": 0.7013,
+      "step": 1014
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938125160297513e-05,
+      "loss": 0.5944,
+      "step": 1015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9380610412926394e-05,
+      "loss": 0.628,
+      "step": 1016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9379969222877665e-05,
+      "loss": 0.6064,
+      "step": 1017
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937932803282893e-05,
+      "loss": 0.6185,
+      "step": 1018
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93786868427802e-05,
+      "loss": 0.5976,
+      "step": 1019
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937804565273147e-05,
+      "loss": 0.6966,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937740446268274e-05,
+      "loss": 0.5636,
+      "step": 1021
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937676327263401e-05,
+      "loss": 0.5755,
+      "step": 1022
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937612208258528e-05,
+      "loss": 0.6425,
+      "step": 1023
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9375480892536554e-05,
+      "loss": 0.645,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937483970248782e-05,
+      "loss": 0.5782,
+      "step": 1025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937419851243909e-05,
+      "loss": 0.6828,
+      "step": 1026
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9373557322390355e-05,
+      "loss": 0.6423,
+      "step": 1027
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937291613234163e-05,
+      "loss": 0.6207,
+      "step": 1028
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93722749422929e-05,
+      "loss": 0.599,
+      "step": 1029
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937163375224417e-05,
+      "loss": 0.6785,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937099256219544e-05,
+      "loss": 0.5936,
+      "step": 1031
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937035137214671e-05,
+      "loss": 0.6905,
+      "step": 1032
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936971018209798e-05,
+      "loss": 0.6007,
+      "step": 1033
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9369068992049244e-05,
+      "loss": 0.6559,
+      "step": 1034
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9368427802000516e-05,
+      "loss": 0.6406,
+      "step": 1035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936778661195178e-05,
+      "loss": 0.6758,
+      "step": 1036
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936714542190305e-05,
+      "loss": 0.6325,
+      "step": 1037
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9366504231854324e-05,
+      "loss": 0.6318,
+      "step": 1038
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9365863041805596e-05,
+      "loss": 0.6581,
+      "step": 1039
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936522185175687e-05,
+      "loss": 0.6115,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936458066170813e-05,
+      "loss": 0.629,
+      "step": 1041
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9363939471659404e-05,
+      "loss": 0.6885,
+      "step": 1042
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936329828161067e-05,
+      "loss": 0.6346,
+      "step": 1043
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936265709156194e-05,
+      "loss": 0.6513,
+      "step": 1044
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9362015901513206e-05,
+      "loss": 0.5622,
+      "step": 1045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9361374711464484e-05,
+      "loss": 0.5957,
+      "step": 1046
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936073352141575e-05,
+      "loss": 0.6285,
+      "step": 1047
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936009233136702e-05,
+      "loss": 0.5611,
+      "step": 1048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935945114131829e-05,
+      "loss": 0.5589,
+      "step": 1049
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935880995126956e-05,
+      "loss": 0.5902,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935816876122083e-05,
+      "loss": 0.6467,
+      "step": 1051
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9357527571172094e-05,
+      "loss": 0.5789,
+      "step": 1052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9356886381123366e-05,
+      "loss": 0.6692,
+      "step": 1053
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935624519107464e-05,
+      "loss": 0.5411,
+      "step": 1054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935560400102591e-05,
+      "loss": 0.5624,
+      "step": 1055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9354962810977174e-05,
+      "loss": 0.7045,
+      "step": 1056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9354321620928446e-05,
+      "loss": 0.6405,
+      "step": 1057
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935368043087972e-05,
+      "loss": 0.6338,
+      "step": 1058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935303924083098e-05,
+      "loss": 0.6587,
+      "step": 1059
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9352398050782254e-05,
+      "loss": 0.6025,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935175686073352e-05,
+      "loss": 0.6361,
+      "step": 1061
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93511156706848e-05,
+      "loss": 0.6261,
+      "step": 1062
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935047448063606e-05,
+      "loss": 0.6134,
+      "step": 1063
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9349833290587334e-05,
+      "loss": 0.6238,
+      "step": 1064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93491921005386e-05,
+      "loss": 0.6982,
+      "step": 1065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934855091048987e-05,
+      "loss": 0.6934,
+      "step": 1066
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934790972044114e-05,
+      "loss": 0.6582,
+      "step": 1067
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934726853039241e-05,
+      "loss": 0.6006,
+      "step": 1068
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934662734034368e-05,
+      "loss": 0.6132,
+      "step": 1069
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934598615029495e-05,
+      "loss": 0.6555,
+      "step": 1070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934534496024622e-05,
+      "loss": 0.647,
+      "step": 1071
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934470377019749e-05,
+      "loss": 0.6107,
+      "step": 1072
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934406258014876e-05,
+      "loss": 0.6296,
+      "step": 1073
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9343421390100024e-05,
+      "loss": 0.6079,
+      "step": 1074
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9342780200051296e-05,
+      "loss": 0.6084,
+      "step": 1075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934213901000257e-05,
+      "loss": 0.5968,
+      "step": 1076
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934149781995383e-05,
+      "loss": 0.6041,
+      "step": 1077
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9340856629905104e-05,
+      "loss": 0.6348,
+      "step": 1078
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9340215439856376e-05,
+      "loss": 0.6837,
+      "step": 1079
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933957424980765e-05,
+      "loss": 0.528,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933893305975891e-05,
+      "loss": 0.5639,
+      "step": 1081
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9338291869710185e-05,
+      "loss": 0.6176,
+      "step": 1082
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933765067966145e-05,
+      "loss": 0.6413,
+      "step": 1083
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933700948961272e-05,
+      "loss": 0.6985,
+      "step": 1084
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933636829956399e-05,
+      "loss": 0.5917,
+      "step": 1085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9335727109515265e-05,
+      "loss": 0.5843,
+      "step": 1086
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9335085919466536e-05,
+      "loss": 0.6769,
+      "step": 1087
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93344447294178e-05,
+      "loss": 0.6577,
+      "step": 1088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933380353936907e-05,
+      "loss": 0.5663,
+      "step": 1089
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933316234932034e-05,
+      "loss": 0.6394,
+      "step": 1090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933252115927161e-05,
+      "loss": 0.6373,
+      "step": 1091
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933187996922288e-05,
+      "loss": 0.5672,
+      "step": 1092
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9331238779174146e-05,
+      "loss": 0.6452,
+      "step": 1093
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933059758912542e-05,
+      "loss": 0.6157,
+      "step": 1094
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932995639907669e-05,
+      "loss": 0.6502,
+      "step": 1095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932931520902796e-05,
+      "loss": 0.6534,
+      "step": 1096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9328674018979226e-05,
+      "loss": 0.6426,
+      "step": 1097
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93280328289305e-05,
+      "loss": 0.7128,
+      "step": 1098
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932739163888176e-05,
+      "loss": 0.6342,
+      "step": 1099
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9326750448833035e-05,
+      "loss": 0.6487,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9326109258784307e-05,
+      "loss": 0.6581,
+      "step": 1101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932546806873558e-05,
+      "loss": 0.6064,
+      "step": 1102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932482687868685e-05,
+      "loss": 0.653,
+      "step": 1103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9324185688638115e-05,
+      "loss": 0.6097,
+      "step": 1104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9323544498589387e-05,
+      "loss": 0.6136,
+      "step": 1105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932290330854065e-05,
+      "loss": 0.6213,
+      "step": 1106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932226211849192e-05,
+      "loss": 0.671,
+      "step": 1107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932162092844319e-05,
+      "loss": 0.6416,
+      "step": 1108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932097973839446e-05,
+      "loss": 0.6239,
+      "step": 1109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932033854834573e-05,
+      "loss": 0.6333,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9319697358297e-05,
+      "loss": 0.5954,
+      "step": 1111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9319056168248275e-05,
+      "loss": 0.6552,
+      "step": 1112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931841497819954e-05,
+      "loss": 0.6534,
+      "step": 1113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931777378815081e-05,
+      "loss": 0.651,
+      "step": 1114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931713259810208e-05,
+      "loss": 0.6218,
+      "step": 1115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931649140805335e-05,
+      "loss": 0.6085,
+      "step": 1116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931585021800461e-05,
+      "loss": 0.6014,
+      "step": 1117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931520902795589e-05,
+      "loss": 0.5863,
+      "step": 1118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9314567837907164e-05,
+      "loss": 0.635,
+      "step": 1119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931392664785843e-05,
+      "loss": 0.6832,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93132854578097e-05,
+      "loss": 0.656,
+      "step": 1121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9312644267760965e-05,
+      "loss": 0.7034,
+      "step": 1122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931200307771224e-05,
+      "loss": 0.6,
+      "step": 1123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93113618876635e-05,
+      "loss": 0.6522,
+      "step": 1124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9310720697614774e-05,
+      "loss": 0.5999,
+      "step": 1125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9310079507566045e-05,
+      "loss": 0.7038,
+      "step": 1126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930943831751732e-05,
+      "loss": 0.616,
+      "step": 1127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930879712746859e-05,
+      "loss": 0.6541,
+      "step": 1128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9308155937419854e-05,
+      "loss": 0.6738,
+      "step": 1129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9307514747371125e-05,
+      "loss": 0.6189,
+      "step": 1130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930687355732239e-05,
+      "loss": 0.6934,
+      "step": 1131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930623236727366e-05,
+      "loss": 0.623,
+      "step": 1132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930559117722493e-05,
+      "loss": 0.6856,
+      "step": 1133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9304949987176205e-05,
+      "loss": 0.6286,
+      "step": 1134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930430879712747e-05,
+      "loss": 0.6758,
+      "step": 1135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930366760707874e-05,
+      "loss": 0.6318,
+      "step": 1136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9303026417030014e-05,
+      "loss": 0.649,
+      "step": 1137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930238522698128e-05,
+      "loss": 0.6923,
+      "step": 1138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930174403693255e-05,
+      "loss": 0.6717,
+      "step": 1139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9301102846883815e-05,
+      "loss": 0.6491,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930046165683509e-05,
+      "loss": 0.6663,
+      "step": 1141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929982046678636e-05,
+      "loss": 0.6723,
+      "step": 1142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929917927673763e-05,
+      "loss": 0.5857,
+      "step": 1143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9298538086688895e-05,
+      "loss": 0.635,
+      "step": 1144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929789689664017e-05,
+      "loss": 0.5941,
+      "step": 1145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929725570659144e-05,
+      "loss": 0.6931,
+      "step": 1146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9296614516542704e-05,
+      "loss": 0.6619,
+      "step": 1147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9295973326493976e-05,
+      "loss": 0.6134,
+      "step": 1148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929533213644524e-05,
+      "loss": 0.6593,
+      "step": 1149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929469094639652e-05,
+      "loss": 0.6358,
+      "step": 1150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9294049756347784e-05,
+      "loss": 0.6118,
+      "step": 1151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9293408566299056e-05,
+      "loss": 0.6388,
+      "step": 1152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929276737625032e-05,
+      "loss": 0.6084,
+      "step": 1153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929212618620159e-05,
+      "loss": 0.6638,
+      "step": 1154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9291484996152864e-05,
+      "loss": 0.6678,
+      "step": 1155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929084380610413e-05,
+      "loss": 0.6467,
+      "step": 1156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92902026160554e-05,
+      "loss": 0.6449,
+      "step": 1157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928956142600667e-05,
+      "loss": 0.5494,
+      "step": 1158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9288920235957944e-05,
+      "loss": 0.6831,
+      "step": 1159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928827904590921e-05,
+      "loss": 0.672,
+      "step": 1160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928763785586048e-05,
+      "loss": 0.6409,
+      "step": 1161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9286996665811746e-05,
+      "loss": 0.5729,
+      "step": 1162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928635547576302e-05,
+      "loss": 0.6389,
+      "step": 1163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928571428571429e-05,
+      "loss": 0.6022,
+      "step": 1164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9285073095665554e-05,
+      "loss": 0.6287,
+      "step": 1165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9284431905616826e-05,
+      "loss": 0.693,
+      "step": 1166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92837907155681e-05,
+      "loss": 0.745,
+      "step": 1167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928314952551937e-05,
+      "loss": 0.635,
+      "step": 1168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9282508335470634e-05,
+      "loss": 0.5731,
+      "step": 1169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9281867145421906e-05,
+      "loss": 0.6408,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.928122595537317e-05,
+      "loss": 0.6551,
+      "step": 1171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.928058476532444e-05,
+      "loss": 0.639,
+      "step": 1172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9279943575275714e-05,
+      "loss": 0.6585,
+      "step": 1173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9279302385226986e-05,
+      "loss": 0.6089,
+      "step": 1174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927866119517826e-05,
+      "loss": 0.6626,
+      "step": 1175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927802000512952e-05,
+      "loss": 0.6871,
+      "step": 1176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9277378815080794e-05,
+      "loss": 0.6376,
+      "step": 1177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927673762503206e-05,
+      "loss": 0.6115,
+      "step": 1178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927609643498333e-05,
+      "loss": 0.6276,
+      "step": 1179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9275455244934596e-05,
+      "loss": 0.6519,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927481405488587e-05,
+      "loss": 0.6937,
+      "step": 1181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927417286483714e-05,
+      "loss": 0.5894,
+      "step": 1182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927353167478841e-05,
+      "loss": 0.6057,
+      "step": 1183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927289048473968e-05,
+      "loss": 0.5605,
+      "step": 1184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927224929469095e-05,
+      "loss": 0.6183,
+      "step": 1185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927160810464222e-05,
+      "loss": 0.6345,
+      "step": 1186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9270966914593484e-05,
+      "loss": 0.6178,
+      "step": 1187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9270325724544756e-05,
+      "loss": 0.6586,
+      "step": 1188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926968453449603e-05,
+      "loss": 0.705,
+      "step": 1189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92690433444473e-05,
+      "loss": 0.6513,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926840215439857e-05,
+      "loss": 0.5945,
+      "step": 1191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9267760964349836e-05,
+      "loss": 0.6358,
+      "step": 1192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926711977430111e-05,
+      "loss": 0.6028,
+      "step": 1193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926647858425237e-05,
+      "loss": 0.5721,
+      "step": 1194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9265837394203645e-05,
+      "loss": 0.5724,
+      "step": 1195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926519620415491e-05,
+      "loss": 0.6459,
+      "step": 1196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926455501410618e-05,
+      "loss": 0.7056,
+      "step": 1197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926391382405745e-05,
+      "loss": 0.6596,
+      "step": 1198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9263272634008725e-05,
+      "loss": 0.6196,
+      "step": 1199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9262631443959996e-05,
+      "loss": 0.7047,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926199025391126e-05,
+      "loss": 0.69,
+      "step": 1201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926134906386253e-05,
+      "loss": 0.6427,
+      "step": 1202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92607078738138e-05,
+      "loss": 0.734,
+      "step": 1203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926006668376507e-05,
+      "loss": 0.5632,
+      "step": 1204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9259425493716335e-05,
+      "loss": 0.6676,
+      "step": 1205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925878430366761e-05,
+      "loss": 0.5978,
+      "step": 1206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925814311361888e-05,
+      "loss": 0.6823,
+      "step": 1207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925750192357015e-05,
+      "loss": 0.6469,
+      "step": 1208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925686073352142e-05,
+      "loss": 0.6398,
+      "step": 1209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9256219543472686e-05,
+      "loss": 0.5871,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925557835342396e-05,
+      "loss": 0.6135,
+      "step": 1211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925493716337522e-05,
+      "loss": 0.5608,
+      "step": 1212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9254295973326495e-05,
+      "loss": 0.6551,
+      "step": 1213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9253654783277767e-05,
+      "loss": 0.6439,
+      "step": 1214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925301359322904e-05,
+      "loss": 0.5674,
+      "step": 1215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92523724031803e-05,
+      "loss": 0.588,
+      "step": 1216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9251731213131575e-05,
+      "loss": 0.5982,
+      "step": 1217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925109002308285e-05,
+      "loss": 0.6641,
+      "step": 1218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925044883303411e-05,
+      "loss": 0.5723,
+      "step": 1219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924980764298538e-05,
+      "loss": 0.658,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924916645293665e-05,
+      "loss": 0.6445,
+      "step": 1221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924852526288793e-05,
+      "loss": 0.5902,
+      "step": 1222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924788407283919e-05,
+      "loss": 0.6475,
+      "step": 1223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9247242882790463e-05,
+      "loss": 0.6893,
+      "step": 1224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924660169274173e-05,
+      "loss": 0.5852,
+      "step": 1225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9245960502693e-05,
+      "loss": 0.5909,
+      "step": 1226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924531931264427e-05,
+      "loss": 0.6027,
+      "step": 1227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924467812259554e-05,
+      "loss": 0.6328,
+      "step": 1228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924403693254681e-05,
+      "loss": 0.6724,
+      "step": 1229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924339574249808e-05,
+      "loss": 0.6299,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924275455244935e-05,
+      "loss": 0.6725,
+      "step": 1231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924211336240062e-05,
+      "loss": 0.5775,
+      "step": 1232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924147217235189e-05,
+      "loss": 0.6055,
+      "step": 1233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9240830982303153e-05,
+      "loss": 0.5909,
+      "step": 1234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9240189792254425e-05,
+      "loss": 0.591,
+      "step": 1235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92395486022057e-05,
+      "loss": 0.6522,
+      "step": 1236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923890741215696e-05,
+      "loss": 0.5668,
+      "step": 1237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9238266222108234e-05,
+      "loss": 0.5922,
+      "step": 1238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9237625032059505e-05,
+      "loss": 0.6538,
+      "step": 1239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923698384201078e-05,
+      "loss": 0.647,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923634265196204e-05,
+      "loss": 0.6113,
+      "step": 1241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9235701461913314e-05,
+      "loss": 0.6678,
+      "step": 1242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923506027186458e-05,
+      "loss": 0.6545,
+      "step": 1243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923441908181585e-05,
+      "loss": 0.5631,
+      "step": 1244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923377789176712e-05,
+      "loss": 0.6112,
+      "step": 1245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9233136701718394e-05,
+      "loss": 0.6783,
+      "step": 1246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9232495511669665e-05,
+      "loss": 0.5764,
+      "step": 1247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923185432162093e-05,
+      "loss": 0.592,
+      "step": 1248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92312131315722e-05,
+      "loss": 0.5283,
+      "step": 1249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923057194152347e-05,
+      "loss": 0.5451,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922993075147474e-05,
+      "loss": 0.6145,
+      "step": 1251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9229289561426004e-05,
+      "loss": 0.6176,
+      "step": 1252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9228648371377275e-05,
+      "loss": 0.5813,
+      "step": 1253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922800718132855e-05,
+      "loss": 0.6836,
+      "step": 1254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922736599127982e-05,
+      "loss": 0.6281,
+      "step": 1255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922672480123109e-05,
+      "loss": 0.623,
+      "step": 1256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9226083611182355e-05,
+      "loss": 0.6133,
+      "step": 1257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922544242113363e-05,
+      "loss": 0.7053,
+      "step": 1258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922480123108489e-05,
+      "loss": 0.6016,
+      "step": 1259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9224160041036164e-05,
+      "loss": 0.6518,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9223518850987436e-05,
+      "loss": 0.6177,
+      "step": 1261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922287766093871e-05,
+      "loss": 0.587,
+      "step": 1262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922223647088998e-05,
+      "loss": 0.5702,
+      "step": 1263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9221595280841244e-05,
+      "loss": 0.6374,
+      "step": 1264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9220954090792516e-05,
+      "loss": 0.6785,
+      "step": 1265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922031290074378e-05,
+      "loss": 0.6157,
+      "step": 1266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921967171069505e-05,
+      "loss": 0.6355,
+      "step": 1267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921903052064632e-05,
+      "loss": 0.6323,
+      "step": 1268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921838933059759e-05,
+      "loss": 0.5838,
+      "step": 1269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921774814054886e-05,
+      "loss": 0.6125,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921710695050013e-05,
+      "loss": 0.6357,
+      "step": 1271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9216465760451404e-05,
+      "loss": 0.6265,
+      "step": 1272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921582457040267e-05,
+      "loss": 0.7485,
+      "step": 1273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921518338035394e-05,
+      "loss": 0.6537,
+      "step": 1274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9214542190305206e-05,
+      "loss": 0.6388,
+      "step": 1275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921390100025648e-05,
+      "loss": 0.607,
+      "step": 1276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921325981020775e-05,
+      "loss": 0.5697,
+      "step": 1277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921261862015902e-05,
+      "loss": 0.6787,
+      "step": 1278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9211977430110286e-05,
+      "loss": 0.641,
+      "step": 1279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921133624006156e-05,
+      "loss": 0.5854,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921069505001283e-05,
+      "loss": 0.6134,
+      "step": 1281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9210053859964094e-05,
+      "loss": 0.6512,
+      "step": 1282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9209412669915366e-05,
+      "loss": 0.6616,
+      "step": 1283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920877147986663e-05,
+      "loss": 0.6472,
+      "step": 1284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92081302898179e-05,
+      "loss": 0.6214,
+      "step": 1285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9207489099769174e-05,
+      "loss": 0.6634,
+      "step": 1286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9206847909720446e-05,
+      "loss": 0.6288,
+      "step": 1287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920620671967172e-05,
+      "loss": 0.6036,
+      "step": 1288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920556552962298e-05,
+      "loss": 0.6674,
+      "step": 1289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9204924339574254e-05,
+      "loss": 0.5812,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920428314952552e-05,
+      "loss": 0.6144,
+      "step": 1291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920364195947679e-05,
+      "loss": 0.6196,
+      "step": 1292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9203000769428056e-05,
+      "loss": 0.6584,
+      "step": 1293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9202359579379334e-05,
+      "loss": 0.7248,
+      "step": 1294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92017183893306e-05,
+      "loss": 0.6769,
+      "step": 1295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920107719928187e-05,
+      "loss": 0.6086,
+      "step": 1296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920043600923314e-05,
+      "loss": 0.5609,
+      "step": 1297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919979481918441e-05,
+      "loss": 0.7002,
+      "step": 1298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919915362913568e-05,
+      "loss": 0.6245,
+      "step": 1299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9198512439086944e-05,
+      "loss": 0.6626,
+      "step": 1300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9197871249038216e-05,
+      "loss": 0.6618,
+      "step": 1301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919723005898949e-05,
+      "loss": 0.6489,
+      "step": 1302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919658886894076e-05,
+      "loss": 0.6577,
+      "step": 1303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9195947678892025e-05,
+      "loss": 0.659,
+      "step": 1304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9195306488843296e-05,
+      "loss": 0.6283,
+      "step": 1305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919466529879457e-05,
+      "loss": 0.7055,
+      "step": 1306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919402410874583e-05,
+      "loss": 0.647,
+      "step": 1307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9193382918697105e-05,
+      "loss": 0.622,
+      "step": 1308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919274172864837e-05,
+      "loss": 0.6822,
+      "step": 1309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919210053859965e-05,
+      "loss": 0.622,
+      "step": 1310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919145934855091e-05,
+      "loss": 0.6557,
+      "step": 1311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9190818158502185e-05,
+      "loss": 0.6302,
+      "step": 1312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919017696845345e-05,
+      "loss": 0.6532,
+      "step": 1313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918953577840472e-05,
+      "loss": 0.5987,
+      "step": 1314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918889458835599e-05,
+      "loss": 0.6085,
+      "step": 1315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918825339830726e-05,
+      "loss": 0.62,
+      "step": 1316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918761220825853e-05,
+      "loss": 0.6285,
+      "step": 1317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91869710182098e-05,
+      "loss": 0.6036,
+      "step": 1318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918632982816107e-05,
+      "loss": 0.5829,
+      "step": 1319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918568863811234e-05,
+      "loss": 0.5784,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918504744806361e-05,
+      "loss": 0.5855,
+      "step": 1321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9184406258014875e-05,
+      "loss": 0.636,
+      "step": 1322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9183765067966146e-05,
+      "loss": 0.6152,
+      "step": 1323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918312387791742e-05,
+      "loss": 0.6826,
+      "step": 1324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918248268786868e-05,
+      "loss": 0.6726,
+      "step": 1325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9181841497819955e-05,
+      "loss": 0.5277,
+      "step": 1326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9181200307771227e-05,
+      "loss": 0.5672,
+      "step": 1327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91805591177225e-05,
+      "loss": 0.5618,
+      "step": 1328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917991792767376e-05,
+      "loss": 0.6732,
+      "step": 1329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9179276737625035e-05,
+      "loss": 0.6796,
+      "step": 1330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91786355475763e-05,
+      "loss": 0.7051,
+      "step": 1331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917799435752757e-05,
+      "loss": 0.7058,
+      "step": 1332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917735316747884e-05,
+      "loss": 0.6149,
+      "step": 1333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9176711977430115e-05,
+      "loss": 0.6348,
+      "step": 1334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917607078738139e-05,
+      "loss": 0.6328,
+      "step": 1335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917542959733265e-05,
+      "loss": 0.6617,
+      "step": 1336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9174788407283923e-05,
+      "loss": 0.7131,
+      "step": 1337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917414721723519e-05,
+      "loss": 0.6842,
+      "step": 1338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917350602718646e-05,
+      "loss": 0.6455,
+      "step": 1339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9172864837137725e-05,
+      "loss": 0.6419,
+      "step": 1340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9172223647089e-05,
+      "loss": 0.6435,
+      "step": 1341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917158245704027e-05,
+      "loss": 0.5967,
+      "step": 1342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917094126699154e-05,
+      "loss": 0.6013,
+      "step": 1343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917030007694281e-05,
+      "loss": 0.6068,
+      "step": 1344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916965888689408e-05,
+      "loss": 0.6448,
+      "step": 1345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916901769684535e-05,
+      "loss": 0.601,
+      "step": 1346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9168376506796613e-05,
+      "loss": 0.617,
+      "step": 1347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9167735316747885e-05,
+      "loss": 0.6143,
+      "step": 1348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916709412669916e-05,
+      "loss": 0.6447,
+      "step": 1349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916645293665043e-05,
+      "loss": 0.603,
+      "step": 1350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91658117466017e-05,
+      "loss": 0.5887,
+      "step": 1351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9165170556552965e-05,
+      "loss": 0.6288,
+      "step": 1352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916452936650424e-05,
+      "loss": 0.6114,
+      "step": 1353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91638881764555e-05,
+      "loss": 0.5982,
+      "step": 1354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9163246986406774e-05,
+      "loss": 0.608,
+      "step": 1355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916260579635804e-05,
+      "loss": 0.631,
+      "step": 1356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916196460630931e-05,
+      "loss": 0.6505,
+      "step": 1357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916132341626058e-05,
+      "loss": 0.6624,
+      "step": 1358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9160682226211854e-05,
+      "loss": 0.6554,
+      "step": 1359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9160041036163125e-05,
+      "loss": 0.6429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915939984611439e-05,
+      "loss": 0.6293,
+      "step": 1361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915875865606566e-05,
+      "loss": 0.57,
+      "step": 1362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915811746601693e-05,
+      "loss": 0.6667,
+      "step": 1363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91574762759682e-05,
+      "loss": 0.5903,
+      "step": 1364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9156835085919464e-05,
+      "loss": 0.6208,
+      "step": 1365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915619389587074e-05,
+      "loss": 0.6872,
+      "step": 1366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915555270582201e-05,
+      "loss": 0.6062,
+      "step": 1367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915491151577328e-05,
+      "loss": 0.6548,
+      "step": 1368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915427032572455e-05,
+      "loss": 0.57,
+      "step": 1369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9153629135675816e-05,
+      "loss": 0.595,
+      "step": 1370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915298794562709e-05,
+      "loss": 0.586,
+      "step": 1371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915234675557835e-05,
+      "loss": 0.5957,
+      "step": 1372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9151705565529624e-05,
+      "loss": 0.6159,
+      "step": 1373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9151064375480896e-05,
+      "loss": 0.5759,
+      "step": 1374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915042318543217e-05,
+      "loss": 0.6943,
+      "step": 1375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914978199538343e-05,
+      "loss": 0.7015,
+      "step": 1376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9149140805334704e-05,
+      "loss": 0.5499,
+      "step": 1377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9148499615285976e-05,
+      "loss": 0.5437,
+      "step": 1378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914785842523724e-05,
+      "loss": 0.5856,
+      "step": 1379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914721723518851e-05,
+      "loss": 0.5968,
+      "step": 1380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914657604513978e-05,
+      "loss": 0.7115,
+      "step": 1381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9145934855091056e-05,
+      "loss": 0.6926,
+      "step": 1382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914529366504232e-05,
+      "loss": 0.5951,
+      "step": 1383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914465247499359e-05,
+      "loss": 0.6251,
+      "step": 1384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914401128494486e-05,
+      "loss": 0.6082,
+      "step": 1385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914337009489613e-05,
+      "loss": 0.6378,
+      "step": 1386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91427289048474e-05,
+      "loss": 0.6871,
+      "step": 1387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9142087714798666e-05,
+      "loss": 0.6214,
+      "step": 1388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914144652474994e-05,
+      "loss": 0.6322,
+      "step": 1389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914080533470121e-05,
+      "loss": 0.6372,
+      "step": 1390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914016414465248e-05,
+      "loss": 0.565,
+      "step": 1391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9139522954603746e-05,
+      "loss": 0.6815,
+      "step": 1392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913888176455502e-05,
+      "loss": 0.6708,
+      "step": 1393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913824057450628e-05,
+      "loss": 0.5909,
+      "step": 1394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9137599384457554e-05,
+      "loss": 0.6585,
+      "step": 1395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9136958194408826e-05,
+      "loss": 0.622,
+      "step": 1396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913631700436009e-05,
+      "loss": 0.6626,
+      "step": 1397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913567581431137e-05,
+      "loss": 0.628,
+      "step": 1398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9135034624262634e-05,
+      "loss": 0.6432,
+      "step": 1399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9134393434213906e-05,
+      "loss": 0.6189,
+      "step": 1400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913375224416517e-05,
+      "loss": 0.6339,
+      "step": 1401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913311105411644e-05,
+      "loss": 0.6565,
+      "step": 1402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913246986406771e-05,
+      "loss": 0.6029,
+      "step": 1403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913182867401898e-05,
+      "loss": 0.6654,
+      "step": 1404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913118748397025e-05,
+      "loss": 0.597,
+      "step": 1405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913054629392152e-05,
+      "loss": 0.6455,
+      "step": 1406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9129905103872795e-05,
+      "loss": 0.6752,
+      "step": 1407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912926391382406e-05,
+      "loss": 0.5796,
+      "step": 1408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912862272377533e-05,
+      "loss": 0.6874,
+      "step": 1409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9127981533726596e-05,
+      "loss": 0.6278,
+      "step": 1410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912734034367787e-05,
+      "loss": 0.5879,
+      "step": 1411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912669915362913e-05,
+      "loss": 0.6165,
+      "step": 1412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9126057963580404e-05,
+      "loss": 0.6514,
+      "step": 1413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9125416773531676e-05,
+      "loss": 0.6353,
+      "step": 1414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912477558348295e-05,
+      "loss": 0.6035,
+      "step": 1415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912413439343422e-05,
+      "loss": 0.6358,
+      "step": 1416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9123493203385485e-05,
+      "loss": 0.6145,
+      "step": 1417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9122852013336756e-05,
+      "loss": 0.6515,
+      "step": 1418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912221082328802e-05,
+      "loss": 0.6028,
+      "step": 1419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912156963323929e-05,
+      "loss": 0.6653,
+      "step": 1420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9120928443190565e-05,
+      "loss": 0.621,
+      "step": 1421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9120287253141836e-05,
+      "loss": 0.6455,
+      "step": 1422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911964606309311e-05,
+      "loss": 0.6475,
+      "step": 1423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911900487304437e-05,
+      "loss": 0.6359,
+      "step": 1424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9118363682995645e-05,
+      "loss": 0.6438,
+      "step": 1425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911772249294691e-05,
+      "loss": 0.6579,
+      "step": 1426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911708130289818e-05,
+      "loss": 0.5775,
+      "step": 1427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9116440112849446e-05,
+      "loss": 0.5902,
+      "step": 1428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911579892280072e-05,
+      "loss": 0.6136,
+      "step": 1429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911515773275199e-05,
+      "loss": 0.5889,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.911451654270326e-05,
+      "loss": 0.6635,
+      "step": 1431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.911387535265453e-05,
+      "loss": 0.5915,
+      "step": 1432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.91132341626058e-05,
+      "loss": 0.6214,
+      "step": 1433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.911259297255707e-05,
+      "loss": 0.6529,
+      "step": 1434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9111951782508335e-05,
+      "loss": 0.5913,
+      "step": 1435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9111310592459607e-05,
+      "loss": 0.6892,
+      "step": 1436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.911066940241088e-05,
+      "loss": 0.6186,
+      "step": 1437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.911002821236215e-05,
+      "loss": 0.6913,
+      "step": 1438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9109387022313415e-05,
+      "loss": 0.609,
+      "step": 1439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910874583226469e-05,
+      "loss": 0.6419,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910810464221596e-05,
+      "loss": 0.6323,
+      "step": 1441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910746345216722e-05,
+      "loss": 0.6244,
+      "step": 1442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9106822262118495e-05,
+      "loss": 0.637,
+      "step": 1443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910618107206976e-05,
+      "loss": 0.6946,
+      "step": 1444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910553988202103e-05,
+      "loss": 0.5923,
+      "step": 1445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.91048986919723e-05,
+      "loss": 0.6135,
+      "step": 1446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9104257501923575e-05,
+      "loss": 0.6177,
+      "step": 1447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910361631187484e-05,
+      "loss": 0.7206,
+      "step": 1448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910297512182611e-05,
+      "loss": 0.6419,
+      "step": 1449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9102333931777383e-05,
+      "loss": 0.6737,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910169274172865e-05,
+      "loss": 0.6063,
+      "step": 1451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910105155167992e-05,
+      "loss": 0.6647,
+      "step": 1452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9100410361631185e-05,
+      "loss": 0.6421,
+      "step": 1453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9099769171582464e-05,
+      "loss": 0.567,
+      "step": 1454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909912798153373e-05,
+      "loss": 0.5998,
+      "step": 1455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9098486791485e-05,
+      "loss": 0.5743,
+      "step": 1456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9097845601436265e-05,
+      "loss": 0.573,
+      "step": 1457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909720441138754e-05,
+      "loss": 0.5804,
+      "step": 1458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909656322133881e-05,
+      "loss": 0.6175,
+      "step": 1459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9095922031290074e-05,
+      "loss": 0.6146,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9095280841241345e-05,
+      "loss": 0.6359,
+      "step": 1461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909463965119262e-05,
+      "loss": 0.5769,
+      "step": 1462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909399846114389e-05,
+      "loss": 0.705,
+      "step": 1463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9093357271095154e-05,
+      "loss": 0.6578,
+      "step": 1464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9092716081046425e-05,
+      "loss": 0.5924,
+      "step": 1465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909207489099769e-05,
+      "loss": 0.5688,
+      "step": 1466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909143370094896e-05,
+      "loss": 0.585,
+      "step": 1467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9090792510900234e-05,
+      "loss": 0.5521,
+      "step": 1468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90901513208515e-05,
+      "loss": 0.62,
+      "step": 1469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908951013080278e-05,
+      "loss": 0.5435,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908886894075404e-05,
+      "loss": 0.6618,
+      "step": 1471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9088227750705314e-05,
+      "loss": 0.644,
+      "step": 1472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908758656065658e-05,
+      "loss": 0.6348,
+      "step": 1473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908694537060785e-05,
+      "loss": 0.5485,
+      "step": 1474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908630418055912e-05,
+      "loss": 0.6092,
+      "step": 1475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908566299051039e-05,
+      "loss": 0.5486,
+      "step": 1476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908502180046166e-05,
+      "loss": 0.5601,
+      "step": 1477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908438061041293e-05,
+      "loss": 0.7027,
+      "step": 1478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90837394203642e-05,
+      "loss": 0.7096,
+      "step": 1479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908309823031547e-05,
+      "loss": 0.7466,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908245704026674e-05,
+      "loss": 0.5599,
+      "step": 1481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9081815850218004e-05,
+      "loss": 0.6452,
+      "step": 1482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9081174660169276e-05,
+      "loss": 0.6201,
+      "step": 1483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908053347012055e-05,
+      "loss": 0.6781,
+      "step": 1484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907989228007181e-05,
+      "loss": 0.6665,
+      "step": 1485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9079251090023084e-05,
+      "loss": 0.6057,
+      "step": 1486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9078609899974356e-05,
+      "loss": 0.6086,
+      "step": 1487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907796870992563e-05,
+      "loss": 0.5958,
+      "step": 1488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907732751987689e-05,
+      "loss": 0.5324,
+      "step": 1489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9076686329828164e-05,
+      "loss": 0.7078,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907604513977943e-05,
+      "loss": 0.6593,
+      "step": 1491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90754039497307e-05,
+      "loss": 0.5809,
+      "step": 1492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907476275968197e-05,
+      "loss": 0.6722,
+      "step": 1493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9074121569633244e-05,
+      "loss": 0.6252,
+      "step": 1494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9073480379584516e-05,
+      "loss": 0.6895,
+      "step": 1495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907283918953578e-05,
+      "loss": 0.6776,
+      "step": 1496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907219799948705e-05,
+      "loss": 0.6233,
+      "step": 1497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907155680943832e-05,
+      "loss": 0.6746,
+      "step": 1498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907091561938959e-05,
+      "loss": 0.6212,
+      "step": 1499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9070274429340854e-05,
+      "loss": 0.6253,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9069633239292126e-05,
+      "loss": 0.6754,
+      "step": 1501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90689920492434e-05,
+      "loss": 0.6604,
+      "step": 1502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906835085919467e-05,
+      "loss": 0.6508,
+      "step": 1503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906770966914594e-05,
+      "loss": 0.627,
+      "step": 1504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9067068479097206e-05,
+      "loss": 0.6426,
+      "step": 1505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906642728904848e-05,
+      "loss": 0.6868,
+      "step": 1506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906578609899974e-05,
+      "loss": 0.6397,
+      "step": 1507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9065144908951014e-05,
+      "loss": 0.6338,
+      "step": 1508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9064503718902286e-05,
+      "loss": 0.6773,
+      "step": 1509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906386252885356e-05,
+      "loss": 0.6633,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906322133880483e-05,
+      "loss": 0.6243,
+      "step": 1511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9062580148756094e-05,
+      "loss": 0.5949,
+      "step": 1512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9061938958707366e-05,
+      "loss": 0.5645,
+      "step": 1513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906129776865863e-05,
+      "loss": 0.6767,
+      "step": 1514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90606565786099e-05,
+      "loss": 0.6083,
+      "step": 1515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906001538856117e-05,
+      "loss": 0.5816,
+      "step": 1516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905937419851244e-05,
+      "loss": 0.5912,
+      "step": 1517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905873300846371e-05,
+      "loss": 0.6104,
+      "step": 1518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905809181841498e-05,
+      "loss": 0.6706,
+      "step": 1519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9057450628366255e-05,
+      "loss": 0.6456,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905680943831752e-05,
+      "loss": 0.6168,
+      "step": 1521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905616824826879e-05,
+      "loss": 0.6606,
+      "step": 1522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9055527058220056e-05,
+      "loss": 0.6383,
+      "step": 1523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905488586817133e-05,
+      "loss": 0.6938,
+      "step": 1524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905424467812259e-05,
+      "loss": 0.6256,
+      "step": 1525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905360348807387e-05,
+      "loss": 0.6126,
+      "step": 1526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9052962298025136e-05,
+      "loss": 0.5779,
+      "step": 1527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905232110797641e-05,
+      "loss": 0.6518,
+      "step": 1528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905167991792768e-05,
+      "loss": 0.6155,
+      "step": 1529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9051038727878945e-05,
+      "loss": 0.5489,
+      "step": 1530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9050397537830216e-05,
+      "loss": 0.6167,
+      "step": 1531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904975634778148e-05,
+      "loss": 0.6366,
+      "step": 1532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904911515773275e-05,
+      "loss": 0.5476,
+      "step": 1533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9048473967684025e-05,
+      "loss": 0.6639,
+      "step": 1534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9047832777635296e-05,
+      "loss": 0.6305,
+      "step": 1535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904719158758656e-05,
+      "loss": 0.6105,
+      "step": 1536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904655039753783e-05,
+      "loss": 0.6404,
+      "step": 1537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9045909207489105e-05,
+      "loss": 0.5228,
+      "step": 1538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904526801744037e-05,
+      "loss": 0.615,
+      "step": 1539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904462682739164e-05,
+      "loss": 0.6747,
+      "step": 1540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9043985637342906e-05,
+      "loss": 0.6394,
+      "step": 1541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9043344447294185e-05,
+      "loss": 0.7077,
+      "step": 1542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904270325724545e-05,
+      "loss": 0.5882,
+      "step": 1543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904206206719672e-05,
+      "loss": 0.6305,
+      "step": 1544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9041420877147986e-05,
+      "loss": 0.5848,
+      "step": 1545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904077968709926e-05,
+      "loss": 0.7045,
+      "step": 1546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904013849705053e-05,
+      "loss": 0.6226,
+      "step": 1547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9039497307001795e-05,
+      "loss": 0.6427,
+      "step": 1548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9038856116953067e-05,
+      "loss": 0.5956,
+      "step": 1549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903821492690434e-05,
+      "loss": 0.6875,
+      "step": 1550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903757373685561e-05,
+      "loss": 0.6297,
+      "step": 1551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9036932546806875e-05,
+      "loss": 0.6821,
+      "step": 1552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903629135675815e-05,
+      "loss": 0.6136,
+      "step": 1553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903565016670941e-05,
+      "loss": 0.5458,
+      "step": 1554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903500897666068e-05,
+      "loss": 0.6467,
+      "step": 1555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9034367786611955e-05,
+      "loss": 0.6171,
+      "step": 1556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903372659656322e-05,
+      "loss": 0.6489,
+      "step": 1557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90330854065145e-05,
+      "loss": 0.6294,
+      "step": 1558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9032444216465763e-05,
+      "loss": 0.6763,
+      "step": 1559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9031803026417035e-05,
+      "loss": 0.6508,
+      "step": 1560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90311618363683e-05,
+      "loss": 0.6122,
+      "step": 1561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903052064631957e-05,
+      "loss": 0.5615,
+      "step": 1562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902987945627084e-05,
+      "loss": 0.6471,
+      "step": 1563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902923826622211e-05,
+      "loss": 0.6647,
+      "step": 1564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902859707617338e-05,
+      "loss": 0.6345,
+      "step": 1565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902795588612465e-05,
+      "loss": 0.5945,
+      "step": 1566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9027314696075924e-05,
+      "loss": 0.6018,
+      "step": 1567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902667350602719e-05,
+      "loss": 0.6818,
+      "step": 1568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902603231597846e-05,
+      "loss": 0.5939,
+      "step": 1569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9025391125929725e-05,
+      "loss": 0.6133,
+      "step": 1570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9024749935881e-05,
+      "loss": 0.7027,
+      "step": 1571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902410874583226e-05,
+      "loss": 0.5554,
+      "step": 1572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9023467555783534e-05,
+      "loss": 0.7075,
+      "step": 1573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9022826365734805e-05,
+      "loss": 0.6406,
+      "step": 1574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902218517568608e-05,
+      "loss": 0.6325,
+      "step": 1575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902154398563735e-05,
+      "loss": 0.6549,
+      "step": 1576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9020902795588614e-05,
+      "loss": 0.6225,
+      "step": 1577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9020261605539885e-05,
+      "loss": 0.6921,
+      "step": 1578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901962041549115e-05,
+      "loss": 0.6311,
+      "step": 1579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901897922544242e-05,
+      "loss": 0.6632,
+      "step": 1580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9018338035393694e-05,
+      "loss": 0.6971,
+      "step": 1581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9017696845344965e-05,
+      "loss": 0.7333,
+      "step": 1582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901705565529624e-05,
+      "loss": 0.6685,
+      "step": 1583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90164144652475e-05,
+      "loss": 0.6219,
+      "step": 1584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9015773275198774e-05,
+      "loss": 0.6196,
+      "step": 1585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901513208515004e-05,
+      "loss": 0.5533,
+      "step": 1586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901449089510131e-05,
+      "loss": 0.6547,
+      "step": 1587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9013849705052575e-05,
+      "loss": 0.5972,
+      "step": 1588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901320851500385e-05,
+      "loss": 0.5945,
+      "step": 1589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901256732495512e-05,
+      "loss": 0.589,
+      "step": 1590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901192613490639e-05,
+      "loss": 0.5729,
+      "step": 1591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901128494485766e-05,
+      "loss": 0.6187,
+      "step": 1592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901064375480893e-05,
+      "loss": 0.6351,
+      "step": 1593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90100025647602e-05,
+      "loss": 0.5812,
+      "step": 1594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9009361374711464e-05,
+      "loss": 0.677,
+      "step": 1595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9008720184662736e-05,
+      "loss": 0.5333,
+      "step": 1596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900807899461401e-05,
+      "loss": 0.6034,
+      "step": 1597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900743780456528e-05,
+      "loss": 0.6371,
+      "step": 1598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9006796614516544e-05,
+      "loss": 0.6656,
+      "step": 1599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9006155424467816e-05,
+      "loss": 0.5699,
+      "step": 1600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900551423441909e-05,
+      "loss": 0.5831,
+      "step": 1601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900487304437035e-05,
+      "loss": 0.7082,
+      "step": 1602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9004231854321624e-05,
+      "loss": 0.7421,
+      "step": 1603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900359066427289e-05,
+      "loss": 0.5927,
+      "step": 1604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900294947422416e-05,
+      "loss": 0.702,
+      "step": 1605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900230828417543e-05,
+      "loss": 0.5389,
+      "step": 1606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9001667094126704e-05,
+      "loss": 0.6463,
+      "step": 1607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900102590407797e-05,
+      "loss": 0.6512,
+      "step": 1608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900038471402924e-05,
+      "loss": 0.5983,
+      "step": 1609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899974352398051e-05,
+      "loss": 0.6095,
+      "step": 1610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899910233393178e-05,
+      "loss": 0.6714,
+      "step": 1611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899846114388305e-05,
+      "loss": 0.6435,
+      "step": 1612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8997819953834314e-05,
+      "loss": 0.6294,
+      "step": 1613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899717876378559e-05,
+      "loss": 0.6043,
+      "step": 1614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899653757373686e-05,
+      "loss": 0.6378,
+      "step": 1615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899589638368813e-05,
+      "loss": 0.5828,
+      "step": 1616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8995255193639394e-05,
+      "loss": 0.6478,
+      "step": 1617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8994614003590666e-05,
+      "loss": 0.6474,
+      "step": 1618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899397281354194e-05,
+      "loss": 0.6378,
+      "step": 1619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.89933316234932e-05,
+      "loss": 0.5968,
+      "step": 1620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8992690433444474e-05,
+      "loss": 0.6027,
+      "step": 1621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8992049243395746e-05,
+      "loss": 0.5646,
+      "step": 1622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899140805334702e-05,
+      "loss": 0.6123,
+      "step": 1623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899076686329828e-05,
+      "loss": 0.6467,
+      "step": 1624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8990125673249554e-05,
+      "loss": 0.5895,
+      "step": 1625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898948448320082e-05,
+      "loss": 0.61,
+      "step": 1626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898884329315209e-05,
+      "loss": 0.5703,
+      "step": 1627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898820210310336e-05,
+      "loss": 0.6541,
+      "step": 1628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898756091305463e-05,
+      "loss": 0.5745,
+      "step": 1629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8986919723005906e-05,
+      "loss": 0.6244,
+      "step": 1630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898627853295717e-05,
+      "loss": 0.5967,
+      "step": 1631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898563734290844e-05,
+      "loss": 0.5921,
+      "step": 1632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898499615285971e-05,
+      "loss": 0.5899,
+      "step": 1633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898435496281098e-05,
+      "loss": 0.7235,
+      "step": 1634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8983713772762244e-05,
+      "loss": 0.5555,
+      "step": 1635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8983072582713516e-05,
+      "loss": 0.6444,
+      "step": 1636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898243139266479e-05,
+      "loss": 0.6408,
+      "step": 1637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898179020261606e-05,
+      "loss": 0.6061,
+      "step": 1638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898114901256733e-05,
+      "loss": 0.5591,
+      "step": 1639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8980507822518596e-05,
+      "loss": 0.6799,
+      "step": 1640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897986663246987e-05,
+      "loss": 0.615,
+      "step": 1641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897922544242113e-05,
+      "loss": 0.5706,
+      "step": 1642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8978584252372405e-05,
+      "loss": 0.6768,
+      "step": 1643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897794306232367e-05,
+      "loss": 0.7311,
+      "step": 1644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897730187227494e-05,
+      "loss": 0.7007,
+      "step": 1645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897666068222621e-05,
+      "loss": 0.7004,
+      "step": 1646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8976019492177485e-05,
+      "loss": 0.588,
+      "step": 1647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8975378302128756e-05,
+      "loss": 0.6163,
+      "step": 1648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897473711208002e-05,
+      "loss": 0.6017,
+      "step": 1649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897409592203129e-05,
+      "loss": 0.6494,
+      "step": 1650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897345473198256e-05,
+      "loss": 0.6948,
+      "step": 1651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897281354193383e-05,
+      "loss": 0.6091,
+      "step": 1652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.89721723518851e-05,
+      "loss": 0.553,
+      "step": 1653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897153116183637e-05,
+      "loss": 0.609,
+      "step": 1654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8970889971787645e-05,
+      "loss": 0.5845,
+      "step": 1655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897024878173891e-05,
+      "loss": 0.6119,
+      "step": 1656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896960759169018e-05,
+      "loss": 0.6245,
+      "step": 1657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8968966401641447e-05,
+      "loss": 0.6395,
+      "step": 1658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896832521159272e-05,
+      "loss": 0.6125,
+      "step": 1659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896768402154398e-05,
+      "loss": 0.5929,
+      "step": 1660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8967042831495255e-05,
+      "loss": 0.6047,
+      "step": 1661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8966401641446527e-05,
+      "loss": 0.6173,
+      "step": 1662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.89657604513978e-05,
+      "loss": 0.5336,
+      "step": 1663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896511926134907e-05,
+      "loss": 0.5639,
+      "step": 1664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8964478071300335e-05,
+      "loss": 0.6123,
+      "step": 1665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896383688125161e-05,
+      "loss": 0.5762,
+      "step": 1666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896319569120287e-05,
+      "loss": 0.7567,
+      "step": 1667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896255450115414e-05,
+      "loss": 0.6202,
+      "step": 1668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8961913311105415e-05,
+      "loss": 0.6349,
+      "step": 1669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896127212105669e-05,
+      "loss": 0.6862,
+      "step": 1670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896063093100796e-05,
+      "loss": 0.6039,
+      "step": 1671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8959989740959223e-05,
+      "loss": 0.575,
+      "step": 1672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8959348550910495e-05,
+      "loss": 0.6856,
+      "step": 1673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895870736086176e-05,
+      "loss": 0.5616,
+      "step": 1674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895806617081303e-05,
+      "loss": 0.709,
+      "step": 1675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.89574249807643e-05,
+      "loss": 0.6329,
+      "step": 1676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895678379071557e-05,
+      "loss": 0.6855,
+      "step": 1677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895614260066684e-05,
+      "loss": 0.729,
+      "step": 1678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895550141061811e-05,
+      "loss": 0.6756,
+      "step": 1679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8954860220569384e-05,
+      "loss": 0.5932,
+      "step": 1680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895421903052065e-05,
+      "loss": 0.6026,
+      "step": 1681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895357784047192e-05,
+      "loss": 0.636,
+      "step": 1682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8952936650423185e-05,
+      "loss": 0.5471,
+      "step": 1683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895229546037446e-05,
+      "loss": 0.6388,
+      "step": 1684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895165427032573e-05,
+      "loss": 0.609,
+      "step": 1685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8951013080277e-05,
+      "loss": 0.6503,
+      "step": 1686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8950371890228265e-05,
+      "loss": 0.6376,
+      "step": 1687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.894973070017954e-05,
+      "loss": 0.5297,
+      "step": 1688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.894908951013081e-05,
+      "loss": 0.5657,
+      "step": 1689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8948448320082074e-05,
+      "loss": 0.6095,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8947807130033345e-05,
+      "loss": 0.633,
+      "step": 1691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894716593998461e-05,
+      "loss": 0.6075,
+      "step": 1692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894652474993588e-05,
+      "loss": 0.6343,
+      "step": 1693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8945883559887154e-05,
+      "loss": 0.5291,
+      "step": 1694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8945242369838425e-05,
+      "loss": 0.6001,
+      "step": 1695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894460117978969e-05,
+      "loss": 0.5824,
+      "step": 1696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894395998974096e-05,
+      "loss": 0.6392,
+      "step": 1697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8943318799692234e-05,
+      "loss": 0.5621,
+      "step": 1698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.89426776096435e-05,
+      "loss": 0.6695,
+      "step": 1699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894203641959477e-05,
+      "loss": 0.6112,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8941395229546035e-05,
+      "loss": 0.6534,
+      "step": 1701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8940754039497314e-05,
+      "loss": 0.6175,
+      "step": 1702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894011284944858e-05,
+      "loss": 0.7202,
+      "step": 1703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893947165939985e-05,
+      "loss": 0.5994,
+      "step": 1704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8938830469351116e-05,
+      "loss": 0.607,
+      "step": 1705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893818927930239e-05,
+      "loss": 0.6859,
+      "step": 1706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893754808925366e-05,
+      "loss": 0.6079,
+      "step": 1707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8936906899204924e-05,
+      "loss": 0.6224,
+      "step": 1708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8936265709156196e-05,
+      "loss": 0.6835,
+      "step": 1709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893562451910747e-05,
+      "loss": 0.637,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893498332905874e-05,
+      "loss": 0.5689,
+      "step": 1711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8934342139010004e-05,
+      "loss": 0.7275,
+      "step": 1712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8933700948961276e-05,
+      "loss": 0.6113,
+      "step": 1713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893305975891254e-05,
+      "loss": 0.5406,
+      "step": 1714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893241856886381e-05,
+      "loss": 0.5479,
+      "step": 1715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8931777378815084e-05,
+      "loss": 0.6797,
+      "step": 1716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893113618876635e-05,
+      "loss": 0.6219,
+      "step": 1717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893049499871763e-05,
+      "loss": 0.6358,
+      "step": 1718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892985380866889e-05,
+      "loss": 0.65,
+      "step": 1719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8929212618620164e-05,
+      "loss": 0.5935,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892857142857143e-05,
+      "loss": 0.6652,
+      "step": 1721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.89279302385227e-05,
+      "loss": 0.6178,
+      "step": 1722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8927289048473966e-05,
+      "loss": 0.6581,
+      "step": 1723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892664785842524e-05,
+      "loss": 0.6344,
+      "step": 1724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892600666837651e-05,
+      "loss": 0.6472,
+      "step": 1725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892536547832778e-05,
+      "loss": 0.6094,
+      "step": 1726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892472428827905e-05,
+      "loss": 0.6105,
+      "step": 1727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892408309823032e-05,
+      "loss": 0.7314,
+      "step": 1728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892344190818159e-05,
+      "loss": 0.6524,
+      "step": 1729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8922800718132854e-05,
+      "loss": 0.6256,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8922159528084126e-05,
+      "loss": 0.6266,
+      "step": 1731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892151833803539e-05,
+      "loss": 0.6736,
+      "step": 1732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892087714798666e-05,
+      "loss": 0.6636,
+      "step": 1733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8920235957937934e-05,
+      "loss": 0.6431,
+      "step": 1734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8919594767889206e-05,
+      "loss": 0.6361,
+      "step": 1735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891895357784048e-05,
+      "loss": 0.6558,
+      "step": 1736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891831238779174e-05,
+      "loss": 0.6108,
+      "step": 1737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8917671197743014e-05,
+      "loss": 0.5897,
+      "step": 1738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891703000769428e-05,
+      "loss": 0.6459,
+      "step": 1739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891638881764555e-05,
+      "loss": 0.6658,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891574762759682e-05,
+      "loss": 0.6137,
+      "step": 1741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8915106437548095e-05,
+      "loss": 0.6194,
+      "step": 1742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8914465247499366e-05,
+      "loss": 0.6039,
+      "step": 1743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891382405745063e-05,
+      "loss": 0.5716,
+      "step": 1744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.89131828674019e-05,
+      "loss": 0.6369,
+      "step": 1745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891254167735317e-05,
+      "loss": 0.5854,
+      "step": 1746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891190048730444e-05,
+      "loss": 0.6977,
+      "step": 1747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8911259297255704e-05,
+      "loss": 0.5641,
+      "step": 1748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8910618107206976e-05,
+      "loss": 0.5781,
+      "step": 1749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890997691715825e-05,
+      "loss": 0.571,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890933572710952e-05,
+      "loss": 0.7375,
+      "step": 1751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890869453706079e-05,
+      "loss": 0.573,
+      "step": 1752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8908053347012056e-05,
+      "loss": 0.6639,
+      "step": 1753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890741215696333e-05,
+      "loss": 0.6445,
+      "step": 1754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890677096691459e-05,
+      "loss": 0.6196,
+      "step": 1755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8906129776865865e-05,
+      "loss": 0.5837,
+      "step": 1756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8905488586817136e-05,
+      "loss": 0.6093,
+      "step": 1757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890484739676841e-05,
+      "loss": 0.6167,
+      "step": 1758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890420620671967e-05,
+      "loss": 0.5782,
+      "step": 1759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8903565016670945e-05,
+      "loss": 0.6434,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8902923826622216e-05,
+      "loss": 0.6465,
+      "step": 1761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890228263657348e-05,
+      "loss": 0.634,
+      "step": 1762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890164144652475e-05,
+      "loss": 0.6448,
+      "step": 1763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890100025647602e-05,
+      "loss": 0.5755,
+      "step": 1764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890035906642729e-05,
+      "loss": 0.5746,
+      "step": 1765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889971787637856e-05,
+      "loss": 0.5936,
+      "step": 1766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889907668632983e-05,
+      "loss": 0.6127,
+      "step": 1767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88984354962811e-05,
+      "loss": 0.6915,
+      "step": 1768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889779430623237e-05,
+      "loss": 0.5974,
+      "step": 1769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889715311618364e-05,
+      "loss": 0.5524,
+      "step": 1770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8896511926134907e-05,
+      "loss": 0.5161,
+      "step": 1771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889587073608618e-05,
+      "loss": 0.5786,
+      "step": 1772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889522954603744e-05,
+      "loss": 0.6008,
+      "step": 1773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889458835598872e-05,
+      "loss": 0.5791,
+      "step": 1774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889394716593999e-05,
+      "loss": 0.5647,
+      "step": 1775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889330597589126e-05,
+      "loss": 0.5631,
+      "step": 1776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889266478584252e-05,
+      "loss": 0.6482,
+      "step": 1777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8892023595793795e-05,
+      "loss": 0.5982,
+      "step": 1778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889138240574507e-05,
+      "loss": 0.5999,
+      "step": 1779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889074121569633e-05,
+      "loss": 0.7286,
+      "step": 1780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88901000256476e-05,
+      "loss": 0.678,
+      "step": 1781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8889458835598875e-05,
+      "loss": 0.6949,
+      "step": 1782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888881764555015e-05,
+      "loss": 0.5565,
+      "step": 1783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888817645550141e-05,
+      "loss": 0.5937,
+      "step": 1784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8887535265452683e-05,
+      "loss": 0.5403,
+      "step": 1785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888689407540395e-05,
+      "loss": 0.6042,
+      "step": 1786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888625288535522e-05,
+      "loss": 0.6813,
+      "step": 1787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888561169530649e-05,
+      "loss": 0.6041,
+      "step": 1788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888497050525776e-05,
+      "loss": 0.6038,
+      "step": 1789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8884329315209035e-05,
+      "loss": 0.5639,
+      "step": 1790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88836881251603e-05,
+      "loss": 0.5949,
+      "step": 1791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888304693511157e-05,
+      "loss": 0.6252,
+      "step": 1792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888240574506284e-05,
+      "loss": 0.5679,
+      "step": 1793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888176455501411e-05,
+      "loss": 0.6862,
+      "step": 1794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8881123364965374e-05,
+      "loss": 0.5273,
+      "step": 1795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8880482174916645e-05,
+      "loss": 0.5925,
+      "step": 1796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887984098486792e-05,
+      "loss": 0.672,
+      "step": 1797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887919979481919e-05,
+      "loss": 0.6115,
+      "step": 1798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887855860477046e-05,
+      "loss": 0.61,
+      "step": 1799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8877917414721725e-05,
+      "loss": 0.6693,
+      "step": 1800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8877276224673e-05,
+      "loss": 0.5672,
+      "step": 1801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887663503462426e-05,
+      "loss": 0.6129,
+      "step": 1802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8875993844575534e-05,
+      "loss": 0.6769,
+      "step": 1803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88753526545268e-05,
+      "loss": 0.6643,
+      "step": 1804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887471146447807e-05,
+      "loss": 0.5615,
+      "step": 1805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887407027442935e-05,
+      "loss": 0.6067,
+      "step": 1806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8873429084380614e-05,
+      "loss": 0.6046,
+      "step": 1807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8872787894331886e-05,
+      "loss": 0.6,
+      "step": 1808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887214670428315e-05,
+      "loss": 0.5904,
+      "step": 1809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887150551423442e-05,
+      "loss": 0.5915,
+      "step": 1810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887086432418569e-05,
+      "loss": 0.5364,
+      "step": 1811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887022313413696e-05,
+      "loss": 0.5867,
+      "step": 1812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886958194408823e-05,
+      "loss": 0.6545,
+      "step": 1813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88689407540395e-05,
+      "loss": 0.5791,
+      "step": 1814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8868299563990774e-05,
+      "loss": 0.5714,
+      "step": 1815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886765837394204e-05,
+      "loss": 0.635,
+      "step": 1816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886701718389331e-05,
+      "loss": 0.6492,
+      "step": 1817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8866375993844576e-05,
+      "loss": 0.5566,
+      "step": 1818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886573480379585e-05,
+      "loss": 0.6237,
+      "step": 1819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886509361374711e-05,
+      "loss": 0.6745,
+      "step": 1820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8864452423698384e-05,
+      "loss": 0.7586,
+      "step": 1821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8863811233649656e-05,
+      "loss": 0.6689,
+      "step": 1822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886317004360093e-05,
+      "loss": 0.5851,
+      "step": 1823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88625288535522e-05,
+      "loss": 0.6302,
+      "step": 1824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8861887663503464e-05,
+      "loss": 0.5105,
+      "step": 1825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8861246473454736e-05,
+      "loss": 0.6356,
+      "step": 1826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8860605283406e-05,
+      "loss": 0.6728,
+      "step": 1827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885996409335727e-05,
+      "loss": 0.585,
+      "step": 1828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8859322903308544e-05,
+      "loss": 0.6197,
+      "step": 1829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8858681713259816e-05,
+      "loss": 0.6467,
+      "step": 1830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885804052321108e-05,
+      "loss": 0.5473,
+      "step": 1831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885739933316235e-05,
+      "loss": 0.5542,
+      "step": 1832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8856758143113624e-05,
+      "loss": 0.6517,
+      "step": 1833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885611695306489e-05,
+      "loss": 0.6377,
+      "step": 1834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885547576301616e-05,
+      "loss": 0.6616,
+      "step": 1835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8854834572967426e-05,
+      "loss": 0.6461,
+      "step": 1836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88541933829187e-05,
+      "loss": 0.5778,
+      "step": 1837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885355219286997e-05,
+      "loss": 0.5698,
+      "step": 1838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885291100282124e-05,
+      "loss": 0.6905,
+      "step": 1839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8852269812772506e-05,
+      "loss": 0.6626,
+      "step": 1840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885162862272378e-05,
+      "loss": 0.7315,
+      "step": 1841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885098743267505e-05,
+      "loss": 0.6443,
+      "step": 1842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8850346242626314e-05,
+      "loss": 0.5984,
+      "step": 1843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8849705052577586e-05,
+      "loss": 0.6006,
+      "step": 1844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884906386252886e-05,
+      "loss": 0.5807,
+      "step": 1845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884842267248013e-05,
+      "loss": 0.6739,
+      "step": 1846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8847781482431394e-05,
+      "loss": 0.5694,
+      "step": 1847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8847140292382666e-05,
+      "loss": 0.6093,
+      "step": 1848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884649910233394e-05,
+      "loss": 0.5458,
+      "step": 1849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88458579122852e-05,
+      "loss": 0.5525,
+      "step": 1850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8845216722236474e-05,
+      "loss": 0.6082,
+      "step": 1851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884457553218774e-05,
+      "loss": 0.5432,
+      "step": 1852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884393434213901e-05,
+      "loss": 0.6128,
+      "step": 1853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884329315209028e-05,
+      "loss": 0.5693,
+      "step": 1854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8842651962041555e-05,
+      "loss": 0.5341,
+      "step": 1855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884201077199282e-05,
+      "loss": 0.5849,
+      "step": 1856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884136958194409e-05,
+      "loss": 0.6836,
+      "step": 1857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884072839189536e-05,
+      "loss": 0.6892,
+      "step": 1858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884008720184663e-05,
+      "loss": 0.6175,
+      "step": 1859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88394460117979e-05,
+      "loss": 0.5449,
+      "step": 1860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8838804821749165e-05,
+      "loss": 0.653,
+      "step": 1861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883816363170044e-05,
+      "loss": 0.5429,
+      "step": 1862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883752244165171e-05,
+      "loss": 0.6077,
+      "step": 1863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883688125160298e-05,
+      "loss": 0.6587,
+      "step": 1864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8836240061554245e-05,
+      "loss": 0.6552,
+      "step": 1865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8835598871505516e-05,
+      "loss": 0.7016,
+      "step": 1866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883495768145679e-05,
+      "loss": 0.8168,
+      "step": 1867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883431649140805e-05,
+      "loss": 0.6538,
+      "step": 1868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8833675301359325e-05,
+      "loss": 0.5996,
+      "step": 1869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8833034111310596e-05,
+      "loss": 0.7295,
+      "step": 1870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883239292126187e-05,
+      "loss": 0.6114,
+      "step": 1871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883175173121313e-05,
+      "loss": 0.5798,
+      "step": 1872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8831110541164405e-05,
+      "loss": 0.5492,
+      "step": 1873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883046935111567e-05,
+      "loss": 0.6663,
+      "step": 1874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882982816106694e-05,
+      "loss": 0.6703,
+      "step": 1875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882918697101821e-05,
+      "loss": 0.6454,
+      "step": 1876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882854578096948e-05,
+      "loss": 0.6421,
+      "step": 1877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8827904590920757e-05,
+      "loss": 0.568,
+      "step": 1878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882726340087202e-05,
+      "loss": 0.6382,
+      "step": 1879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882662221082329e-05,
+      "loss": 0.6171,
+      "step": 1880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882598102077456e-05,
+      "loss": 0.5733,
+      "step": 1881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882533983072583e-05,
+      "loss": 0.637,
+      "step": 1882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8824698640677095e-05,
+      "loss": 0.5999,
+      "step": 1883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8824057450628367e-05,
+      "loss": 0.7039,
+      "step": 1884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882341626057964e-05,
+      "loss": 0.6252,
+      "step": 1885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882277507053091e-05,
+      "loss": 0.6408,
+      "step": 1886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882213388048218e-05,
+      "loss": 0.6055,
+      "step": 1887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882149269043345e-05,
+      "loss": 0.6588,
+      "step": 1888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882085150038472e-05,
+      "loss": 0.5946,
+      "step": 1889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882021031033598e-05,
+      "loss": 0.6608,
+      "step": 1890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8819569120287255e-05,
+      "loss": 0.6791,
+      "step": 1891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881892793023852e-05,
+      "loss": 0.6113,
+      "step": 1892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881828674018979e-05,
+      "loss": 0.5347,
+      "step": 1893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8817645550141063e-05,
+      "loss": 0.5989,
+      "step": 1894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8817004360092335e-05,
+      "loss": 0.5798,
+      "step": 1895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881636317004361e-05,
+      "loss": 0.6573,
+      "step": 1896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881572197999487e-05,
+      "loss": 0.599,
+      "step": 1897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8815080789946143e-05,
+      "loss": 0.6671,
+      "step": 1898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881443959989741e-05,
+      "loss": 0.6578,
+      "step": 1899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881379840984868e-05,
+      "loss": 0.6153,
+      "step": 1900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881315721979995e-05,
+      "loss": 0.6497,
+      "step": 1901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8812516029751224e-05,
+      "loss": 0.6371,
+      "step": 1902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8811874839702495e-05,
+      "loss": 0.5744,
+      "step": 1903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881123364965376e-05,
+      "loss": 0.5692,
+      "step": 1904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881059245960503e-05,
+      "loss": 0.5892,
+      "step": 1905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88099512695563e-05,
+      "loss": 0.5309,
+      "step": 1906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880931007950757e-05,
+      "loss": 0.7456,
+      "step": 1907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8808668889458834e-05,
+      "loss": 0.6542,
+      "step": 1908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8808027699410105e-05,
+      "loss": 0.5666,
+      "step": 1909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880738650936138e-05,
+      "loss": 0.6197,
+      "step": 1910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880674531931265e-05,
+      "loss": 0.5662,
+      "step": 1911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880610412926392e-05,
+      "loss": 0.656,
+      "step": 1912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8805462939215185e-05,
+      "loss": 0.6348,
+      "step": 1913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880482174916646e-05,
+      "loss": 0.5762,
+      "step": 1914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880418055911772e-05,
+      "loss": 0.5963,
+      "step": 1915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8803539369068994e-05,
+      "loss": 0.7337,
+      "step": 1916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8802898179020265e-05,
+      "loss": 0.6742,
+      "step": 1917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880225698897154e-05,
+      "loss": 0.6299,
+      "step": 1918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88016157989228e-05,
+      "loss": 0.573,
+      "step": 1919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8800974608874074e-05,
+      "loss": 0.652,
+      "step": 1920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8800333418825346e-05,
+      "loss": 0.5417,
+      "step": 1921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879969222877661e-05,
+      "loss": 0.6001,
+      "step": 1922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879905103872788e-05,
+      "loss": 0.6307,
+      "step": 1923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879840984867915e-05,
+      "loss": 0.5833,
+      "step": 1924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879776865863042e-05,
+      "loss": 0.583,
+      "step": 1925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879712746858169e-05,
+      "loss": 0.565,
+      "step": 1926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879648627853296e-05,
+      "loss": 0.5528,
+      "step": 1927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879584508848423e-05,
+      "loss": 0.595,
+      "step": 1928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.87952038984355e-05,
+      "loss": 0.6222,
+      "step": 1929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879456270838677e-05,
+      "loss": 0.7301,
+      "step": 1930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8793921518338036e-05,
+      "loss": 0.6074,
+      "step": 1931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879328032828931e-05,
+      "loss": 0.6113,
+      "step": 1932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879263913824057e-05,
+      "loss": 0.5621,
+      "step": 1933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879199794819185e-05,
+      "loss": 0.5604,
+      "step": 1934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8791356758143116e-05,
+      "loss": 0.6102,
+      "step": 1935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879071556809439e-05,
+      "loss": 0.5538,
+      "step": 1936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879007437804565e-05,
+      "loss": 0.5715,
+      "step": 1937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8789433187996924e-05,
+      "loss": 0.5472,
+      "step": 1938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8788791997948196e-05,
+      "loss": 0.5779,
+      "step": 1939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878815080789946e-05,
+      "loss": 0.6519,
+      "step": 1940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878750961785073e-05,
+      "loss": 0.6301,
+      "step": 1941
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8786868427802004e-05,
+      "loss": 0.5443,
+      "step": 1942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8786227237753276e-05,
+      "loss": 0.6331,
+      "step": 1943
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878558604770454e-05,
+      "loss": 0.5716,
+      "step": 1944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878494485765581e-05,
+      "loss": 0.664,
+      "step": 1945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878430366760708e-05,
+      "loss": 0.5228,
+      "step": 1946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878366247755835e-05,
+      "loss": 0.5835,
+      "step": 1947
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878302128750962e-05,
+      "loss": 0.5758,
+      "step": 1948
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8782380097460886e-05,
+      "loss": 0.6526,
+      "step": 1949
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8781738907412164e-05,
+      "loss": 0.584,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.878109771736343e-05,
+      "loss": 0.5634,
+      "step": 1951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87804565273147e-05,
+      "loss": 0.5962,
+      "step": 1952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8779815337265966e-05,
+      "loss": 0.6422,
+      "step": 1953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877917414721724e-05,
+      "loss": 0.6658,
+      "step": 1954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87785329571685e-05,
+      "loss": 0.5725,
+      "step": 1955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8777891767119774e-05,
+      "loss": 0.6181,
+      "step": 1956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8777250577071046e-05,
+      "loss": 0.6111,
+      "step": 1957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877660938702232e-05,
+      "loss": 0.5611,
+      "step": 1958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877596819697359e-05,
+      "loss": 0.6343,
+      "step": 1959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8775327006924854e-05,
+      "loss": 0.6908,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8774685816876126e-05,
+      "loss": 0.595,
+      "step": 1961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877404462682739e-05,
+      "loss": 0.6135,
+      "step": 1962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877340343677866e-05,
+      "loss": 0.6177,
+      "step": 1963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877276224672993e-05,
+      "loss": 0.6507,
+      "step": 1964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87721210566812e-05,
+      "loss": 0.5859,
+      "step": 1965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877147986663248e-05,
+      "loss": 0.6024,
+      "step": 1966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877083867658374e-05,
+      "loss": 0.5839,
+      "step": 1967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8770197486535015e-05,
+      "loss": 0.6807,
+      "step": 1968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876955629648628e-05,
+      "loss": 0.6424,
+      "step": 1969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876891510643755e-05,
+      "loss": 0.5972,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8768273916388816e-05,
+      "loss": 0.6742,
+      "step": 1971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876763272634009e-05,
+      "loss": 0.5999,
+      "step": 1972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876699153629136e-05,
+      "loss": 0.542,
+      "step": 1973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876635034624263e-05,
+      "loss": 0.6847,
+      "step": 1974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87657091561939e-05,
+      "loss": 0.6067,
+      "step": 1975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876506796614517e-05,
+      "loss": 0.6505,
+      "step": 1976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876442677609644e-05,
+      "loss": 0.5813,
+      "step": 1977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8763785586047705e-05,
+      "loss": 0.5854,
+      "step": 1978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8763144395998976e-05,
+      "loss": 0.6133,
+      "step": 1979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876250320595024e-05,
+      "loss": 0.5854,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876186201590151e-05,
+      "loss": 0.6058,
+      "step": 1981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8761220825852785e-05,
+      "loss": 0.7788,
+      "step": 1982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8760579635804056e-05,
+      "loss": 0.6281,
+      "step": 1983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875993844575533e-05,
+      "loss": 0.6201,
+      "step": 1984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875929725570659e-05,
+      "loss": 0.605,
+      "step": 1985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8758656065657865e-05,
+      "loss": 0.702,
+      "step": 1986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875801487560913e-05,
+      "loss": 0.6272,
+      "step": 1987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87573736855604e-05,
+      "loss": 0.5936,
+      "step": 1988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875673249551167e-05,
+      "loss": 0.5849,
+      "step": 1989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8756091305462945e-05,
+      "loss": 0.585,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875545011541421e-05,
+      "loss": 0.5868,
+      "step": 1991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875480892536548e-05,
+      "loss": 0.5701,
+      "step": 1992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875416773531675e-05,
+      "loss": 0.6718,
+      "step": 1993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875352654526802e-05,
+      "loss": 0.6909,
+      "step": 1994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875288535521929e-05,
+      "loss": 0.5308,
+      "step": 1995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8752244165170555e-05,
+      "loss": 0.5639,
+      "step": 1996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8751602975121827e-05,
+      "loss": 0.6452,
+      "step": 1997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87509617850731e-05,
+      "loss": 0.6824,
+      "step": 1998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875032059502437e-05,
+      "loss": 0.6205,
+      "step": 1999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8749679404975635e-05,
+      "loss": 0.6132,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874903821492691e-05,
+      "loss": 0.5818,
+      "step": 2001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874839702487818e-05,
+      "loss": 0.6644,
+      "step": 2002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874775583482944e-05,
+      "loss": 0.6289,
+      "step": 2003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8747114644780715e-05,
+      "loss": 0.6919,
+      "step": 2004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874647345473199e-05,
+      "loss": 0.6568,
+      "step": 2005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874583226468326e-05,
+      "loss": 0.5291,
+      "step": 2006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8745191074634523e-05,
+      "loss": 0.5969,
+      "step": 2007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8744549884585795e-05,
+      "loss": 0.6067,
+      "step": 2008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874390869453706e-05,
+      "loss": 0.7035,
+      "step": 2009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874326750448833e-05,
+      "loss": 0.5688,
+      "step": 2010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8742626314439604e-05,
+      "loss": 0.63,
+      "step": 2011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874198512439087e-05,
+      "loss": 0.6587,
+      "step": 2012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874134393434214e-05,
+      "loss": 0.657,
+      "step": 2013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874070274429341e-05,
+      "loss": 0.6463,
+      "step": 2014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8740061554244684e-05,
+      "loss": 0.5687,
+      "step": 2015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873942036419595e-05,
+      "loss": 0.7023,
+      "step": 2016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873877917414722e-05,
+      "loss": 0.6479,
+      "step": 2017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8738137984098485e-05,
+      "loss": 0.6219,
+      "step": 2018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873749679404976e-05,
+      "loss": 0.6684,
+      "step": 2019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873685560400103e-05,
+      "loss": 0.5434,
+      "step": 2020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8736214413952294e-05,
+      "loss": 0.5572,
+      "step": 2021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873557322390357e-05,
+      "loss": 0.6421,
+      "step": 2022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873493203385484e-05,
+      "loss": 0.7051,
+      "step": 2023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873429084380611e-05,
+      "loss": 0.5695,
+      "step": 2024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8733649653757374e-05,
+      "loss": 0.616,
+      "step": 2025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8733008463708645e-05,
+      "loss": 0.6535,
+      "step": 2026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873236727365991e-05,
+      "loss": 0.6629,
+      "step": 2027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873172608361118e-05,
+      "loss": 0.6259,
+      "step": 2028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8731084893562454e-05,
+      "loss": 0.6701,
+      "step": 2029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8730443703513725e-05,
+      "loss": 0.5325,
+      "step": 2030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8729802513465e-05,
+      "loss": 0.6557,
+      "step": 2031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872916132341626e-05,
+      "loss": 0.5836,
+      "step": 2032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8728520133367534e-05,
+      "loss": 0.6297,
+      "step": 2033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87278789433188e-05,
+      "loss": 0.6296,
+      "step": 2034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872723775327007e-05,
+      "loss": 0.6093,
+      "step": 2035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8726596563221335e-05,
+      "loss": 0.5989,
+      "step": 2036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872595537317261e-05,
+      "loss": 0.5982,
+      "step": 2037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8725314183123886e-05,
+      "loss": 0.5456,
+      "step": 2038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872467299307515e-05,
+      "loss": 0.5665,
+      "step": 2039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872403180302642e-05,
+      "loss": 0.5833,
+      "step": 2040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872339061297769e-05,
+      "loss": 0.5769,
+      "step": 2041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872274942292896e-05,
+      "loss": 0.6448,
+      "step": 2042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8722108232880224e-05,
+      "loss": 0.6299,
+      "step": 2043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8721467042831496e-05,
+      "loss": 0.5972,
+      "step": 2044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872082585278277e-05,
+      "loss": 0.549,
+      "step": 2045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872018466273404e-05,
+      "loss": 0.6087,
+      "step": 2046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871954347268531e-05,
+      "loss": 0.5925,
+      "step": 2047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8718902282636576e-05,
+      "loss": 0.5374,
+      "step": 2048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871826109258785e-05,
+      "loss": 0.579,
+      "step": 2049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871761990253911e-05,
+      "loss": 0.5307,
+      "step": 2050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8716978712490384e-05,
+      "loss": 0.6172,
+      "step": 2051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871633752244165e-05,
+      "loss": 0.6372,
+      "step": 2052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871569633239292e-05,
+      "loss": 0.64,
+      "step": 2053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87150551423442e-05,
+      "loss": 0.7051,
+      "step": 2054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8714413952295464e-05,
+      "loss": 0.6517,
+      "step": 2055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8713772762246736e-05,
+      "loss": 0.6921,
+      "step": 2056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8713131572198e-05,
+      "loss": 0.5696,
+      "step": 2057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871249038214927e-05,
+      "loss": 0.6658,
+      "step": 2058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871184919210054e-05,
+      "loss": 0.5546,
+      "step": 2059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871120800205181e-05,
+      "loss": 0.5976,
+      "step": 2060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871056681200308e-05,
+      "loss": 0.6195,
+      "step": 2061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870992562195435e-05,
+      "loss": 0.6153,
+      "step": 2062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8709284431905624e-05,
+      "loss": 0.6809,
+      "step": 2063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870864324185689e-05,
+      "loss": 0.6701,
+      "step": 2064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870800205180816e-05,
+      "loss": 0.6335,
+      "step": 2065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8707360861759426e-05,
+      "loss": 0.6249,
+      "step": 2066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87067196717107e-05,
+      "loss": 0.6994,
+      "step": 2067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870607848166196e-05,
+      "loss": 0.6526,
+      "step": 2068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8705437291613234e-05,
+      "loss": 0.6362,
+      "step": 2069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8704796101564506e-05,
+      "loss": 0.5909,
+      "step": 2070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870415491151578e-05,
+      "loss": 0.5845,
+      "step": 2071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870351372146705e-05,
+      "loss": 0.651,
+      "step": 2072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8702872531418314e-05,
+      "loss": 0.6038,
+      "step": 2073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8702231341369586e-05,
+      "loss": 0.589,
+      "step": 2074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870159015132085e-05,
+      "loss": 0.5965,
+      "step": 2075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870094896127212e-05,
+      "loss": 0.5941,
+      "step": 2076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8700307771223395e-05,
+      "loss": 0.6019,
+      "step": 2077
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8699666581174666e-05,
+      "loss": 0.6082,
+      "step": 2078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869902539112593e-05,
+      "loss": 0.6446,
+      "step": 2079
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86983842010772e-05,
+      "loss": 0.6271,
+      "step": 2080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8697743011028475e-05,
+      "loss": 0.5734,
+      "step": 2081
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869710182097974e-05,
+      "loss": 0.5891,
+      "step": 2082
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869646063093101e-05,
+      "loss": 0.6945,
+      "step": 2083
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8695819440882276e-05,
+      "loss": 0.5222,
+      "step": 2084
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869517825083355e-05,
+      "loss": 0.6309,
+      "step": 2085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869453706078482e-05,
+      "loss": 0.6462,
+      "step": 2086
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869389587073609e-05,
+      "loss": 0.6294,
+      "step": 2087
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8693254680687356e-05,
+      "loss": 0.6094,
+      "step": 2088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869261349063863e-05,
+      "loss": 0.6052,
+      "step": 2089
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86919723005899e-05,
+      "loss": 0.5657,
+      "step": 2090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8691331110541165e-05,
+      "loss": 0.5609,
+      "step": 2091
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8690689920492436e-05,
+      "loss": 0.6405,
+      "step": 2092
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869004873044371e-05,
+      "loss": 0.706,
+      "step": 2093
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868940754039498e-05,
+      "loss": 0.6146,
+      "step": 2094
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8688766350346245e-05,
+      "loss": 0.6173,
+      "step": 2095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8688125160297516e-05,
+      "loss": 0.5499,
+      "step": 2096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868748397024878e-05,
+      "loss": 0.6117,
+      "step": 2097
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868684278020005e-05,
+      "loss": 0.6316,
+      "step": 2098
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8686201590151325e-05,
+      "loss": 0.6414,
+      "step": 2099
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868556040010259e-05,
+      "loss": 0.5267,
+      "step": 2100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868491921005386e-05,
+      "loss": 0.6117,
+      "step": 2101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868427802000513e-05,
+      "loss": 0.6,
+      "step": 2102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8683636829956405e-05,
+      "loss": 0.6201,
+      "step": 2103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868299563990767e-05,
+      "loss": 0.6936,
+      "step": 2104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868235444985894e-05,
+      "loss": 0.6606,
+      "step": 2105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8681713259810207e-05,
+      "loss": 0.6144,
+      "step": 2106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868107206976148e-05,
+      "loss": 0.618,
+      "step": 2107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868043087971275e-05,
+      "loss": 0.5864,
+      "step": 2108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8679789689664015e-05,
+      "loss": 0.5251,
+      "step": 2109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8679148499615293e-05,
+      "loss": 0.5311,
+      "step": 2110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867850730956656e-05,
+      "loss": 0.6838,
+      "step": 2111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867786611951783e-05,
+      "loss": 0.6209,
+      "step": 2112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8677224929469095e-05,
+      "loss": 0.664,
+      "step": 2113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867658373942037e-05,
+      "loss": 0.6296,
+      "step": 2114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867594254937163e-05,
+      "loss": 0.6189,
+      "step": 2115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86753013593229e-05,
+      "loss": 0.5662,
+      "step": 2116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8674660169274175e-05,
+      "loss": 0.6322,
+      "step": 2117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867401897922545e-05,
+      "loss": 0.6927,
+      "step": 2118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867337778917672e-05,
+      "loss": 0.6293,
+      "step": 2119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8672736599127983e-05,
+      "loss": 0.6963,
+      "step": 2120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8672095409079255e-05,
+      "loss": 0.6762,
+      "step": 2121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867145421903052e-05,
+      "loss": 0.6835,
+      "step": 2122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867081302898179e-05,
+      "loss": 0.6095,
+      "step": 2123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867017183893306e-05,
+      "loss": 0.5673,
+      "step": 2124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866953064888433e-05,
+      "loss": 0.5191,
+      "step": 2125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866888945883561e-05,
+      "loss": 0.6553,
+      "step": 2126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866824826878687e-05,
+      "loss": 0.6594,
+      "step": 2127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8667607078738144e-05,
+      "loss": 0.5812,
+      "step": 2128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866696588868941e-05,
+      "loss": 0.5523,
+      "step": 2129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866632469864068e-05,
+      "loss": 0.5867,
+      "step": 2130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8665683508591945e-05,
+      "loss": 0.6198,
+      "step": 2131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866504231854322e-05,
+      "loss": 0.654,
+      "step": 2132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866440112849449e-05,
+      "loss": 0.6445,
+      "step": 2133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866375993844576e-05,
+      "loss": 0.6036,
+      "step": 2134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866311874839703e-05,
+      "loss": 0.5906,
+      "step": 2135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86624775583483e-05,
+      "loss": 0.5831,
+      "step": 2136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866183636829957e-05,
+      "loss": 0.6036,
+      "step": 2137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8661195178250834e-05,
+      "loss": 0.6155,
+      "step": 2138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8660553988202105e-05,
+      "loss": 0.5803,
+      "step": 2139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865991279815337e-05,
+      "loss": 0.6205,
+      "step": 2140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865927160810464e-05,
+      "loss": 0.6661,
+      "step": 2141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8658630418055914e-05,
+      "loss": 0.5773,
+      "step": 2142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8657989228007186e-05,
+      "loss": 0.5538,
+      "step": 2143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865734803795846e-05,
+      "loss": 0.5649,
+      "step": 2144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865670684790972e-05,
+      "loss": 0.5696,
+      "step": 2145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8656065657860994e-05,
+      "loss": 0.5144,
+      "step": 2146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865542446781226e-05,
+      "loss": 0.5694,
+      "step": 2147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865478327776353e-05,
+      "loss": 0.5939,
+      "step": 2148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86541420877148e-05,
+      "loss": 0.5872,
+      "step": 2149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8653500897666074e-05,
+      "loss": 0.6207,
+      "step": 2150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865285970761734e-05,
+      "loss": 0.5524,
+      "step": 2151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865221851756861e-05,
+      "loss": 0.5911,
+      "step": 2152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865157732751988e-05,
+      "loss": 0.7259,
+      "step": 2153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865093613747115e-05,
+      "loss": 0.6452,
+      "step": 2154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.865029494742242e-05,
+      "loss": 0.4494,
+      "step": 2155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8649653757373684e-05,
+      "loss": 0.6007,
+      "step": 2156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8649012567324956e-05,
+      "loss": 0.5753,
+      "step": 2157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864837137727623e-05,
+      "loss": 0.6679,
+      "step": 2158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86477301872275e-05,
+      "loss": 0.6274,
+      "step": 2159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8647088997178764e-05,
+      "loss": 0.6741,
+      "step": 2160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8646447807130036e-05,
+      "loss": 0.6151,
+      "step": 2161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864580661708131e-05,
+      "loss": 0.6264,
+      "step": 2162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864516542703257e-05,
+      "loss": 0.6311,
+      "step": 2163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8644524236983844e-05,
+      "loss": 0.6161,
+      "step": 2164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8643883046935116e-05,
+      "loss": 0.5773,
+      "step": 2165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864324185688639e-05,
+      "loss": 0.6097,
+      "step": 2166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864260066683765e-05,
+      "loss": 0.5846,
+      "step": 2167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8641959476788924e-05,
+      "loss": 0.6785,
+      "step": 2168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864131828674019e-05,
+      "loss": 0.5607,
+      "step": 2169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864067709669146e-05,
+      "loss": 0.6124,
+      "step": 2170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864003590664273e-05,
+      "loss": 0.6033,
+      "step": 2171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8639394716594e-05,
+      "loss": 0.5021,
+      "step": 2172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.863875352654527e-05,
+      "loss": 0.5789,
+      "step": 2173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.863811233649654e-05,
+      "loss": 0.6078,
+      "step": 2174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.863747114644781e-05,
+      "loss": 0.5639,
+      "step": 2175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.863682995639908e-05,
+      "loss": 0.6212,
+      "step": 2176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.863618876635035e-05,
+      "loss": 0.5788,
+      "step": 2177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8635547576301614e-05,
+      "loss": 0.5021,
+      "step": 2178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8634906386252886e-05,
+      "loss": 0.5758,
+      "step": 2179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.863426519620416e-05,
+      "loss": 0.6485,
+      "step": 2180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.863362400615542e-05,
+      "loss": 0.6013,
+      "step": 2181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86329828161067e-05,
+      "loss": 0.656,
+      "step": 2182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8632341626057966e-05,
+      "loss": 0.6913,
+      "step": 2183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.863170043600924e-05,
+      "loss": 0.6421,
+      "step": 2184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86310592459605e-05,
+      "loss": 0.6415,
+      "step": 2185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8630418055911774e-05,
+      "loss": 0.5955,
+      "step": 2186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862977686586304e-05,
+      "loss": 0.5389,
+      "step": 2187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862913567581431e-05,
+      "loss": 0.7388,
+      "step": 2188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862849448576558e-05,
+      "loss": 0.614,
+      "step": 2189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8627853295716855e-05,
+      "loss": 0.6931,
+      "step": 2190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8627212105668126e-05,
+      "loss": 0.5913,
+      "step": 2191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862657091561939e-05,
+      "loss": 0.6201,
+      "step": 2192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862592972557066e-05,
+      "loss": 0.5762,
+      "step": 2193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862528853552193e-05,
+      "loss": 0.5514,
+      "step": 2194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86246473454732e-05,
+      "loss": 0.6147,
+      "step": 2195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8624006155424465e-05,
+      "loss": 0.5656,
+      "step": 2196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8623364965375736e-05,
+      "loss": 0.6131,
+      "step": 2197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8622723775327015e-05,
+      "loss": 0.649,
+      "step": 2198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862208258527828e-05,
+      "loss": 0.6293,
+      "step": 2199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862144139522955e-05,
+      "loss": 0.5807,
+      "step": 2200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8620800205180816e-05,
+      "loss": 0.675,
+      "step": 2201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.862015901513209e-05,
+      "loss": 0.5949,
+      "step": 2202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.861951782508335e-05,
+      "loss": 0.6271,
+      "step": 2203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8618876635034625e-05,
+      "loss": 0.6587,
+      "step": 2204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8618235444985896e-05,
+      "loss": 0.635,
+      "step": 2205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.861759425493717e-05,
+      "loss": 0.6052,
+      "step": 2206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.861695306488844e-05,
+      "loss": 0.6112,
+      "step": 2207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8616311874839705e-05,
+      "loss": 0.6892,
+      "step": 2208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8615670684790977e-05,
+      "loss": 0.5682,
+      "step": 2209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.861502949474224e-05,
+      "loss": 0.6415,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861438830469351e-05,
+      "loss": 0.5642,
+      "step": 2211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861374711464478e-05,
+      "loss": 0.6398,
+      "step": 2212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861310592459605e-05,
+      "loss": 0.6433,
+      "step": 2213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861246473454732e-05,
+      "loss": 0.6022,
+      "step": 2214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861182354449859e-05,
+      "loss": 0.7042,
+      "step": 2215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8611182354449865e-05,
+      "loss": 0.6465,
+      "step": 2216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861054116440113e-05,
+      "loss": 0.6248,
+      "step": 2217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.86098999743524e-05,
+      "loss": 0.687,
+      "step": 2218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8609258784303667e-05,
+      "loss": 0.6054,
+      "step": 2219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860861759425494e-05,
+      "loss": 0.6891,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860797640420621e-05,
+      "loss": 0.647,
+      "step": 2221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860733521415748e-05,
+      "loss": 0.6693,
+      "step": 2222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860669402410875e-05,
+      "loss": 0.6419,
+      "step": 2223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860605283406002e-05,
+      "loss": 0.6725,
+      "step": 2224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860541164401129e-05,
+      "loss": 0.5911,
+      "step": 2225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8604770453962555e-05,
+      "loss": 0.6126,
+      "step": 2226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860412926391383e-05,
+      "loss": 0.5843,
+      "step": 2227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860348807386509e-05,
+      "loss": 0.6526,
+      "step": 2228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8602846883816363e-05,
+      "loss": 0.6444,
+      "step": 2229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8602205693767635e-05,
+      "loss": 0.6158,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860156450371891e-05,
+      "loss": 0.6544,
+      "step": 2231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860092331367018e-05,
+      "loss": 0.6796,
+      "step": 2232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8600282123621444e-05,
+      "loss": 0.7162,
+      "step": 2233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8599640933572715e-05,
+      "loss": 0.5953,
+      "step": 2234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859899974352398e-05,
+      "loss": 0.5041,
+      "step": 2235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859835855347525e-05,
+      "loss": 0.6489,
+      "step": 2236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8597717363426524e-05,
+      "loss": 0.6956,
+      "step": 2237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8597076173377795e-05,
+      "loss": 0.6327,
+      "step": 2238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859643498332906e-05,
+      "loss": 0.6763,
+      "step": 2239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859579379328033e-05,
+      "loss": 0.6409,
+      "step": 2240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8595152603231604e-05,
+      "loss": 0.5568,
+      "step": 2241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859451141318287e-05,
+      "loss": 0.5881,
+      "step": 2242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859387022313414e-05,
+      "loss": 0.63,
+      "step": 2243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8593229033085405e-05,
+      "loss": 0.6065,
+      "step": 2244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859258784303668e-05,
+      "loss": 0.6482,
+      "step": 2245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859194665298795e-05,
+      "loss": 0.599,
+      "step": 2246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859130546293922e-05,
+      "loss": 0.5569,
+      "step": 2247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8590664272890485e-05,
+      "loss": 0.6408,
+      "step": 2248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859002308284176e-05,
+      "loss": 0.6984,
+      "step": 2249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858938189279303e-05,
+      "loss": 0.6598,
+      "step": 2250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8588740702744294e-05,
+      "loss": 0.5853,
+      "step": 2251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8588099512695565e-05,
+      "loss": 0.5966,
+      "step": 2252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858745832264684e-05,
+      "loss": 0.564,
+      "step": 2253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858681713259811e-05,
+      "loss": 0.6112,
+      "step": 2254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8586175942549374e-05,
+      "loss": 0.5761,
+      "step": 2255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8585534752500646e-05,
+      "loss": 0.5189,
+      "step": 2256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858489356245191e-05,
+      "loss": 0.5592,
+      "step": 2257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858425237240318e-05,
+      "loss": 0.6267,
+      "step": 2258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8583611182354454e-05,
+      "loss": 0.5668,
+      "step": 2259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858296999230572e-05,
+      "loss": 0.5899,
+      "step": 2260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858232880225699e-05,
+      "loss": 0.676,
+      "step": 2261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858168761220826e-05,
+      "loss": 0.7028,
+      "step": 2262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8581046422159534e-05,
+      "loss": 0.7725,
+      "step": 2263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85804052321108e-05,
+      "loss": 0.6076,
+      "step": 2264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857976404206207e-05,
+      "loss": 0.5816,
+      "step": 2265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8579122852013336e-05,
+      "loss": 0.4928,
+      "step": 2266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857848166196461e-05,
+      "loss": 0.6202,
+      "step": 2267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857784047191588e-05,
+      "loss": 0.5996,
+      "step": 2268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8577199281867144e-05,
+      "loss": 0.6668,
+      "step": 2269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857655809181842e-05,
+      "loss": 0.629,
+      "step": 2270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857591690176969e-05,
+      "loss": 0.6755,
+      "step": 2271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857527571172096e-05,
+      "loss": 0.7112,
+      "step": 2272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8574634521672224e-05,
+      "loss": 0.7101,
+      "step": 2273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8573993331623496e-05,
+      "loss": 0.6397,
+      "step": 2274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857335214157476e-05,
+      "loss": 0.5886,
+      "step": 2275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857271095152603e-05,
+      "loss": 0.621,
+      "step": 2276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8572069761477304e-05,
+      "loss": 0.6247,
+      "step": 2277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8571428571428576e-05,
+      "loss": 0.6524,
+      "step": 2278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857078738137985e-05,
+      "loss": 0.6474,
+      "step": 2279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857014619133111e-05,
+      "loss": 0.6488,
+      "step": 2280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8569505001282384e-05,
+      "loss": 0.6678,
+      "step": 2281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856886381123365e-05,
+      "loss": 0.6529,
+      "step": 2282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856822262118492e-05,
+      "loss": 0.6093,
+      "step": 2283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8567581431136186e-05,
+      "loss": 0.6129,
+      "step": 2284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856694024108746e-05,
+      "loss": 0.6316,
+      "step": 2285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8566299051038736e-05,
+      "loss": 0.6037,
+      "step": 2286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856565786099e-05,
+      "loss": 0.5908,
+      "step": 2287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856501667094127e-05,
+      "loss": 0.6138,
+      "step": 2288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856437548089254e-05,
+      "loss": 0.5909,
+      "step": 2289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856373429084381e-05,
+      "loss": 0.6259,
+      "step": 2290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8563093100795074e-05,
+      "loss": 0.6401,
+      "step": 2291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8562451910746346e-05,
+      "loss": 0.6609,
+      "step": 2292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856181072069762e-05,
+      "loss": 0.6468,
+      "step": 2293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856116953064889e-05,
+      "loss": 0.6027,
+      "step": 2294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856052834060016e-05,
+      "loss": 0.6181,
+      "step": 2295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8559887150551426e-05,
+      "loss": 0.6634,
+      "step": 2296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85592459605027e-05,
+      "loss": 0.6874,
+      "step": 2297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855860477045396e-05,
+      "loss": 0.5787,
+      "step": 2298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8557963580405235e-05,
+      "loss": 0.5517,
+      "step": 2299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85573223903565e-05,
+      "loss": 0.6391,
+      "step": 2300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855668120030777e-05,
+      "loss": 0.6284,
+      "step": 2301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855604001025904e-05,
+      "loss": 0.5329,
+      "step": 2302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8555398820210315e-05,
+      "loss": 0.563,
+      "step": 2303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8554757630161586e-05,
+      "loss": 0.6569,
+      "step": 2304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855411644011285e-05,
+      "loss": 0.6351,
+      "step": 2305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855347525006412e-05,
+      "loss": 0.6396,
+      "step": 2306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855283406001539e-05,
+      "loss": 0.627,
+      "step": 2307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855219286996666e-05,
+      "loss": 0.5643,
+      "step": 2308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855155167991793e-05,
+      "loss": 0.6363,
+      "step": 2309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85509104898692e-05,
+      "loss": 0.6331,
+      "step": 2310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855026929982047e-05,
+      "loss": 0.659,
+      "step": 2311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854962810977174e-05,
+      "loss": 0.6275,
+      "step": 2312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854898691972301e-05,
+      "loss": 0.6737,
+      "step": 2313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8548345729674276e-05,
+      "loss": 0.6587,
+      "step": 2314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854770453962555e-05,
+      "loss": 0.6252,
+      "step": 2315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854706334957681e-05,
+      "loss": 0.5079,
+      "step": 2316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8546422159528085e-05,
+      "loss": 0.561,
+      "step": 2317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8545780969479356e-05,
+      "loss": 0.5993,
+      "step": 2318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854513977943063e-05,
+      "loss": 0.6644,
+      "step": 2319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854449858938189e-05,
+      "loss": 0.6057,
+      "step": 2320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8543857399333165e-05,
+      "loss": 0.5927,
+      "step": 2321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8543216209284437e-05,
+      "loss": 0.6378,
+      "step": 2322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85425750192357e-05,
+      "loss": 0.5755,
+      "step": 2323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854193382918697e-05,
+      "loss": 0.5502,
+      "step": 2324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8541292639138245e-05,
+      "loss": 0.6007,
+      "step": 2325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854065144908952e-05,
+      "loss": 0.6069,
+      "step": 2326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854001025904078e-05,
+      "loss": 0.5689,
+      "step": 2327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853936906899205e-05,
+      "loss": 0.6137,
+      "step": 2328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853872787894332e-05,
+      "loss": 0.5437,
+      "step": 2329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853808668889459e-05,
+      "loss": 0.5873,
+      "step": 2330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853744549884586e-05,
+      "loss": 0.6183,
+      "step": 2331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8536804308797127e-05,
+      "loss": 0.6845,
+      "step": 2332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85361631187484e-05,
+      "loss": 0.693,
+      "step": 2333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853552192869967e-05,
+      "loss": 0.7116,
+      "step": 2334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853488073865094e-05,
+      "loss": 0.5882,
+      "step": 2335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853423954860221e-05,
+      "loss": 0.5694,
+      "step": 2336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853359835855348e-05,
+      "loss": 0.5918,
+      "step": 2337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853295716850474e-05,
+      "loss": 0.6417,
+      "step": 2338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8532315978456015e-05,
+      "loss": 0.5841,
+      "step": 2339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853167478840729e-05,
+      "loss": 0.5638,
+      "step": 2340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853103359835856e-05,
+      "loss": 0.5541,
+      "step": 2341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853039240830983e-05,
+      "loss": 0.6391,
+      "step": 2342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8529751218261095e-05,
+      "loss": 0.6685,
+      "step": 2343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852911002821237e-05,
+      "loss": 0.5924,
+      "step": 2344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852846883816363e-05,
+      "loss": 0.6065,
+      "step": 2345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8527827648114904e-05,
+      "loss": 0.6033,
+      "step": 2346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852718645806617e-05,
+      "loss": 0.6347,
+      "step": 2347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852654526801744e-05,
+      "loss": 0.5403,
+      "step": 2348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852590407796871e-05,
+      "loss": 0.5924,
+      "step": 2349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8525262887919984e-05,
+      "loss": 0.6247,
+      "step": 2350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8524621697871255e-05,
+      "loss": 0.5713,
+      "step": 2351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852398050782252e-05,
+      "loss": 0.5204,
+      "step": 2352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852333931777379e-05,
+      "loss": 0.4871,
+      "step": 2353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852269812772506e-05,
+      "loss": 0.7046,
+      "step": 2354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.852205693767633e-05,
+      "loss": 0.6716,
+      "step": 2355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8521415747627594e-05,
+      "loss": 0.6924,
+      "step": 2356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8520774557578865e-05,
+      "loss": 0.5835,
+      "step": 2357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8520133367530144e-05,
+      "loss": 0.6427,
+      "step": 2358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851949217748141e-05,
+      "loss": 0.6313,
+      "step": 2359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851885098743268e-05,
+      "loss": 0.5891,
+      "step": 2360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8518209797383945e-05,
+      "loss": 0.6414,
+      "step": 2361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851756860733522e-05,
+      "loss": 0.5705,
+      "step": 2362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851692741728648e-05,
+      "loss": 0.6142,
+      "step": 2363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8516286227237754e-05,
+      "loss": 0.6857,
+      "step": 2364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8515645037189025e-05,
+      "loss": 0.6452,
+      "step": 2365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85150038471403e-05,
+      "loss": 0.5818,
+      "step": 2366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851436265709157e-05,
+      "loss": 0.5243,
+      "step": 2367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8513721467042834e-05,
+      "loss": 0.6333,
+      "step": 2368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8513080276994106e-05,
+      "loss": 0.5597,
+      "step": 2369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851243908694537e-05,
+      "loss": 0.7061,
+      "step": 2370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851179789689664e-05,
+      "loss": 0.6934,
+      "step": 2371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851115670684791e-05,
+      "loss": 0.6182,
+      "step": 2372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851051551679918e-05,
+      "loss": 0.6586,
+      "step": 2373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850987432675045e-05,
+      "loss": 0.5994,
+      "step": 2374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850923313670172e-05,
+      "loss": 0.6198,
+      "step": 2375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8508591946652994e-05,
+      "loss": 0.5751,
+      "step": 2376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850795075660426e-05,
+      "loss": 0.6235,
+      "step": 2377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850730956655553e-05,
+      "loss": 0.6116,
+      "step": 2378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8506668376506796e-05,
+      "loss": 0.6413,
+      "step": 2379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850602718645807e-05,
+      "loss": 0.5869,
+      "step": 2380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850538599640934e-05,
+      "loss": 0.6344,
+      "step": 2381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850474480636061e-05,
+      "loss": 0.6852,
+      "step": 2382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8504103616311876e-05,
+      "loss": 0.5876,
+      "step": 2383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850346242626315e-05,
+      "loss": 0.6559,
+      "step": 2384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850282123621442e-05,
+      "loss": 0.5929,
+      "step": 2385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8502180046165684e-05,
+      "loss": 0.6077,
+      "step": 2386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8501538856116956e-05,
+      "loss": 0.6322,
+      "step": 2387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850089766606822e-05,
+      "loss": 0.592,
+      "step": 2388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.850025647601949e-05,
+      "loss": 0.5373,
+      "step": 2389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8499615285970764e-05,
+      "loss": 0.7603,
+      "step": 2390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8498974095922036e-05,
+      "loss": 0.6687,
+      "step": 2391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.84983329058733e-05,
+      "loss": 0.5658,
+      "step": 2392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.849769171582457e-05,
+      "loss": 0.6713,
+      "step": 2393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8497050525775844e-05,
+      "loss": 0.4786,
+      "step": 2394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.849640933572711e-05,
+      "loss": 0.6855,
+      "step": 2395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.849576814567838e-05,
+      "loss": 0.612,
+      "step": 2396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.849512695562965e-05,
+      "loss": 0.5938,
+      "step": 2397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8494485765580924e-05,
+      "loss": 0.63,
+      "step": 2398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.849384457553219e-05,
+      "loss": 0.6266,
+      "step": 2399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.849320338548346e-05,
+      "loss": 0.63,
+      "step": 2400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8492562195434726e-05,
+      "loss": 0.5501,
+      "step": 2401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8491921005386e-05,
+      "loss": 0.581,
+      "step": 2402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.849127981533727e-05,
+      "loss": 0.6147,
+      "step": 2403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8490638625288534e-05,
+      "loss": 0.6283,
+      "step": 2404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8489997435239806e-05,
+      "loss": 0.6994,
+      "step": 2405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848935624519108e-05,
+      "loss": 0.6216,
+      "step": 2406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848871505514235e-05,
+      "loss": 0.6704,
+      "step": 2407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8488073865093614e-05,
+      "loss": 0.6755,
+      "step": 2408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8487432675044886e-05,
+      "loss": 0.5179,
+      "step": 2409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848679148499615e-05,
+      "loss": 0.5712,
+      "step": 2410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848615029494742e-05,
+      "loss": 0.6466,
+      "step": 2411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8485509104898695e-05,
+      "loss": 0.5277,
+      "step": 2412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8484867914849966e-05,
+      "loss": 0.6169,
+      "step": 2413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848422672480124e-05,
+      "loss": 0.5766,
+      "step": 2414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.84835855347525e-05,
+      "loss": 0.6517,
+      "step": 2415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8482944344703775e-05,
+      "loss": 0.7031,
+      "step": 2416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848230315465504e-05,
+      "loss": 0.6008,
+      "step": 2417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848166196460631e-05,
+      "loss": 0.5552,
+      "step": 2418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848102077455758e-05,
+      "loss": 0.6552,
+      "step": 2419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848037958450885e-05,
+      "loss": 0.6762,
+      "step": 2420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847973839446012e-05,
+      "loss": 0.6046,
+      "step": 2421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847909720441139e-05,
+      "loss": 0.6071,
+      "step": 2422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847845601436266e-05,
+      "loss": 0.6087,
+      "step": 2423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847781482431393e-05,
+      "loss": 0.6446,
+      "step": 2424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.84771736342652e-05,
+      "loss": 0.5699,
+      "step": 2425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8476532444216465e-05,
+      "loss": 0.5204,
+      "step": 2426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8475891254167736e-05,
+      "loss": 0.6482,
+      "step": 2427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847525006411901e-05,
+      "loss": 0.6246,
+      "step": 2428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847460887407027e-05,
+      "loss": 0.6576,
+      "step": 2429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847396768402155e-05,
+      "loss": 0.6998,
+      "step": 2430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8473326493972816e-05,
+      "loss": 0.5757,
+      "step": 2431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847268530392409e-05,
+      "loss": 0.5635,
+      "step": 2432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847204411387535e-05,
+      "loss": 0.5692,
+      "step": 2433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8471402923826625e-05,
+      "loss": 0.6281,
+      "step": 2434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847076173377789e-05,
+      "loss": 0.6075,
+      "step": 2435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847012054372916e-05,
+      "loss": 0.5741,
+      "step": 2436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846947935368043e-05,
+      "loss": 0.6414,
+      "step": 2437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8468838163631705e-05,
+      "loss": 0.6316,
+      "step": 2438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846819697358298e-05,
+      "loss": 0.5855,
+      "step": 2439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846755578353424e-05,
+      "loss": 0.65,
+      "step": 2440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846691459348551e-05,
+      "loss": 0.6501,
+      "step": 2441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846627340343678e-05,
+      "loss": 0.6412,
+      "step": 2442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846563221338805e-05,
+      "loss": 0.6516,
+      "step": 2443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8464991023339315e-05,
+      "loss": 0.574,
+      "step": 2444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846434983329059e-05,
+      "loss": 0.5815,
+      "step": 2445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8463708643241865e-05,
+      "loss": 0.6076,
+      "step": 2446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846306745319313e-05,
+      "loss": 0.6383,
+      "step": 2447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.84624262631444e-05,
+      "loss": 0.7021,
+      "step": 2448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846178507309567e-05,
+      "loss": 0.5549,
+      "step": 2449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846114388304694e-05,
+      "loss": 0.6817,
+      "step": 2450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8460502692998203e-05,
+      "loss": 0.6481,
+      "step": 2451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8459861502949475e-05,
+      "loss": 0.5839,
+      "step": 2452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845922031290075e-05,
+      "loss": 0.617,
+      "step": 2453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845857912285202e-05,
+      "loss": 0.6308,
+      "step": 2454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845793793280329e-05,
+      "loss": 0.5999,
+      "step": 2455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8457296742754555e-05,
+      "loss": 0.5733,
+      "step": 2456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845665555270583e-05,
+      "loss": 0.5887,
+      "step": 2457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845601436265709e-05,
+      "loss": 0.55,
+      "step": 2458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8455373172608364e-05,
+      "loss": 0.5702,
+      "step": 2459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845473198255963e-05,
+      "loss": 0.5764,
+      "step": 2460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.84540907925109e-05,
+      "loss": 0.5558,
+      "step": 2461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845344960246217e-05,
+      "loss": 0.5971,
+      "step": 2462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8452808412413444e-05,
+      "loss": 0.6227,
+      "step": 2463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8452167222364715e-05,
+      "loss": 0.68,
+      "step": 2464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845152603231598e-05,
+      "loss": 0.5733,
+      "step": 2465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845088484226725e-05,
+      "loss": 0.6785,
+      "step": 2466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845024365221852e-05,
+      "loss": 0.652,
+      "step": 2467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.844960246216979e-05,
+      "loss": 0.637,
+      "step": 2468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.844896127212106e-05,
+      "loss": 0.603,
+      "step": 2469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.844832008207233e-05,
+      "loss": 0.5757,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84476788920236e-05,
+      "loss": 0.6481,
+      "step": 2471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844703770197487e-05,
+      "loss": 0.6586,
+      "step": 2472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844639651192614e-05,
+      "loss": 0.5794,
+      "step": 2473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8445755321877405e-05,
+      "loss": 0.6427,
+      "step": 2474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844511413182868e-05,
+      "loss": 0.5743,
+      "step": 2475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844447294177994e-05,
+      "loss": 0.6421,
+      "step": 2476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8443831751731214e-05,
+      "loss": 0.5911,
+      "step": 2477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8443190561682486e-05,
+      "loss": 0.5927,
+      "step": 2478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844254937163376e-05,
+      "loss": 0.6799,
+      "step": 2479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844190818158502e-05,
+      "loss": 0.6733,
+      "step": 2480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8441266991536294e-05,
+      "loss": 0.7262,
+      "step": 2481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8440625801487566e-05,
+      "loss": 0.576,
+      "step": 2482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843998461143883e-05,
+      "loss": 0.5426,
+      "step": 2483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84393434213901e-05,
+      "loss": 0.6422,
+      "step": 2484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8438702231341374e-05,
+      "loss": 0.6475,
+      "step": 2485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8438061041292646e-05,
+      "loss": 0.6503,
+      "step": 2486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843741985124391e-05,
+      "loss": 0.5621,
+      "step": 2487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843677866119518e-05,
+      "loss": 0.6803,
+      "step": 2488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843613747114645e-05,
+      "loss": 0.6457,
+      "step": 2489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843549628109772e-05,
+      "loss": 0.5811,
+      "step": 2490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843485509104899e-05,
+      "loss": 0.5991,
+      "step": 2491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8434213901000256e-05,
+      "loss": 0.636,
+      "step": 2492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843357271095153e-05,
+      "loss": 0.5851,
+      "step": 2493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84329315209028e-05,
+      "loss": 0.6803,
+      "step": 2494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843229033085407e-05,
+      "loss": 0.6255,
+      "step": 2495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8431649140805336e-05,
+      "loss": 0.6352,
+      "step": 2496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843100795075661e-05,
+      "loss": 0.6284,
+      "step": 2497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843036676070787e-05,
+      "loss": 0.6519,
+      "step": 2498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8429725570659144e-05,
+      "loss": 0.627,
+      "step": 2499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8429084380610416e-05,
+      "loss": 0.635,
+      "step": 2500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842844319056169e-05,
+      "loss": 0.6574,
+      "step": 2501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842780200051296e-05,
+      "loss": 0.5769,
+      "step": 2502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8427160810464224e-05,
+      "loss": 0.601,
+      "step": 2503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8426519620415496e-05,
+      "loss": 0.6209,
+      "step": 2504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842587843036676e-05,
+      "loss": 0.6434,
+      "step": 2505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842523724031803e-05,
+      "loss": 0.6291,
+      "step": 2506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84245960502693e-05,
+      "loss": 0.5923,
+      "step": 2507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842395486022057e-05,
+      "loss": 0.6192,
+      "step": 2508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842331367017184e-05,
+      "loss": 0.6277,
+      "step": 2509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842267248012311e-05,
+      "loss": 0.5482,
+      "step": 2510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8422031290074384e-05,
+      "loss": 0.6588,
+      "step": 2511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842139010002565e-05,
+      "loss": 0.5471,
+      "step": 2512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842074890997692e-05,
+      "loss": 0.5699,
+      "step": 2513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8420107719928186e-05,
+      "loss": 0.546,
+      "step": 2514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841946652987946e-05,
+      "loss": 0.5595,
+      "step": 2515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841882533983072e-05,
+      "loss": 0.6171,
+      "step": 2516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8418184149781994e-05,
+      "loss": 0.6397,
+      "step": 2517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841754295973327e-05,
+      "loss": 0.5407,
+      "step": 2518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841690176968454e-05,
+      "loss": 0.668,
+      "step": 2519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841626057963581e-05,
+      "loss": 0.5989,
+      "step": 2520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8415619389587074e-05,
+      "loss": 0.6019,
+      "step": 2521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8414978199538346e-05,
+      "loss": 0.6194,
+      "step": 2522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841433700948961e-05,
+      "loss": 0.6456,
+      "step": 2523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841369581944088e-05,
+      "loss": 0.6635,
+      "step": 2524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8413054629392155e-05,
+      "loss": 0.6142,
+      "step": 2525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8412413439343426e-05,
+      "loss": 0.6152,
+      "step": 2526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84117722492947e-05,
+      "loss": 0.6466,
+      "step": 2527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841113105924596e-05,
+      "loss": 0.5871,
+      "step": 2528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8410489869197235e-05,
+      "loss": 0.6296,
+      "step": 2529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84098486791485e-05,
+      "loss": 0.5383,
+      "step": 2530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840920748909977e-05,
+      "loss": 0.6021,
+      "step": 2531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8408566299051036e-05,
+      "loss": 0.5892,
+      "step": 2532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840792510900231e-05,
+      "loss": 0.6262,
+      "step": 2533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840728391895358e-05,
+      "loss": 0.5905,
+      "step": 2534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840664272890485e-05,
+      "loss": 0.6469,
+      "step": 2535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840600153885612e-05,
+      "loss": 0.6304,
+      "step": 2536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840536034880739e-05,
+      "loss": 0.6575,
+      "step": 2537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840471915875866e-05,
+      "loss": 0.6326,
+      "step": 2538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8404077968709925e-05,
+      "loss": 0.6822,
+      "step": 2539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8403436778661196e-05,
+      "loss": 0.6274,
+      "step": 2540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840279558861247e-05,
+      "loss": 0.4992,
+      "step": 2541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840215439856374e-05,
+      "loss": 0.5355,
+      "step": 2542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8401513208515005e-05,
+      "loss": 0.576,
+      "step": 2543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8400872018466277e-05,
+      "loss": 0.5992,
+      "step": 2544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840023082841755e-05,
+      "loss": 0.72,
+      "step": 2545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839958963836881e-05,
+      "loss": 0.6533,
+      "step": 2546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8398948448320085e-05,
+      "loss": 0.5774,
+      "step": 2547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839830725827135e-05,
+      "loss": 0.6521,
+      "step": 2548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839766606822262e-05,
+      "loss": 0.5891,
+      "step": 2549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839702487817389e-05,
+      "loss": 0.5717,
+      "step": 2550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8396383688125165e-05,
+      "loss": 0.7153,
+      "step": 2551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839574249807643e-05,
+      "loss": 0.6342,
+      "step": 2552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83951013080277e-05,
+      "loss": 0.6422,
+      "step": 2553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839446011797897e-05,
+      "loss": 0.6327,
+      "step": 2554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839381892793024e-05,
+      "loss": 0.6183,
+      "step": 2555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839317773788151e-05,
+      "loss": 0.5797,
+      "step": 2556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839253654783278e-05,
+      "loss": 0.5807,
+      "step": 2557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8391895357784053e-05,
+      "loss": 0.7585,
+      "step": 2558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839125416773532e-05,
+      "loss": 0.5524,
+      "step": 2559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839061297768659e-05,
+      "loss": 0.5888,
+      "step": 2560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8389971787637855e-05,
+      "loss": 0.6792,
+      "step": 2561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838933059758913e-05,
+      "loss": 0.5442,
+      "step": 2562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83886894075404e-05,
+      "loss": 0.6722,
+      "step": 2563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8388048217491663e-05,
+      "loss": 0.5331,
+      "step": 2564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8387407027442935e-05,
+      "loss": 0.6282,
+      "step": 2565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838676583739421e-05,
+      "loss": 0.5224,
+      "step": 2566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838612464734548e-05,
+      "loss": 0.6259,
+      "step": 2567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8385483457296744e-05,
+      "loss": 0.6882,
+      "step": 2568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8384842267248015e-05,
+      "loss": 0.5863,
+      "step": 2569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838420107719928e-05,
+      "loss": 0.6598,
+      "step": 2570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838355988715055e-05,
+      "loss": 0.621,
+      "step": 2571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8382918697101824e-05,
+      "loss": 0.5723,
+      "step": 2572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8382277507053095e-05,
+      "loss": 0.5418,
+      "step": 2573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838163631700437e-05,
+      "loss": 0.5863,
+      "step": 2574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838099512695563e-05,
+      "loss": 0.6449,
+      "step": 2575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8380353936906904e-05,
+      "loss": 0.5531,
+      "step": 2576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837971274685817e-05,
+      "loss": 0.6895,
+      "step": 2577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837907155680944e-05,
+      "loss": 0.5555,
+      "step": 2578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8378430366760705e-05,
+      "loss": 0.5685,
+      "step": 2579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837778917671198e-05,
+      "loss": 0.6086,
+      "step": 2580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837714798666325e-05,
+      "loss": 0.5661,
+      "step": 2581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837650679661452e-05,
+      "loss": 0.702,
+      "step": 2582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837586560656579e-05,
+      "loss": 0.5788,
+      "step": 2583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837522441651706e-05,
+      "loss": 0.5632,
+      "step": 2584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837458322646833e-05,
+      "loss": 0.5503,
+      "step": 2585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8373942036419594e-05,
+      "loss": 0.6682,
+      "step": 2586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8373300846370865e-05,
+      "loss": 0.5739,
+      "step": 2587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837265965632213e-05,
+      "loss": 0.6104,
+      "step": 2588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83720184662734e-05,
+      "loss": 0.6155,
+      "step": 2589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837137727622468e-05,
+      "loss": 0.5106,
+      "step": 2590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8370736086175946e-05,
+      "loss": 0.6891,
+      "step": 2591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837009489612722e-05,
+      "loss": 0.5708,
+      "step": 2592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836945370607848e-05,
+      "loss": 0.5647,
+      "step": 2593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8368812516029754e-05,
+      "loss": 0.6405,
+      "step": 2594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836817132598102e-05,
+      "loss": 0.612,
+      "step": 2595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836753013593229e-05,
+      "loss": 0.5935,
+      "step": 2596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836688894588356e-05,
+      "loss": 0.5648,
+      "step": 2597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8366247755834834e-05,
+      "loss": 0.7012,
+      "step": 2598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8365606565786106e-05,
+      "loss": 0.5952,
+      "step": 2599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836496537573737e-05,
+      "loss": 0.6949,
+      "step": 2600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836432418568864e-05,
+      "loss": 0.5561,
+      "step": 2601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836368299563991e-05,
+      "loss": 0.5267,
+      "step": 2602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836304180559118e-05,
+      "loss": 0.7005,
+      "step": 2603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8362400615542444e-05,
+      "loss": 0.6048,
+      "step": 2604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8361759425493716e-05,
+      "loss": 0.6365,
+      "step": 2605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8361118235444994e-05,
+      "loss": 0.5982,
+      "step": 2606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836047704539626e-05,
+      "loss": 0.6054,
+      "step": 2607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835983585534753e-05,
+      "loss": 0.6156,
+      "step": 2608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8359194665298796e-05,
+      "loss": 0.5841,
+      "step": 2609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835855347525007e-05,
+      "loss": 0.564,
+      "step": 2610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835791228520133e-05,
+      "loss": 0.5996,
+      "step": 2611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8357271095152604e-05,
+      "loss": 0.5457,
+      "step": 2612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8356629905103876e-05,
+      "loss": 0.6527,
+      "step": 2613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835598871505515e-05,
+      "loss": 0.547,
+      "step": 2614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835534752500642e-05,
+      "loss": 0.5768,
+      "step": 2615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8354706334957684e-05,
+      "loss": 0.6149,
+      "step": 2616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8354065144908956e-05,
+      "loss": 0.574,
+      "step": 2617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835342395486022e-05,
+      "loss": 0.6757,
+      "step": 2618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835278276481149e-05,
+      "loss": 0.5539,
+      "step": 2619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835214157476276e-05,
+      "loss": 0.5923,
+      "step": 2620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835150038471403e-05,
+      "loss": 0.6408,
+      "step": 2621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83508591946653e-05,
+      "loss": 0.6969,
+      "step": 2622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835021800461657e-05,
+      "loss": 0.606,
+      "step": 2623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8349576814567844e-05,
+      "loss": 0.5734,
+      "step": 2624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.834893562451911e-05,
+      "loss": 0.4745,
+      "step": 2625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.834829443447038e-05,
+      "loss": 0.6543,
+      "step": 2626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8347653244421646e-05,
+      "loss": 0.5765,
+      "step": 2627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.834701205437292e-05,
+      "loss": 0.5725,
+      "step": 2628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.834637086432419e-05,
+      "loss": 0.7563,
+      "step": 2629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.834572967427546e-05,
+      "loss": 0.5741,
+      "step": 2630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8345088484226726e-05,
+      "loss": 0.7573,
+      "step": 2631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8344447294178e-05,
+      "loss": 0.5905,
+      "step": 2632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.834380610412927e-05,
+      "loss": 0.5704,
+      "step": 2633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8343164914080535e-05,
+      "loss": 0.6546,
+      "step": 2634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8342523724031806e-05,
+      "loss": 0.6003,
+      "step": 2635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.834188253398307e-05,
+      "loss": 0.6059,
+      "step": 2636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.834124134393434e-05,
+      "loss": 0.566,
+      "step": 2637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8340600153885615e-05,
+      "loss": 0.6115,
+      "step": 2638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8339958963836886e-05,
+      "loss": 0.5746,
+      "step": 2639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833931777378815e-05,
+      "loss": 0.6346,
+      "step": 2640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833867658373942e-05,
+      "loss": 0.5324,
+      "step": 2641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8338035393690695e-05,
+      "loss": 0.5775,
+      "step": 2642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833739420364196e-05,
+      "loss": 0.6161,
+      "step": 2643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833675301359323e-05,
+      "loss": 0.5861,
+      "step": 2644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83361118235445e-05,
+      "loss": 0.628,
+      "step": 2645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8335470633495775e-05,
+      "loss": 0.605,
+      "step": 2646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833482944344704e-05,
+      "loss": 0.5298,
+      "step": 2647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833418825339831e-05,
+      "loss": 0.5698,
+      "step": 2648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8333547063349576e-05,
+      "loss": 0.6307,
+      "step": 2649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833290587330085e-05,
+      "loss": 0.5793,
+      "step": 2650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833226468325212e-05,
+      "loss": 0.6046,
+      "step": 2651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8331623493203385e-05,
+      "loss": 0.6921,
+      "step": 2652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8330982303154656e-05,
+      "loss": 0.5393,
+      "step": 2653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.833034111310593e-05,
+      "loss": 0.6346,
+      "step": 2654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83296999230572e-05,
+      "loss": 0.5497,
+      "step": 2655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8329058733008465e-05,
+      "loss": 0.6673,
+      "step": 2656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8328417542959737e-05,
+      "loss": 0.561,
+      "step": 2657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8327776352911e-05,
+      "loss": 0.5441,
+      "step": 2658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832713516286227e-05,
+      "loss": 0.5566,
+      "step": 2659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8326493972813545e-05,
+      "loss": 0.5574,
+      "step": 2660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832585278276482e-05,
+      "loss": 0.5166,
+      "step": 2661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832521159271609e-05,
+      "loss": 0.6454,
+      "step": 2662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832457040266735e-05,
+      "loss": 0.6047,
+      "step": 2663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8323929212618625e-05,
+      "loss": 0.7754,
+      "step": 2664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832328802256989e-05,
+      "loss": 0.6438,
+      "step": 2665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832264683252116e-05,
+      "loss": 0.676,
+      "step": 2666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832200564247243e-05,
+      "loss": 0.6369,
+      "step": 2667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83213644524237e-05,
+      "loss": 0.5368,
+      "step": 2668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832072326237497e-05,
+      "loss": 0.649,
+      "step": 2669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832008207232624e-05,
+      "loss": 0.5439,
+      "step": 2670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8319440882277513e-05,
+      "loss": 0.5839,
+      "step": 2671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831879969222878e-05,
+      "loss": 0.5691,
+      "step": 2672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831815850218005e-05,
+      "loss": 0.6706,
+      "step": 2673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8317517312131315e-05,
+      "loss": 0.6193,
+      "step": 2674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831687612208259e-05,
+      "loss": 0.6145,
+      "step": 2675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831623493203385e-05,
+      "loss": 0.5759,
+      "step": 2676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8315593741985123e-05,
+      "loss": 0.6655,
+      "step": 2677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83149525519364e-05,
+      "loss": 0.7086,
+      "step": 2678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831431136188767e-05,
+      "loss": 0.6029,
+      "step": 2679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831367017183894e-05,
+      "loss": 0.6175,
+      "step": 2680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8313028981790204e-05,
+      "loss": 0.6298,
+      "step": 2681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8312387791741475e-05,
+      "loss": 0.6068,
+      "step": 2682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831174660169274e-05,
+      "loss": 0.5562,
+      "step": 2683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831110541164401e-05,
+      "loss": 0.6325,
+      "step": 2684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8310464221595284e-05,
+      "loss": 0.5959,
+      "step": 2685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8309823031546555e-05,
+      "loss": 0.5889,
+      "step": 2686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830918184149783e-05,
+      "loss": 0.5245,
+      "step": 2687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830854065144909e-05,
+      "loss": 0.7133,
+      "step": 2688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8307899461400364e-05,
+      "loss": 0.6122,
+      "step": 2689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830725827135163e-05,
+      "loss": 0.6502,
+      "step": 2690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83066170813029e-05,
+      "loss": 0.7062,
+      "step": 2691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8305975891254165e-05,
+      "loss": 0.6643,
+      "step": 2692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830533470120544e-05,
+      "loss": 0.5784,
+      "step": 2693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830469351115671e-05,
+      "loss": 0.5818,
+      "step": 2694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830405232110798e-05,
+      "loss": 0.6207,
+      "step": 2695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830341113105925e-05,
+      "loss": 0.5741,
+      "step": 2696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830276994101052e-05,
+      "loss": 0.5582,
+      "step": 2697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.830212875096179e-05,
+      "loss": 0.5828,
+      "step": 2698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8301487560913054e-05,
+      "loss": 0.5538,
+      "step": 2699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8300846370864326e-05,
+      "loss": 0.5617,
+      "step": 2700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.83002051808156e-05,
+      "loss": 0.5872,
+      "step": 2701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829956399076687e-05,
+      "loss": 0.5561,
+      "step": 2702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8298922800718134e-05,
+      "loss": 0.6239,
+      "step": 2703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8298281610669406e-05,
+      "loss": 0.6421,
+      "step": 2704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829764042062068e-05,
+      "loss": 0.6947,
+      "step": 2705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829699923057194e-05,
+      "loss": 0.5283,
+      "step": 2706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8296358040523214e-05,
+      "loss": 0.5945,
+      "step": 2707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829571685047448e-05,
+      "loss": 0.5909,
+      "step": 2708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829507566042575e-05,
+      "loss": 0.6507,
+      "step": 2709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829443447037702e-05,
+      "loss": 0.6503,
+      "step": 2710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8293793280328294e-05,
+      "loss": 0.5312,
+      "step": 2711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829315209027956e-05,
+      "loss": 0.5675,
+      "step": 2712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829251090023083e-05,
+      "loss": 0.5432,
+      "step": 2713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.82918697101821e-05,
+      "loss": 0.5833,
+      "step": 2714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829122852013337e-05,
+      "loss": 0.5931,
+      "step": 2715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829058733008464e-05,
+      "loss": 0.585,
+      "step": 2716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828994614003591e-05,
+      "loss": 0.5368,
+      "step": 2717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828930494998718e-05,
+      "loss": 0.5897,
+      "step": 2718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828866375993845e-05,
+      "loss": 0.553,
+      "step": 2719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828802256988972e-05,
+      "loss": 0.6818,
+      "step": 2720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8287381379840984e-05,
+      "loss": 0.543,
+      "step": 2721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8286740189792256e-05,
+      "loss": 0.5774,
+      "step": 2722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828609899974353e-05,
+      "loss": 0.6244,
+      "step": 2723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828545780969479e-05,
+      "loss": 0.6153,
+      "step": 2724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8284816619646064e-05,
+      "loss": 0.6049,
+      "step": 2725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8284175429597336e-05,
+      "loss": 0.6603,
+      "step": 2726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828353423954861e-05,
+      "loss": 0.6293,
+      "step": 2727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828289304949987e-05,
+      "loss": 0.5752,
+      "step": 2728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8282251859451144e-05,
+      "loss": 0.5907,
+      "step": 2729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828161066940241e-05,
+      "loss": 0.5979,
+      "step": 2730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828096947935368e-05,
+      "loss": 0.6068,
+      "step": 2731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.828032828930495e-05,
+      "loss": 0.5727,
+      "step": 2732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8279687099256224e-05,
+      "loss": 0.6181,
+      "step": 2733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8279045909207496e-05,
+      "loss": 0.642,
+      "step": 2734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827840471915876e-05,
+      "loss": 0.5774,
+      "step": 2735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827776352911003e-05,
+      "loss": 0.6597,
+      "step": 2736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82771223390613e-05,
+      "loss": 0.6662,
+      "step": 2737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827648114901257e-05,
+      "loss": 0.6261,
+      "step": 2738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8275839958963834e-05,
+      "loss": 0.533,
+      "step": 2739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8275198768915106e-05,
+      "loss": 0.6659,
+      "step": 2740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827455757886638e-05,
+      "loss": 0.6633,
+      "step": 2741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827391638881765e-05,
+      "loss": 0.5725,
+      "step": 2742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827327519876892e-05,
+      "loss": 0.5961,
+      "step": 2743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8272634008720186e-05,
+      "loss": 0.6891,
+      "step": 2744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827199281867146e-05,
+      "loss": 0.5911,
+      "step": 2745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827135162862272e-05,
+      "loss": 0.5732,
+      "step": 2746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8270710438573995e-05,
+      "loss": 0.7139,
+      "step": 2747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827006924852526e-05,
+      "loss": 0.5869,
+      "step": 2748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826942805847654e-05,
+      "loss": 0.6558,
+      "step": 2749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826878686842781e-05,
+      "loss": 0.5731,
+      "step": 2750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8268145678379075e-05,
+      "loss": 0.6155,
+      "step": 2751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8267504488330346e-05,
+      "loss": 0.6226,
+      "step": 2752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826686329828161e-05,
+      "loss": 0.5601,
+      "step": 2753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826622210823288e-05,
+      "loss": 0.5867,
+      "step": 2754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826558091818415e-05,
+      "loss": 0.5604,
+      "step": 2755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826493972813542e-05,
+      "loss": 0.5707,
+      "step": 2756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826429853808669e-05,
+      "loss": 0.5231,
+      "step": 2757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826365734803796e-05,
+      "loss": 0.7331,
+      "step": 2758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8263016157989235e-05,
+      "loss": 0.537,
+      "step": 2759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82623749679405e-05,
+      "loss": 0.5532,
+      "step": 2760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826173377789177e-05,
+      "loss": 0.6193,
+      "step": 2761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8261092587843036e-05,
+      "loss": 0.5266,
+      "step": 2762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826045139779431e-05,
+      "loss": 0.5672,
+      "step": 2763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825981020774557e-05,
+      "loss": 0.6226,
+      "step": 2764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8259169017696845e-05,
+      "loss": 0.6217,
+      "step": 2765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8258527827648117e-05,
+      "loss": 0.6108,
+      "step": 2766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825788663759939e-05,
+      "loss": 0.5335,
+      "step": 2767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825724544755066e-05,
+      "loss": 0.5643,
+      "step": 2768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8256604257501925e-05,
+      "loss": 0.6882,
+      "step": 2769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8255963067453197e-05,
+      "loss": 0.7252,
+      "step": 2770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825532187740446e-05,
+      "loss": 0.6466,
+      "step": 2771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825468068735573e-05,
+      "loss": 0.6332,
+      "step": 2772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8254039497307005e-05,
+      "loss": 0.5505,
+      "step": 2773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825339830725828e-05,
+      "loss": 0.5707,
+      "step": 2774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825275711720954e-05,
+      "loss": 0.4838,
+      "step": 2775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825211592716081e-05,
+      "loss": 0.6431,
+      "step": 2776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8251474737112085e-05,
+      "loss": 0.5509,
+      "step": 2777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825083354706335e-05,
+      "loss": 0.6864,
+      "step": 2778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825019235701462e-05,
+      "loss": 0.5604,
+      "step": 2779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824955116696589e-05,
+      "loss": 0.6631,
+      "step": 2780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824890997691716e-05,
+      "loss": 0.5618,
+      "step": 2781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824826878686843e-05,
+      "loss": 0.6284,
+      "step": 2782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82476275968197e-05,
+      "loss": 0.5846,
+      "step": 2783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824698640677097e-05,
+      "loss": 0.6087,
+      "step": 2784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824634521672224e-05,
+      "loss": 0.5855,
+      "step": 2785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824570402667351e-05,
+      "loss": 0.6115,
+      "step": 2786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8245062836624775e-05,
+      "loss": 0.696,
+      "step": 2787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824442164657605e-05,
+      "loss": 0.5793,
+      "step": 2788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824378045652732e-05,
+      "loss": 0.5863,
+      "step": 2789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824313926647859e-05,
+      "loss": 0.6209,
+      "step": 2790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8242498076429855e-05,
+      "loss": 0.6366,
+      "step": 2791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824185688638113e-05,
+      "loss": 0.588,
+      "step": 2792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824121569633239e-05,
+      "loss": 0.6022,
+      "step": 2793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8240574506283664e-05,
+      "loss": 0.6549,
+      "step": 2794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8239933316234935e-05,
+      "loss": 0.644,
+      "step": 2795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82392921261862e-05,
+      "loss": 0.6327,
+      "step": 2796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823865093613747e-05,
+      "loss": 0.6594,
+      "step": 2797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8238009746088744e-05,
+      "loss": 0.6104,
+      "step": 2798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8237368556040015e-05,
+      "loss": 0.5564,
+      "step": 2799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823672736599128e-05,
+      "loss": 0.6167,
+      "step": 2800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823608617594255e-05,
+      "loss": 0.5898,
+      "step": 2801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8235444985893824e-05,
+      "loss": 0.5908,
+      "step": 2802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823480379584509e-05,
+      "loss": 0.6133,
+      "step": 2803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823416260579636e-05,
+      "loss": 0.5954,
+      "step": 2804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823352141574763e-05,
+      "loss": 0.6622,
+      "step": 2805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8232880225698904e-05,
+      "loss": 0.6413,
+      "step": 2806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823223903565017e-05,
+      "loss": 0.5431,
+      "step": 2807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823159784560144e-05,
+      "loss": 0.6559,
+      "step": 2808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8230956655552705e-05,
+      "loss": 0.5468,
+      "step": 2809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823031546550398e-05,
+      "loss": 0.5726,
+      "step": 2810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822967427545525e-05,
+      "loss": 0.5833,
+      "step": 2811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8229033085406514e-05,
+      "loss": 0.5681,
+      "step": 2812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8228391895357786e-05,
+      "loss": 0.6491,
+      "step": 2813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822775070530906e-05,
+      "loss": 0.6424,
+      "step": 2814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822710951526033e-05,
+      "loss": 0.647,
+      "step": 2815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8226468325211594e-05,
+      "loss": 0.5628,
+      "step": 2816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8225827135162866e-05,
+      "loss": 0.6356,
+      "step": 2817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822518594511413e-05,
+      "loss": 0.5638,
+      "step": 2818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82245447550654e-05,
+      "loss": 0.5554,
+      "step": 2819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8223903565016674e-05,
+      "loss": 0.6588,
+      "step": 2820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8223262374967946e-05,
+      "loss": 0.5748,
+      "step": 2821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822262118491922e-05,
+      "loss": 0.7032,
+      "step": 2822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822197999487048e-05,
+      "loss": 0.7085,
+      "step": 2823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8221338804821754e-05,
+      "loss": 0.5667,
+      "step": 2824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822069761477302e-05,
+      "loss": 0.5511,
+      "step": 2825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822005642472429e-05,
+      "loss": 0.6383,
+      "step": 2826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8219415234675556e-05,
+      "loss": 0.6517,
+      "step": 2827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821877404462683e-05,
+      "loss": 0.6,
+      "step": 2828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82181328545781e-05,
+      "loss": 0.5052,
+      "step": 2829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821749166452937e-05,
+      "loss": 0.6272,
+      "step": 2830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821685047448064e-05,
+      "loss": 0.5066,
+      "step": 2831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821620928443191e-05,
+      "loss": 0.601,
+      "step": 2832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821556809438318e-05,
+      "loss": 0.5526,
+      "step": 2833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8214926904334444e-05,
+      "loss": 0.5614,
+      "step": 2834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.5487,
+      "step": 2835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821364452423698e-05,
+      "loss": 0.5203,
+      "step": 2836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821300333418825e-05,
+      "loss": 0.6025,
+      "step": 2837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821236214413953e-05,
+      "loss": 0.6287,
+      "step": 2838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8211720954090796e-05,
+      "loss": 0.6595,
+      "step": 2839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821107976404207e-05,
+      "loss": 0.5079,
+      "step": 2840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821043857399333e-05,
+      "loss": 0.7033,
+      "step": 2841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8209797383944604e-05,
+      "loss": 0.4991,
+      "step": 2842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820915619389587e-05,
+      "loss": 0.5895,
+      "step": 2843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820851500384714e-05,
+      "loss": 0.6176,
+      "step": 2844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820787381379841e-05,
+      "loss": 0.6382,
+      "step": 2845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8207232623749684e-05,
+      "loss": 0.601,
+      "step": 2846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8206591433700956e-05,
+      "loss": 0.6089,
+      "step": 2847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820595024365222e-05,
+      "loss": 0.6423,
+      "step": 2848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820530905360349e-05,
+      "loss": 0.5894,
+      "step": 2849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820466786355476e-05,
+      "loss": 0.6899,
+      "step": 2850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820402667350603e-05,
+      "loss": 0.6189,
+      "step": 2851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8203385483457294e-05,
+      "loss": 0.6685,
+      "step": 2852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8202744293408566e-05,
+      "loss": 0.5562,
+      "step": 2853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820210310335984e-05,
+      "loss": 0.5904,
+      "step": 2854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820146191331111e-05,
+      "loss": 0.5035,
+      "step": 2855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820082072326238e-05,
+      "loss": 0.62,
+      "step": 2856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8200179533213646e-05,
+      "loss": 0.6277,
+      "step": 2857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819953834316492e-05,
+      "loss": 0.5465,
+      "step": 2858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819889715311618e-05,
+      "loss": 0.6101,
+      "step": 2859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8198255963067455e-05,
+      "loss": 0.6991,
+      "step": 2860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8197614773018726e-05,
+      "loss": 0.6433,
+      "step": 2861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819697358297e-05,
+      "loss": 0.635,
+      "step": 2862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819633239292126e-05,
+      "loss": 0.5229,
+      "step": 2863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8195691202872535e-05,
+      "loss": 0.5714,
+      "step": 2864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8195050012823806e-05,
+      "loss": 0.5807,
+      "step": 2865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819440882277507e-05,
+      "loss": 0.7208,
+      "step": 2866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819376763272634e-05,
+      "loss": 0.6496,
+      "step": 2867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819312644267761e-05,
+      "loss": 0.6608,
+      "step": 2868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819248525262888e-05,
+      "loss": 0.545,
+      "step": 2869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819184406258015e-05,
+      "loss": 0.4993,
+      "step": 2870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819120287253142e-05,
+      "loss": 0.6209,
+      "step": 2871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.819056168248269e-05,
+      "loss": 0.6118,
+      "step": 2872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818992049243396e-05,
+      "loss": 0.5993,
+      "step": 2873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818927930238523e-05,
+      "loss": 0.644,
+      "step": 2874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8188638112336496e-05,
+      "loss": 0.6399,
+      "step": 2875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818799692228777e-05,
+      "loss": 0.6929,
+      "step": 2876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818735573223904e-05,
+      "loss": 0.6498,
+      "step": 2877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818671454219031e-05,
+      "loss": 0.6478,
+      "step": 2878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8186073352141577e-05,
+      "loss": 0.5597,
+      "step": 2879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818543216209285e-05,
+      "loss": 0.5828,
+      "step": 2880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818479097204411e-05,
+      "loss": 0.6842,
+      "step": 2881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8184149781995385e-05,
+      "loss": 0.5975,
+      "step": 2882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818350859194666e-05,
+      "loss": 0.6956,
+      "step": 2883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818286740189792e-05,
+      "loss": 0.6114,
+      "step": 2884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818222621184919e-05,
+      "loss": 0.6296,
+      "step": 2885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8181585021800465e-05,
+      "loss": 0.5806,
+      "step": 2886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.818094383175174e-05,
+      "loss": 0.6447,
+      "step": 2887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8180302641703e-05,
+      "loss": 0.6931,
+      "step": 2888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817966145165427e-05,
+      "loss": 0.5979,
+      "step": 2889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817902026160554e-05,
+      "loss": 0.6468,
+      "step": 2890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817837907155681e-05,
+      "loss": 0.6368,
+      "step": 2891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817773788150808e-05,
+      "loss": 0.6699,
+      "step": 2892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8177096691459353e-05,
+      "loss": 0.5966,
+      "step": 2893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8176455501410625e-05,
+      "loss": 0.5513,
+      "step": 2894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817581431136189e-05,
+      "loss": 0.6738,
+      "step": 2895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817517312131316e-05,
+      "loss": 0.5904,
+      "step": 2896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817453193126443e-05,
+      "loss": 0.6049,
+      "step": 2897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.81738907412157e-05,
+      "loss": 0.5826,
+      "step": 2898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8173249551166963e-05,
+      "loss": 0.652,
+      "step": 2899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8172608361118235e-05,
+      "loss": 0.667,
+      "step": 2900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817196717106951e-05,
+      "loss": 0.6141,
+      "step": 2901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817132598102078e-05,
+      "loss": 0.6781,
+      "step": 2902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817068479097205e-05,
+      "loss": 0.5849,
+      "step": 2903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8170043600923315e-05,
+      "loss": 0.6586,
+      "step": 2904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816940241087459e-05,
+      "loss": 0.602,
+      "step": 2905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816876122082585e-05,
+      "loss": 0.6121,
+      "step": 2906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8168120030777124e-05,
+      "loss": 0.651,
+      "step": 2907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816747884072839e-05,
+      "loss": 0.6755,
+      "step": 2908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816683765067967e-05,
+      "loss": 0.628,
+      "step": 2909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816619646063094e-05,
+      "loss": 0.5866,
+      "step": 2910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8165555270582204e-05,
+      "loss": 0.6611,
+      "step": 2911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8164914080533475e-05,
+      "loss": 0.5656,
+      "step": 2912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816427289048474e-05,
+      "loss": 0.5851,
+      "step": 2913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816363170043601e-05,
+      "loss": 0.5861,
+      "step": 2914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816299051038728e-05,
+      "loss": 0.6883,
+      "step": 2915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816234932033855e-05,
+      "loss": 0.6101,
+      "step": 2916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816170813028982e-05,
+      "loss": 0.6191,
+      "step": 2917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816106694024109e-05,
+      "loss": 0.6505,
+      "step": 2918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8160425750192364e-05,
+      "loss": 0.7212,
+      "step": 2919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.815978456014363e-05,
+      "loss": 0.6198,
+      "step": 2920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.81591433700949e-05,
+      "loss": 0.6226,
+      "step": 2921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8158502180046165e-05,
+      "loss": 0.6156,
+      "step": 2922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.815786098999744e-05,
+      "loss": 0.6381,
+      "step": 2923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.81572197999487e-05,
+      "loss": 0.606,
+      "step": 2924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8156578609899974e-05,
+      "loss": 0.6124,
+      "step": 2925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8155937419851246e-05,
+      "loss": 0.5828,
+      "step": 2926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.815529622980252e-05,
+      "loss": 0.617,
+      "step": 2927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.815465503975379e-05,
+      "loss": 0.6559,
+      "step": 2928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8154013849705054e-05,
+      "loss": 0.563,
+      "step": 2929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8153372659656326e-05,
+      "loss": 0.6352,
+      "step": 2930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.815273146960759e-05,
+      "loss": 0.6729,
+      "step": 2931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.815209027955886e-05,
+      "loss": 0.6244,
+      "step": 2932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8151449089510134e-05,
+      "loss": 0.6027,
+      "step": 2933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8150807899461406e-05,
+      "loss": 0.631,
+      "step": 2934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.815016670941267e-05,
+      "loss": 0.548,
+      "step": 2935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814952551936394e-05,
+      "loss": 0.5959,
+      "step": 2936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8148884329315214e-05,
+      "loss": 0.5802,
+      "step": 2937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814824313926648e-05,
+      "loss": 0.6362,
+      "step": 2938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814760194921775e-05,
+      "loss": 0.6076,
+      "step": 2939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8146960759169016e-05,
+      "loss": 0.5652,
+      "step": 2940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814631956912029e-05,
+      "loss": 0.5973,
+      "step": 2941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814567837907156e-05,
+      "loss": 0.5709,
+      "step": 2942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814503718902283e-05,
+      "loss": 0.6402,
+      "step": 2943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8144395998974096e-05,
+      "loss": 0.6445,
+      "step": 2944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814375480892537e-05,
+      "loss": 0.5905,
+      "step": 2945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814311361887664e-05,
+      "loss": 0.5192,
+      "step": 2946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8142472428827904e-05,
+      "loss": 0.6331,
+      "step": 2947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8141831238779176e-05,
+      "loss": 0.5151,
+      "step": 2948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814119004873045e-05,
+      "loss": 0.619,
+      "step": 2949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814054885868172e-05,
+      "loss": 0.6083,
+      "step": 2950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8139907668632984e-05,
+      "loss": 0.5158,
+      "step": 2951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8139266478584256e-05,
+      "loss": 0.6135,
+      "step": 2952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813862528853552e-05,
+      "loss": 0.5415,
+      "step": 2953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813798409848679e-05,
+      "loss": 0.6406,
+      "step": 2954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8137342908438064e-05,
+      "loss": 0.5842,
+      "step": 2955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813670171838933e-05,
+      "loss": 0.6806,
+      "step": 2956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.81360605283406e-05,
+      "loss": 0.6058,
+      "step": 2957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813541933829187e-05,
+      "loss": 0.6466,
+      "step": 2958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8134778148243144e-05,
+      "loss": 0.5783,
+      "step": 2959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813413695819441e-05,
+      "loss": 0.6344,
+      "step": 2960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813349576814568e-05,
+      "loss": 0.6575,
+      "step": 2961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8132854578096946e-05,
+      "loss": 0.6069,
+      "step": 2962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813221338804822e-05,
+      "loss": 0.5447,
+      "step": 2963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813157219799949e-05,
+      "loss": 0.7508,
+      "step": 2964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813093100795076e-05,
+      "loss": 0.6558,
+      "step": 2965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.813028981790203e-05,
+      "loss": 0.5367,
+      "step": 2966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.81296486278533e-05,
+      "loss": 0.6356,
+      "step": 2967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.812900743780457e-05,
+      "loss": 0.5604,
+      "step": 2968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8128366247755835e-05,
+      "loss": 0.6462,
+      "step": 2969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8127725057707106e-05,
+      "loss": 0.6353,
+      "step": 2970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.812708386765837e-05,
+      "loss": 0.5342,
+      "step": 2971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.812644267760964e-05,
+      "loss": 0.5803,
+      "step": 2972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8125801487560915e-05,
+      "loss": 0.6289,
+      "step": 2973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8125160297512186e-05,
+      "loss": 0.6035,
+      "step": 2974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.812451910746346e-05,
+      "loss": 0.578,
+      "step": 2975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.812387791741472e-05,
+      "loss": 0.6948,
+      "step": 2976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8123236727365995e-05,
+      "loss": 0.5908,
+      "step": 2977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.812259553731726e-05,
+      "loss": 0.626,
+      "step": 2978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.812195434726853e-05,
+      "loss": 0.611,
+      "step": 2979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8121313157219796e-05,
+      "loss": 0.6115,
+      "step": 2980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8120671967171075e-05,
+      "loss": 0.5476,
+      "step": 2981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8120030777122347e-05,
+      "loss": 0.6723,
+      "step": 2982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.811938958707361e-05,
+      "loss": 0.5919,
+      "step": 2983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.811874839702488e-05,
+      "loss": 0.5312,
+      "step": 2984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.811810720697615e-05,
+      "loss": 0.5351,
+      "step": 2985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.811746601692742e-05,
+      "loss": 0.5858,
+      "step": 2986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8116824826878685e-05,
+      "loss": 0.6524,
+      "step": 2987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8116183636829956e-05,
+      "loss": 0.5887,
+      "step": 2988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.811554244678123e-05,
+      "loss": 0.6442,
+      "step": 2989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.81149012567325e-05,
+      "loss": 0.5482,
+      "step": 2990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.811426006668377e-05,
+      "loss": 0.571,
+      "step": 2991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8113618876635037e-05,
+      "loss": 0.6136,
+      "step": 2992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811297768658631e-05,
+      "loss": 0.551,
+      "step": 2993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811233649653757e-05,
+      "loss": 0.6743,
+      "step": 2994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8111695306488845e-05,
+      "loss": 0.6765,
+      "step": 2995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811105411644011e-05,
+      "loss": 0.6995,
+      "step": 2996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811041292639138e-05,
+      "loss": 0.5916,
+      "step": 2997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810977173634266e-05,
+      "loss": 0.4935,
+      "step": 2998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8109130546293925e-05,
+      "loss": 0.736,
+      "step": 2999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.81084893562452e-05,
+      "loss": 0.615,
+      "step": 3000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810784816619646e-05,
+      "loss": 0.5856,
+      "step": 3001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8107206976147733e-05,
+      "loss": 0.5849,
+      "step": 3002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8106565786099e-05,
+      "loss": 0.544,
+      "step": 3003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810592459605027e-05,
+      "loss": 0.5699,
+      "step": 3004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810528340600154e-05,
+      "loss": 0.6792,
+      "step": 3005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8104642215952813e-05,
+      "loss": 0.5704,
+      "step": 3006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8104001025904085e-05,
+      "loss": 0.5725,
+      "step": 3007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810335983585535e-05,
+      "loss": 0.5894,
+      "step": 3008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810271864580662e-05,
+      "loss": 0.6052,
+      "step": 3009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810207745575789e-05,
+      "loss": 0.6745,
+      "step": 3010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810143626570916e-05,
+      "loss": 0.6395,
+      "step": 3011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8100795075660423e-05,
+      "loss": 0.5997,
+      "step": 3012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8100153885611695e-05,
+      "loss": 0.6049,
+      "step": 3013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809951269556297e-05,
+      "loss": 0.5928,
+      "step": 3014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809887150551424e-05,
+      "loss": 0.5229,
+      "step": 3015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809823031546551e-05,
+      "loss": 0.5184,
+      "step": 3016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8097589125416775e-05,
+      "loss": 0.5862,
+      "step": 3017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809694793536805e-05,
+      "loss": 0.5489,
+      "step": 3018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809630674531931e-05,
+      "loss": 0.6231,
+      "step": 3019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8095665555270584e-05,
+      "loss": 0.5552,
+      "step": 3020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8095024365221855e-05,
+      "loss": 0.5526,
+      "step": 3021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809438317517313e-05,
+      "loss": 0.6294,
+      "step": 3022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809374198512439e-05,
+      "loss": 0.5669,
+      "step": 3023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8093100795075664e-05,
+      "loss": 0.6091,
+      "step": 3024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8092459605026935e-05,
+      "loss": 0.6419,
+      "step": 3025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80918184149782e-05,
+      "loss": 0.5443,
+      "step": 3026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809117722492947e-05,
+      "loss": 0.628,
+      "step": 3027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809053603488074e-05,
+      "loss": 0.511,
+      "step": 3028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808989484483201e-05,
+      "loss": 0.6491,
+      "step": 3029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808925365478328e-05,
+      "loss": 0.5309,
+      "step": 3030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808861246473455e-05,
+      "loss": 0.5819,
+      "step": 3031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808797127468582e-05,
+      "loss": 0.6367,
+      "step": 3032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808733008463709e-05,
+      "loss": 0.5768,
+      "step": 3033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808668889458836e-05,
+      "loss": 0.6466,
+      "step": 3034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8086047704539626e-05,
+      "loss": 0.6834,
+      "step": 3035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80854065144909e-05,
+      "loss": 0.5699,
+      "step": 3036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808476532444217e-05,
+      "loss": 0.5967,
+      "step": 3037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808412413439344e-05,
+      "loss": 0.6904,
+      "step": 3038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8083482944344706e-05,
+      "loss": 0.654,
+      "step": 3039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808284175429598e-05,
+      "loss": 0.5711,
+      "step": 3040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808220056424724e-05,
+      "loss": 0.6394,
+      "step": 3041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8081559374198514e-05,
+      "loss": 0.6171,
+      "step": 3042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8080918184149786e-05,
+      "loss": 0.7088,
+      "step": 3043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808027699410105e-05,
+      "loss": 0.5951,
+      "step": 3044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807963580405232e-05,
+      "loss": 0.6325,
+      "step": 3045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8078994614003594e-05,
+      "loss": 0.6295,
+      "step": 3046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8078353423954866e-05,
+      "loss": 0.6659,
+      "step": 3047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807771223390613e-05,
+      "loss": 0.682,
+      "step": 3048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80770710438574e-05,
+      "loss": 0.5896,
+      "step": 3049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807642985380867e-05,
+      "loss": 0.6238,
+      "step": 3050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807578866375994e-05,
+      "loss": 0.6323,
+      "step": 3051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807514747371121e-05,
+      "loss": 0.6326,
+      "step": 3052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807450628366248e-05,
+      "loss": 0.5884,
+      "step": 3053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8073865093613754e-05,
+      "loss": 0.548,
+      "step": 3054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807322390356502e-05,
+      "loss": 0.6529,
+      "step": 3055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807258271351629e-05,
+      "loss": 0.585,
+      "step": 3056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8071941523467556e-05,
+      "loss": 0.65,
+      "step": 3057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807130033341883e-05,
+      "loss": 0.5924,
+      "step": 3058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807065914337009e-05,
+      "loss": 0.5309,
+      "step": 3059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8070017953321364e-05,
+      "loss": 0.5491,
+      "step": 3060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8069376763272636e-05,
+      "loss": 0.6194,
+      "step": 3061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806873557322391e-05,
+      "loss": 0.5653,
+      "step": 3062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806809438317518e-05,
+      "loss": 0.5774,
+      "step": 3063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8067453193126444e-05,
+      "loss": 0.5753,
+      "step": 3064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8066812003077716e-05,
+      "loss": 0.6248,
+      "step": 3065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806617081302898e-05,
+      "loss": 0.5918,
+      "step": 3066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806552962298025e-05,
+      "loss": 0.5243,
+      "step": 3067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806488843293152e-05,
+      "loss": 0.5649,
+      "step": 3068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8064247242882796e-05,
+      "loss": 0.5345,
+      "step": 3069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806360605283407e-05,
+      "loss": 0.6475,
+      "step": 3070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806296486278533e-05,
+      "loss": 0.6431,
+      "step": 3071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8062323672736604e-05,
+      "loss": 0.6128,
+      "step": 3072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806168248268787e-05,
+      "loss": 0.6429,
+      "step": 3073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806104129263914e-05,
+      "loss": 0.5803,
+      "step": 3074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8060400102590406e-05,
+      "loss": 0.6141,
+      "step": 3075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805975891254168e-05,
+      "loss": 0.6094,
+      "step": 3076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805911772249295e-05,
+      "loss": 0.5262,
+      "step": 3077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805847653244422e-05,
+      "loss": 0.562,
+      "step": 3078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805783534239549e-05,
+      "loss": 0.6812,
+      "step": 3079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805719415234676e-05,
+      "loss": 0.5678,
+      "step": 3080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805655296229803e-05,
+      "loss": 0.5988,
+      "step": 3081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8055911772249295e-05,
+      "loss": 0.5972,
+      "step": 3082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8055270582200566e-05,
+      "loss": 0.6466,
+      "step": 3083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805462939215183e-05,
+      "loss": 0.6161,
+      "step": 3084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80539882021031e-05,
+      "loss": 0.5933,
+      "step": 3085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8053347012054375e-05,
+      "loss": 0.6014,
+      "step": 3086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8052705822005646e-05,
+      "loss": 0.54,
+      "step": 3087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805206463195692e-05,
+      "loss": 0.6164,
+      "step": 3088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805142344190818e-05,
+      "loss": 0.5847,
+      "step": 3089
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8050782251859455e-05,
+      "loss": 0.5734,
+      "step": 3090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805014106181072e-05,
+      "loss": 0.5868,
+      "step": 3091
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804949987176199e-05,
+      "loss": 0.5161,
+      "step": 3092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804885868171326e-05,
+      "loss": 0.6001,
+      "step": 3093
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8048217491664535e-05,
+      "loss": 0.6159,
+      "step": 3094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80475763016158e-05,
+      "loss": 0.6444,
+      "step": 3095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804693511156707e-05,
+      "loss": 0.6333,
+      "step": 3096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804629392151834e-05,
+      "loss": 0.5417,
+      "step": 3097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804565273146961e-05,
+      "loss": 0.733,
+      "step": 3098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804501154142088e-05,
+      "loss": 0.5393,
+      "step": 3099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8044370351372145e-05,
+      "loss": 0.6819,
+      "step": 3100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8043729161323417e-05,
+      "loss": 0.5558,
+      "step": 3101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804308797127469e-05,
+      "loss": 0.6018,
+      "step": 3102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804244678122596e-05,
+      "loss": 0.5652,
+      "step": 3103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8041805591177225e-05,
+      "loss": 0.647,
+      "step": 3104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8041164401128497e-05,
+      "loss": 0.6136,
+      "step": 3105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804052321107977e-05,
+      "loss": 0.62,
+      "step": 3106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803988202103103e-05,
+      "loss": 0.6523,
+      "step": 3107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8039240830982305e-05,
+      "loss": 0.6305,
+      "step": 3108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803859964093358e-05,
+      "loss": 0.6652,
+      "step": 3109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803795845088485e-05,
+      "loss": 0.5564,
+      "step": 3110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803731726083611e-05,
+      "loss": 0.5607,
+      "step": 3111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8036676070787385e-05,
+      "loss": 0.633,
+      "step": 3112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803603488073865e-05,
+      "loss": 0.6865,
+      "step": 3113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803539369068992e-05,
+      "loss": 0.5189,
+      "step": 3114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8034752500641193e-05,
+      "loss": 0.6033,
+      "step": 3115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803411131059246e-05,
+      "loss": 0.565,
+      "step": 3116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803347012054373e-05,
+      "loss": 0.6253,
+      "step": 3117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8032828930495e-05,
+      "loss": 0.5689,
+      "step": 3118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8032187740446274e-05,
+      "loss": 0.6514,
+      "step": 3119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803154655039754e-05,
+      "loss": 0.6443,
+      "step": 3120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803090536034881e-05,
+      "loss": 0.6624,
+      "step": 3121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8030264170300075e-05,
+      "loss": 0.7009,
+      "step": 3122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802962298025135e-05,
+      "loss": 0.5989,
+      "step": 3123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802898179020262e-05,
+      "loss": 0.6454,
+      "step": 3124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802834060015389e-05,
+      "loss": 0.6046,
+      "step": 3125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802769941010516e-05,
+      "loss": 0.5573,
+      "step": 3126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802705822005643e-05,
+      "loss": 0.5935,
+      "step": 3127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80264170300077e-05,
+      "loss": 0.5208,
+      "step": 3128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8025775839958964e-05,
+      "loss": 0.6065,
+      "step": 3129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8025134649910235e-05,
+      "loss": 0.5683,
+      "step": 3130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80244934598615e-05,
+      "loss": 0.6327,
+      "step": 3131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802385226981277e-05,
+      "loss": 0.5938,
+      "step": 3132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8023211079764044e-05,
+      "loss": 0.5962,
+      "step": 3133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8022569889715315e-05,
+      "loss": 0.5505,
+      "step": 3134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802192869966659e-05,
+      "loss": 0.6212,
+      "step": 3135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802128750961785e-05,
+      "loss": 0.6158,
+      "step": 3136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8020646319569124e-05,
+      "loss": 0.5215,
+      "step": 3137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802000512952039e-05,
+      "loss": 0.6526,
+      "step": 3138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801936393947166e-05,
+      "loss": 0.5649,
+      "step": 3139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8018722749422925e-05,
+      "loss": 0.5454,
+      "step": 3140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8018081559374204e-05,
+      "loss": 0.6875,
+      "step": 3141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8017440369325476e-05,
+      "loss": 0.5836,
+      "step": 3142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801679917927674e-05,
+      "loss": 0.6105,
+      "step": 3143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801615798922801e-05,
+      "loss": 0.5632,
+      "step": 3144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801551679917928e-05,
+      "loss": 0.5537,
+      "step": 3145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801487560913055e-05,
+      "loss": 0.5009,
+      "step": 3146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8014234419081814e-05,
+      "loss": 0.5758,
+      "step": 3147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8013593229033086e-05,
+      "loss": 0.5773,
+      "step": 3148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801295203898436e-05,
+      "loss": 0.633,
+      "step": 3149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801231084893563e-05,
+      "loss": 0.7365,
+      "step": 3150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80116696588869e-05,
+      "loss": 0.6112,
+      "step": 3151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8011028468838166e-05,
+      "loss": 0.6926,
+      "step": 3152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801038727878944e-05,
+      "loss": 0.5993,
+      "step": 3153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80097460887407e-05,
+      "loss": 0.5915,
+      "step": 3154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8009104898691974e-05,
+      "loss": 0.535,
+      "step": 3155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800846370864324e-05,
+      "loss": 0.6594,
+      "step": 3156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800782251859452e-05,
+      "loss": 0.6401,
+      "step": 3157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800718132854578e-05,
+      "loss": 0.6396,
+      "step": 3158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8006540138497054e-05,
+      "loss": 0.5599,
+      "step": 3159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8005898948448326e-05,
+      "loss": 0.6055,
+      "step": 3160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800525775839959e-05,
+      "loss": 0.5738,
+      "step": 3161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800461656835086e-05,
+      "loss": 0.6558,
+      "step": 3162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800397537830213e-05,
+      "loss": 0.5668,
+      "step": 3163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80033341882534e-05,
+      "loss": 0.6372,
+      "step": 3164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800269299820467e-05,
+      "loss": 0.5727,
+      "step": 3165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800205180815594e-05,
+      "loss": 0.7476,
+      "step": 3166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800141061810721e-05,
+      "loss": 0.6534,
+      "step": 3167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800076942805848e-05,
+      "loss": 0.5795,
+      "step": 3168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.800012823800975e-05,
+      "loss": 0.5947,
+      "step": 3169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7999487047961016e-05,
+      "loss": 0.6103,
+      "step": 3170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.799884585791229e-05,
+      "loss": 0.595,
+      "step": 3171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.799820466786355e-05,
+      "loss": 0.5449,
+      "step": 3172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7997563477814824e-05,
+      "loss": 0.6104,
+      "step": 3173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7996922287766096e-05,
+      "loss": 0.5528,
+      "step": 3174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.799628109771737e-05,
+      "loss": 0.5726,
+      "step": 3175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.799563990766864e-05,
+      "loss": 0.6386,
+      "step": 3176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7994998717619904e-05,
+      "loss": 0.5394,
+      "step": 3177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7994357527571176e-05,
+      "loss": 0.5692,
+      "step": 3178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.799371633752244e-05,
+      "loss": 0.5951,
+      "step": 3179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.799307514747371e-05,
+      "loss": 0.5475,
+      "step": 3180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7992433957424984e-05,
+      "loss": 0.5864,
+      "step": 3181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7991792767376256e-05,
+      "loss": 0.5977,
+      "step": 3182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.799115157732752e-05,
+      "loss": 0.6622,
+      "step": 3183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.799051038727879e-05,
+      "loss": 0.5659,
+      "step": 3184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7989869197230065e-05,
+      "loss": 0.6036,
+      "step": 3185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798922800718133e-05,
+      "loss": 0.5337,
+      "step": 3186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.79885868171326e-05,
+      "loss": 0.6624,
+      "step": 3187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7987945627083866e-05,
+      "loss": 0.6045,
+      "step": 3188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798730443703514e-05,
+      "loss": 0.571,
+      "step": 3189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798666324698641e-05,
+      "loss": 0.6289,
+      "step": 3190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798602205693768e-05,
+      "loss": 0.5022,
+      "step": 3191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7985380866888946e-05,
+      "loss": 0.5648,
+      "step": 3192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798473967684022e-05,
+      "loss": 0.6449,
+      "step": 3193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798409848679149e-05,
+      "loss": 0.6586,
+      "step": 3194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7983457296742755e-05,
+      "loss": 0.6116,
+      "step": 3195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7982816106694026e-05,
+      "loss": 0.6219,
+      "step": 3196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.79821749166453e-05,
+      "loss": 0.6038,
+      "step": 3197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798153372659657e-05,
+      "loss": 0.6274,
+      "step": 3198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7980892536547835e-05,
+      "loss": 0.5436,
+      "step": 3199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7980251346499106e-05,
+      "loss": 0.58,
+      "step": 3200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797961015645037e-05,
+      "loss": 0.6542,
+      "step": 3201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797896896640164e-05,
+      "loss": 0.5981,
+      "step": 3202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7978327776352915e-05,
+      "loss": 0.6354,
+      "step": 3203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797768658630418e-05,
+      "loss": 0.5521,
+      "step": 3204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797704539625545e-05,
+      "loss": 0.6605,
+      "step": 3205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797640420620672e-05,
+      "loss": 0.5499,
+      "step": 3206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7975763016157995e-05,
+      "loss": 0.5991,
+      "step": 3207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797512182610926e-05,
+      "loss": 0.6151,
+      "step": 3208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797448063606053e-05,
+      "loss": 0.6849,
+      "step": 3209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7973839446011796e-05,
+      "loss": 0.6853,
+      "step": 3210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797319825596307e-05,
+      "loss": 0.6103,
+      "step": 3211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797255706591434e-05,
+      "loss": 0.5883,
+      "step": 3212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797191587586561e-05,
+      "loss": 0.6259,
+      "step": 3213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797127468581688e-05,
+      "loss": 0.6578,
+      "step": 3214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.797063349576815e-05,
+      "loss": 0.5454,
+      "step": 3215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796999230571942e-05,
+      "loss": 0.5899,
+      "step": 3216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7969351115670685e-05,
+      "loss": 0.6161,
+      "step": 3217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796870992562196e-05,
+      "loss": 0.5952,
+      "step": 3218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796806873557322e-05,
+      "loss": 0.647,
+      "step": 3219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796742754552449e-05,
+      "loss": 0.5909,
+      "step": 3220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7966786355475765e-05,
+      "loss": 0.6136,
+      "step": 3221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796614516542704e-05,
+      "loss": 0.612,
+      "step": 3222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796550397537831e-05,
+      "loss": 0.6103,
+      "step": 3223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796486278532957e-05,
+      "loss": 0.5713,
+      "step": 3224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7964221595280845e-05,
+      "loss": 0.5418,
+      "step": 3225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796358040523211e-05,
+      "loss": 0.6185,
+      "step": 3226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796293921518338e-05,
+      "loss": 0.6479,
+      "step": 3227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796229802513465e-05,
+      "loss": 0.6522,
+      "step": 3228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7961656835085925e-05,
+      "loss": 0.5781,
+      "step": 3229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.79610156450372e-05,
+      "loss": 0.6331,
+      "step": 3230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.796037445498846e-05,
+      "loss": 0.6201,
+      "step": 3231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7959733264939734e-05,
+      "loss": 0.6247,
+      "step": 3232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7959092074891e-05,
+      "loss": 0.5439,
+      "step": 3233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795845088484227e-05,
+      "loss": 0.6085,
+      "step": 3234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7957809694793535e-05,
+      "loss": 0.5434,
+      "step": 3235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795716850474481e-05,
+      "loss": 0.5181,
+      "step": 3236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795652731469608e-05,
+      "loss": 0.6093,
+      "step": 3237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795588612464735e-05,
+      "loss": 0.5799,
+      "step": 3238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795524493459862e-05,
+      "loss": 0.558,
+      "step": 3239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795460374454989e-05,
+      "loss": 0.6896,
+      "step": 3240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795396255450116e-05,
+      "loss": 0.6266,
+      "step": 3241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7953321364452424e-05,
+      "loss": 0.6813,
+      "step": 3242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7952680174403695e-05,
+      "loss": 0.5868,
+      "step": 3243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795203898435496e-05,
+      "loss": 0.6677,
+      "step": 3244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.795139779430623e-05,
+      "loss": 0.5816,
+      "step": 3245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7950756604257504e-05,
+      "loss": 0.6025,
+      "step": 3246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7950115414208775e-05,
+      "loss": 0.6285,
+      "step": 3247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.794947422416005e-05,
+      "loss": 0.6645,
+      "step": 3248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.794883303411131e-05,
+      "loss": 0.623,
+      "step": 3249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7948191844062584e-05,
+      "loss": 0.5449,
+      "step": 3250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.794755065401385e-05,
+      "loss": 0.624,
+      "step": 3251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794690946396512e-05,
+      "loss": 0.5597,
+      "step": 3252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794626827391639e-05,
+      "loss": 0.582,
+      "step": 3253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7945627083867664e-05,
+      "loss": 0.5266,
+      "step": 3254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794498589381893e-05,
+      "loss": 0.6048,
+      "step": 3255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79443447037702e-05,
+      "loss": 0.7219,
+      "step": 3256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794370351372147e-05,
+      "loss": 0.6589,
+      "step": 3257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794306232367274e-05,
+      "loss": 0.5728,
+      "step": 3258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794242113362401e-05,
+      "loss": 0.6379,
+      "step": 3259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7941779943575274e-05,
+      "loss": 0.6256,
+      "step": 3260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7941138753526546e-05,
+      "loss": 0.6981,
+      "step": 3261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794049756347782e-05,
+      "loss": 0.6538,
+      "step": 3262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793985637342909e-05,
+      "loss": 0.6105,
+      "step": 3263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7939215183380354e-05,
+      "loss": 0.5636,
+      "step": 3264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7938573993331626e-05,
+      "loss": 0.5981,
+      "step": 3265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79379328032829e-05,
+      "loss": 0.5728,
+      "step": 3266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793729161323416e-05,
+      "loss": 0.6081,
+      "step": 3267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7936650423185434e-05,
+      "loss": 0.5961,
+      "step": 3268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7936009233136706e-05,
+      "loss": 0.5984,
+      "step": 3269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793536804308798e-05,
+      "loss": 0.635,
+      "step": 3270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793472685303924e-05,
+      "loss": 0.608,
+      "step": 3271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7934085662990514e-05,
+      "loss": 0.5942,
+      "step": 3272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793344447294178e-05,
+      "loss": 0.6118,
+      "step": 3273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793280328289305e-05,
+      "loss": 0.5594,
+      "step": 3274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793216209284432e-05,
+      "loss": 0.6025,
+      "step": 3275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793152090279559e-05,
+      "loss": 0.6169,
+      "step": 3276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793087971274686e-05,
+      "loss": 0.6089,
+      "step": 3277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793023852269813e-05,
+      "loss": 0.6567,
+      "step": 3278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79295973326494e-05,
+      "loss": 0.5361,
+      "step": 3279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792895614260067e-05,
+      "loss": 0.6247,
+      "step": 3280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792831495255194e-05,
+      "loss": 0.5877,
+      "step": 3281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7927673762503204e-05,
+      "loss": 0.6482,
+      "step": 3282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7927032572454476e-05,
+      "loss": 0.5394,
+      "step": 3283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792639138240575e-05,
+      "loss": 0.6693,
+      "step": 3284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792575019235702e-05,
+      "loss": 0.6527,
+      "step": 3285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792510900230829e-05,
+      "loss": 0.5254,
+      "step": 3286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7924467812259556e-05,
+      "loss": 0.6128,
+      "step": 3287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792382662221083e-05,
+      "loss": 0.5416,
+      "step": 3288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792318543216209e-05,
+      "loss": 0.6424,
+      "step": 3289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7922544242113364e-05,
+      "loss": 0.5477,
+      "step": 3290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792190305206463e-05,
+      "loss": 0.5556,
+      "step": 3291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79212618620159e-05,
+      "loss": 0.58,
+      "step": 3292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792062067196717e-05,
+      "loss": 0.5974,
+      "step": 3293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7919979481918444e-05,
+      "loss": 0.5674,
+      "step": 3294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7919338291869716e-05,
+      "loss": 0.6526,
+      "step": 3295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791869710182098e-05,
+      "loss": 0.7134,
+      "step": 3296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791805591177225e-05,
+      "loss": 0.6921,
+      "step": 3297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791741472172352e-05,
+      "loss": 0.6833,
+      "step": 3298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791677353167479e-05,
+      "loss": 0.5606,
+      "step": 3299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7916132341626054e-05,
+      "loss": 0.6493,
+      "step": 3300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791549115157733e-05,
+      "loss": 0.6064,
+      "step": 3301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7914849961528605e-05,
+      "loss": 0.5726,
+      "step": 3302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791420877147987e-05,
+      "loss": 0.583,
+      "step": 3303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791356758143114e-05,
+      "loss": 0.6372,
+      "step": 3304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7912926391382406e-05,
+      "loss": 0.6086,
+      "step": 3305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791228520133368e-05,
+      "loss": 0.6623,
+      "step": 3306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791164401128494e-05,
+      "loss": 0.5057,
+      "step": 3307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7911002821236215e-05,
+      "loss": 0.5421,
+      "step": 3308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7910361631187486e-05,
+      "loss": 0.502,
+      "step": 3309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790972044113876e-05,
+      "loss": 0.5597,
+      "step": 3310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790907925109003e-05,
+      "loss": 0.6641,
+      "step": 3311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7908438061041295e-05,
+      "loss": 0.6874,
+      "step": 3312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7907796870992566e-05,
+      "loss": 0.6303,
+      "step": 3313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790715568094383e-05,
+      "loss": 0.6674,
+      "step": 3314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79065144908951e-05,
+      "loss": 0.5769,
+      "step": 3315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790587330084637e-05,
+      "loss": 0.6952,
+      "step": 3316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7905232110797647e-05,
+      "loss": 0.6051,
+      "step": 3317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790459092074891e-05,
+      "loss": 0.5959,
+      "step": 3318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790394973070018e-05,
+      "loss": 0.6388,
+      "step": 3319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7903308540651455e-05,
+      "loss": 0.5984,
+      "step": 3320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790266735060272e-05,
+      "loss": 0.5767,
+      "step": 3321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790202616055399e-05,
+      "loss": 0.6219,
+      "step": 3322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7901384970505256e-05,
+      "loss": 0.6597,
+      "step": 3323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790074378045653e-05,
+      "loss": 0.5944,
+      "step": 3324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79001025904078e-05,
+      "loss": 0.6376,
+      "step": 3325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789946140035907e-05,
+      "loss": 0.5667,
+      "step": 3326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7898820210310337e-05,
+      "loss": 0.7188,
+      "step": 3327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789817902026161e-05,
+      "loss": 0.6505,
+      "step": 3328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789753783021288e-05,
+      "loss": 0.6144,
+      "step": 3329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7896896640164145e-05,
+      "loss": 0.5842,
+      "step": 3330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789625545011542e-05,
+      "loss": 0.6254,
+      "step": 3331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789561426006668e-05,
+      "loss": 0.655,
+      "step": 3332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789497307001795e-05,
+      "loss": 0.6262,
+      "step": 3333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7894331879969225e-05,
+      "loss": 0.6092,
+      "step": 3334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78936906899205e-05,
+      "loss": 0.6073,
+      "step": 3335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789304949987176e-05,
+      "loss": 0.6226,
+      "step": 3336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7892408309823033e-05,
+      "loss": 0.6267,
+      "step": 3337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7891767119774305e-05,
+      "loss": 0.5936,
+      "step": 3338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789112592972557e-05,
+      "loss": 0.6444,
+      "step": 3339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789048473967684e-05,
+      "loss": 0.5452,
+      "step": 3340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7889843549628114e-05,
+      "loss": 0.5656,
+      "step": 3341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7889202359579385e-05,
+      "loss": 0.5852,
+      "step": 3342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788856116953065e-05,
+      "loss": 0.5939,
+      "step": 3343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788791997948192e-05,
+      "loss": 0.6323,
+      "step": 3344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788727878943319e-05,
+      "loss": 0.5372,
+      "step": 3345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788663759938446e-05,
+      "loss": 0.6205,
+      "step": 3346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788599640933573e-05,
+      "loss": 0.5579,
+      "step": 3347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7885355219286995e-05,
+      "loss": 0.6428,
+      "step": 3348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788471402923827e-05,
+      "loss": 0.5769,
+      "step": 3349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788407283918954e-05,
+      "loss": 0.6171,
+      "step": 3350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788343164914081e-05,
+      "loss": 0.6857,
+      "step": 3351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7882790459092075e-05,
+      "loss": 0.6197,
+      "step": 3352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788214926904335e-05,
+      "loss": 0.5914,
+      "step": 3353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788150807899461e-05,
+      "loss": 0.6237,
+      "step": 3354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7880866888945884e-05,
+      "loss": 0.6727,
+      "step": 3355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7880225698897155e-05,
+      "loss": 0.6071,
+      "step": 3356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787958450884843e-05,
+      "loss": 0.5476,
+      "step": 3357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78789433187997e-05,
+      "loss": 0.6158,
+      "step": 3358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7878302128750964e-05,
+      "loss": 0.558,
+      "step": 3359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7877660938702235e-05,
+      "loss": 0.5556,
+      "step": 3360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78770197486535e-05,
+      "loss": 0.662,
+      "step": 3361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787637855860477e-05,
+      "loss": 0.6168,
+      "step": 3362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7875737368556044e-05,
+      "loss": 0.5374,
+      "step": 3363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787509617850731e-05,
+      "loss": 0.506,
+      "step": 3364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787445498845858e-05,
+      "loss": 0.6628,
+      "step": 3365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787381379840985e-05,
+      "loss": 0.4524,
+      "step": 3366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7873172608361124e-05,
+      "loss": 0.6065,
+      "step": 3367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787253141831239e-05,
+      "loss": 0.6723,
+      "step": 3368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787189022826366e-05,
+      "loss": 0.6131,
+      "step": 3369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7871249038214926e-05,
+      "loss": 0.6114,
+      "step": 3370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78706078481662e-05,
+      "loss": 0.588,
+      "step": 3371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786996665811747e-05,
+      "loss": 0.6444,
+      "step": 3372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786932546806874e-05,
+      "loss": 0.5971,
+      "step": 3373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786868427802001e-05,
+      "loss": 0.5748,
+      "step": 3374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786804308797128e-05,
+      "loss": 0.6699,
+      "step": 3375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786740189792255e-05,
+      "loss": 0.6888,
+      "step": 3376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7866760707873814e-05,
+      "loss": 0.6156,
+      "step": 3377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7866119517825086e-05,
+      "loss": 0.6402,
+      "step": 3378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786547832777635e-05,
+      "loss": 0.6284,
+      "step": 3379
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786483713772762e-05,
+      "loss": 0.5217,
+      "step": 3380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7864195947678894e-05,
+      "loss": 0.6463,
+      "step": 3381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7863554757630166e-05,
+      "loss": 0.5703,
+      "step": 3382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786291356758144e-05,
+      "loss": 0.6059,
+      "step": 3383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78622723775327e-05,
+      "loss": 0.5818,
+      "step": 3384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7861631187483974e-05,
+      "loss": 0.6879,
+      "step": 3385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786098999743524e-05,
+      "loss": 0.5622,
+      "step": 3386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786034880738651e-05,
+      "loss": 0.6216,
+      "step": 3387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7859707617337776e-05,
+      "loss": 0.6751,
+      "step": 3388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7859066427289054e-05,
+      "loss": 0.5506,
+      "step": 3389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7858425237240326e-05,
+      "loss": 0.5492,
+      "step": 3390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785778404719159e-05,
+      "loss": 0.6195,
+      "step": 3391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785714285714286e-05,
+      "loss": 0.6158,
+      "step": 3392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785650166709413e-05,
+      "loss": 0.6035,
+      "step": 3393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78558604770454e-05,
+      "loss": 0.6335,
+      "step": 3394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7855219286996664e-05,
+      "loss": 0.534,
+      "step": 3395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7854578096947936e-05,
+      "loss": 0.6331,
+      "step": 3396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785393690689921e-05,
+      "loss": 0.6922,
+      "step": 3397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785329571685048e-05,
+      "loss": 0.5735,
+      "step": 3398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785265452680175e-05,
+      "loss": 0.5906,
+      "step": 3399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7852013336753016e-05,
+      "loss": 0.5334,
+      "step": 3400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785137214670429e-05,
+      "loss": 0.6192,
+      "step": 3401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785073095665555e-05,
+      "loss": 0.5569,
+      "step": 3402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7850089766606824e-05,
+      "loss": 0.5837,
+      "step": 3403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784944857655809e-05,
+      "loss": 0.6226,
+      "step": 3404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784880738650936e-05,
+      "loss": 0.5841,
+      "step": 3405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784816619646063e-05,
+      "loss": 0.5473,
+      "step": 3406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7847525006411904e-05,
+      "loss": 0.5731,
+      "step": 3407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7846883816363176e-05,
+      "loss": 0.6182,
+      "step": 3408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784624262631444e-05,
+      "loss": 0.5833,
+      "step": 3409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784560143626571e-05,
+      "loss": 0.6883,
+      "step": 3410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784496024621698e-05,
+      "loss": 0.6945,
+      "step": 3411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784431905616825e-05,
+      "loss": 0.6704,
+      "step": 3412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784367786611952e-05,
+      "loss": 0.5589,
+      "step": 3413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784303667607079e-05,
+      "loss": 0.5927,
+      "step": 3414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784239548602206e-05,
+      "loss": 0.6042,
+      "step": 3415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784175429597333e-05,
+      "loss": 0.6035,
+      "step": 3416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78411131059246e-05,
+      "loss": 0.6054,
+      "step": 3417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7840471915875866e-05,
+      "loss": 0.6665,
+      "step": 3418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783983072582714e-05,
+      "loss": 0.651,
+      "step": 3419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78391895357784e-05,
+      "loss": 0.6212,
+      "step": 3420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7838548345729675e-05,
+      "loss": 0.5811,
+      "step": 3421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7837907155680946e-05,
+      "loss": 0.5158,
+      "step": 3422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783726596563222e-05,
+      "loss": 0.5397,
+      "step": 3423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783662477558348e-05,
+      "loss": 0.5386,
+      "step": 3424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7835983585534755e-05,
+      "loss": 0.5972,
+      "step": 3425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7835342395486026e-05,
+      "loss": 0.6178,
+      "step": 3426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783470120543729e-05,
+      "loss": 0.5773,
+      "step": 3427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783406001538856e-05,
+      "loss": 0.6169,
+      "step": 3428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7833418825339835e-05,
+      "loss": 0.6272,
+      "step": 3429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7832777635291107e-05,
+      "loss": 0.5746,
+      "step": 3430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783213644524237e-05,
+      "loss": 0.6819,
+      "step": 3431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783149525519364e-05,
+      "loss": 0.6359,
+      "step": 3432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783085406514491e-05,
+      "loss": 0.6052,
+      "step": 3433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783021287509618e-05,
+      "loss": 0.5729,
+      "step": 3434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782957168504745e-05,
+      "loss": 0.5986,
+      "step": 3435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7828930494998717e-05,
+      "loss": 0.5815,
+      "step": 3436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782828930494999e-05,
+      "loss": 0.5886,
+      "step": 3437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782764811490126e-05,
+      "loss": 0.6268,
+      "step": 3438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782700692485253e-05,
+      "loss": 0.6803,
+      "step": 3439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7826365734803797e-05,
+      "loss": 0.5337,
+      "step": 3440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782572454475507e-05,
+      "loss": 0.6191,
+      "step": 3441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782508335470633e-05,
+      "loss": 0.5815,
+      "step": 3442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7824442164657605e-05,
+      "loss": 0.5757,
+      "step": 3443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782380097460888e-05,
+      "loss": 0.6882,
+      "step": 3444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782315978456015e-05,
+      "loss": 0.6204,
+      "step": 3445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782251859451142e-05,
+      "loss": 0.5776,
+      "step": 3446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7821877404462685e-05,
+      "loss": 0.5166,
+      "step": 3447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782123621441396e-05,
+      "loss": 0.6135,
+      "step": 3448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782059502436522e-05,
+      "loss": 0.6128,
+      "step": 3449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7819953834316493e-05,
+      "loss": 0.7172,
+      "step": 3450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781931264426776e-05,
+      "loss": 0.576,
+      "step": 3451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781867145421903e-05,
+      "loss": 0.6315,
+      "step": 3452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78180302641703e-05,
+      "loss": 0.5707,
+      "step": 3453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7817389074121574e-05,
+      "loss": 0.6101,
+      "step": 3454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7816747884072845e-05,
+      "loss": 0.5604,
+      "step": 3455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781610669402411e-05,
+      "loss": 0.5777,
+      "step": 3456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781546550397538e-05,
+      "loss": 0.5303,
+      "step": 3457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781482431392665e-05,
+      "loss": 0.5472,
+      "step": 3458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781418312387792e-05,
+      "loss": 0.5932,
+      "step": 3459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7813541933829184e-05,
+      "loss": 0.5833,
+      "step": 3460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781290074378046e-05,
+      "loss": 0.6528,
+      "step": 3461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7812259553731734e-05,
+      "loss": 0.578,
+      "step": 3462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7811618363683e-05,
+      "loss": 0.536,
+      "step": 3463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781097717363427e-05,
+      "loss": 0.5432,
+      "step": 3464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7810335983585535e-05,
+      "loss": 0.5785,
+      "step": 3465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780969479353681e-05,
+      "loss": 0.6791,
+      "step": 3466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780905360348807e-05,
+      "loss": 0.5796,
+      "step": 3467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7808412413439344e-05,
+      "loss": 0.5814,
+      "step": 3468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7807771223390615e-05,
+      "loss": 0.5134,
+      "step": 3469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780713003334189e-05,
+      "loss": 0.5947,
+      "step": 3470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780648884329316e-05,
+      "loss": 0.4875,
+      "step": 3471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7805847653244424e-05,
+      "loss": 0.6572,
+      "step": 3472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7805206463195695e-05,
+      "loss": 0.5536,
+      "step": 3473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780456527314696e-05,
+      "loss": 0.5897,
+      "step": 3474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780392408309823e-05,
+      "loss": 0.5771,
+      "step": 3475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78032828930495e-05,
+      "loss": 0.6344,
+      "step": 3476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7802641703000776e-05,
+      "loss": 0.6524,
+      "step": 3477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780200051295204e-05,
+      "loss": 0.6201,
+      "step": 3478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780135932290331e-05,
+      "loss": 0.5894,
+      "step": 3479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7800718132854584e-05,
+      "loss": 0.6245,
+      "step": 3480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780007694280585e-05,
+      "loss": 0.5745,
+      "step": 3481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779943575275712e-05,
+      "loss": 0.7138,
+      "step": 3482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7798794562708386e-05,
+      "loss": 0.5445,
+      "step": 3483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779815337265966e-05,
+      "loss": 0.5681,
+      "step": 3484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779751218261093e-05,
+      "loss": 0.6588,
+      "step": 3485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.77968709925622e-05,
+      "loss": 0.6091,
+      "step": 3486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7796229802513466e-05,
+      "loss": 0.591,
+      "step": 3487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779558861246474e-05,
+      "loss": 0.6504,
+      "step": 3488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779494742241601e-05,
+      "loss": 0.5786,
+      "step": 3489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7794306232367274e-05,
+      "loss": 0.5509,
+      "step": 3490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7793665042318546e-05,
+      "loss": 0.556,
+      "step": 3491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779302385226981e-05,
+      "loss": 0.5571,
+      "step": 3492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779238266222108e-05,
+      "loss": 0.6531,
+      "step": 3493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7791741472172354e-05,
+      "loss": 0.6009,
+      "step": 3494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7791100282123626e-05,
+      "loss": 0.6239,
+      "step": 3495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779045909207489e-05,
+      "loss": 0.5676,
+      "step": 3496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778981790202616e-05,
+      "loss": 0.6384,
+      "step": 3497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7789176711977434e-05,
+      "loss": 0.5087,
+      "step": 3498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.77885355219287e-05,
+      "loss": 0.5241,
+      "step": 3499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778789433187997e-05,
+      "loss": 0.6632,
+      "step": 3500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778725314183124e-05,
+      "loss": 0.605,
+      "step": 3501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7786611951782514e-05,
+      "loss": 0.5453,
+      "step": 3502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778597076173378e-05,
+      "loss": 0.5842,
+      "step": 3503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778532957168505e-05,
+      "loss": 0.6038,
+      "step": 3504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7784688381636316e-05,
+      "loss": 0.636,
+      "step": 3505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778404719158759e-05,
+      "loss": 0.6276,
+      "step": 3506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778340600153886e-05,
+      "loss": 0.5201,
+      "step": 3507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7782764811490124e-05,
+      "loss": 0.6036,
+      "step": 3508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7782123621441396e-05,
+      "loss": 0.6309,
+      "step": 3509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778148243139267e-05,
+      "loss": 0.5521,
+      "step": 3510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778084124134394e-05,
+      "loss": 0.5682,
+      "step": 3511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7780200051295204e-05,
+      "loss": 0.5951,
+      "step": 3512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7779558861246476e-05,
+      "loss": 0.5931,
+      "step": 3513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777891767119774e-05,
+      "loss": 0.6143,
+      "step": 3514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777827648114901e-05,
+      "loss": 0.6134,
+      "step": 3515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7777635291100284e-05,
+      "loss": 0.5735,
+      "step": 3516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7776994101051556e-05,
+      "loss": 0.5178,
+      "step": 3517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777635291100283e-05,
+      "loss": 0.5869,
+      "step": 3518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777571172095409e-05,
+      "loss": 0.5633,
+      "step": 3519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7775070530905365e-05,
+      "loss": 0.6178,
+      "step": 3520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777442934085663e-05,
+      "loss": 0.652,
+      "step": 3521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77737881508079e-05,
+      "loss": 0.539,
+      "step": 3522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7773146960759166e-05,
+      "loss": 0.5502,
+      "step": 3523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777250577071044e-05,
+      "loss": 0.6021,
+      "step": 3524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777186458066171e-05,
+      "loss": 0.7079,
+      "step": 3525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777122339061298e-05,
+      "loss": 0.6137,
+      "step": 3526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777058220056425e-05,
+      "loss": 0.6351,
+      "step": 3527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776994101051552e-05,
+      "loss": 0.5177,
+      "step": 3528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776929982046679e-05,
+      "loss": 0.6541,
+      "step": 3529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7768658630418055e-05,
+      "loss": 0.6157,
+      "step": 3530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7768017440369326e-05,
+      "loss": 0.7362,
+      "step": 3531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776737625032059e-05,
+      "loss": 0.6922,
+      "step": 3532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776673506027187e-05,
+      "loss": 0.6382,
+      "step": 3533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776609387022314e-05,
+      "loss": 0.58,
+      "step": 3534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7765452680174406e-05,
+      "loss": 0.7287,
+      "step": 3535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776481149012568e-05,
+      "loss": 0.598,
+      "step": 3536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776417030007694e-05,
+      "loss": 0.6132,
+      "step": 3537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7763529110028215e-05,
+      "loss": 0.6001,
+      "step": 3538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776288791997948e-05,
+      "loss": 0.5689,
+      "step": 3539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776224672993075e-05,
+      "loss": 0.5662,
+      "step": 3540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776160553988202e-05,
+      "loss": 0.5593,
+      "step": 3541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7760964349833295e-05,
+      "loss": 0.6265,
+      "step": 3542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7760323159784567e-05,
+      "loss": 0.6231,
+      "step": 3543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775968196973583e-05,
+      "loss": 0.6123,
+      "step": 3544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77590407796871e-05,
+      "loss": 0.7177,
+      "step": 3545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775839958963837e-05,
+      "loss": 0.5221,
+      "step": 3546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775775839958964e-05,
+      "loss": 0.5669,
+      "step": 3547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7757117209540905e-05,
+      "loss": 0.6125,
+      "step": 3548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775647601949218e-05,
+      "loss": 0.6091,
+      "step": 3549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775583482944345e-05,
+      "loss": 0.6057,
+      "step": 3550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775519363939472e-05,
+      "loss": 0.5907,
+      "step": 3551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775455244934599e-05,
+      "loss": 0.5864,
+      "step": 3552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775391125929726e-05,
+      "loss": 0.6337,
+      "step": 3553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775327006924853e-05,
+      "loss": 0.5737,
+      "step": 3554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775262887919979e-05,
+      "loss": 0.5324,
+      "step": 3555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7751987689151065e-05,
+      "loss": 0.4973,
+      "step": 3556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775134649910234e-05,
+      "loss": 0.6675,
+      "step": 3557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775070530905361e-05,
+      "loss": 0.6196,
+      "step": 3558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775006411900488e-05,
+      "loss": 0.5964,
+      "step": 3559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7749422928956145e-05,
+      "loss": 0.6583,
+      "step": 3560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774878173890742e-05,
+      "loss": 0.594,
+      "step": 3561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774814054885868e-05,
+      "loss": 0.5299,
+      "step": 3562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7747499358809953e-05,
+      "loss": 0.8001,
+      "step": 3563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774685816876122e-05,
+      "loss": 0.5688,
+      "step": 3564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77462169787125e-05,
+      "loss": 0.6908,
+      "step": 3565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774557578866376e-05,
+      "loss": 0.5371,
+      "step": 3566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7744934598615034e-05,
+      "loss": 0.6379,
+      "step": 3567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7744293408566305e-05,
+      "loss": 0.6161,
+      "step": 3568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774365221851757e-05,
+      "loss": 0.5356,
+      "step": 3569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774301102846884e-05,
+      "loss": 0.6512,
+      "step": 3570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774236983842011e-05,
+      "loss": 0.6013,
+      "step": 3571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774172864837138e-05,
+      "loss": 0.6948,
+      "step": 3572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774108745832265e-05,
+      "loss": 0.5854,
+      "step": 3573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774044626827392e-05,
+      "loss": 0.6865,
+      "step": 3574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773980507822519e-05,
+      "loss": 0.6222,
+      "step": 3575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773916388817646e-05,
+      "loss": 0.5871,
+      "step": 3576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773852269812773e-05,
+      "loss": 0.6247,
+      "step": 3577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7737881508078995e-05,
+      "loss": 0.621,
+      "step": 3578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773724031803027e-05,
+      "loss": 0.5371,
+      "step": 3579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773659912798153e-05,
+      "loss": 0.5661,
+      "step": 3580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7735957937932804e-05,
+      "loss": 0.6779,
+      "step": 3581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7735316747884075e-05,
+      "loss": 0.5849,
+      "step": 3582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773467555783535e-05,
+      "loss": 0.6428,
+      "step": 3583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773403436778661e-05,
+      "loss": 0.6095,
+      "step": 3584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7733393177737884e-05,
+      "loss": 0.591,
+      "step": 3585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7732751987689156e-05,
+      "loss": 0.5325,
+      "step": 3586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773211079764042e-05,
+      "loss": 0.5921,
+      "step": 3587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773146960759169e-05,
+      "loss": 0.6079,
+      "step": 3588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7730828417542964e-05,
+      "loss": 0.5376,
+      "step": 3589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7730187227494236e-05,
+      "loss": 0.6749,
+      "step": 3590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77295460374455e-05,
+      "loss": 0.5283,
+      "step": 3591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772890484739677e-05,
+      "loss": 0.5894,
+      "step": 3592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772826365734804e-05,
+      "loss": 0.6036,
+      "step": 3593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772762246729931e-05,
+      "loss": 0.6062,
+      "step": 3594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772698127725058e-05,
+      "loss": 0.6397,
+      "step": 3595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7726340087201846e-05,
+      "loss": 0.6739,
+      "step": 3596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772569889715312e-05,
+      "loss": 0.6572,
+      "step": 3597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772505770710439e-05,
+      "loss": 0.5768,
+      "step": 3598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772441651705566e-05,
+      "loss": 0.6443,
+      "step": 3599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7723775327006926e-05,
+      "loss": 0.6456,
+      "step": 3600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77231341369582e-05,
+      "loss": 0.5903,
+      "step": 3601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772249294690946e-05,
+      "loss": 0.6364,
+      "step": 3602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7721851756860734e-05,
+      "loss": 0.6835,
+      "step": 3603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7721210566812006e-05,
+      "loss": 0.5237,
+      "step": 3604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772056937676328e-05,
+      "loss": 0.5233,
+      "step": 3605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771992818671455e-05,
+      "loss": 0.6058,
+      "step": 3606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7719286996665814e-05,
+      "loss": 0.735,
+      "step": 3607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7718645806617086e-05,
+      "loss": 0.6085,
+      "step": 3608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771800461656835e-05,
+      "loss": 0.5757,
+      "step": 3609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771736342651962e-05,
+      "loss": 0.5785,
+      "step": 3610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771672223647089e-05,
+      "loss": 0.5943,
+      "step": 3611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771608104642216e-05,
+      "loss": 0.5703,
+      "step": 3612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771543985637343e-05,
+      "loss": 0.638,
+      "step": 3613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77147986663247e-05,
+      "loss": 0.5821,
+      "step": 3614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7714157476275974e-05,
+      "loss": 0.6589,
+      "step": 3615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771351628622724e-05,
+      "loss": 0.7318,
+      "step": 3616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771287509617851e-05,
+      "loss": 0.6087,
+      "step": 3617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7712233906129776e-05,
+      "loss": 0.6587,
+      "step": 3618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771159271608105e-05,
+      "loss": 0.6079,
+      "step": 3619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771095152603231e-05,
+      "loss": 0.5609,
+      "step": 3620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771031033598359e-05,
+      "loss": 0.5956,
+      "step": 3621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770966914593486e-05,
+      "loss": 0.5719,
+      "step": 3622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770902795588613e-05,
+      "loss": 0.6246,
+      "step": 3623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77083867658374e-05,
+      "loss": 0.6679,
+      "step": 3624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7707745575788664e-05,
+      "loss": 0.5433,
+      "step": 3625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7707104385739936e-05,
+      "loss": 0.6139,
+      "step": 3626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77064631956912e-05,
+      "loss": 0.6552,
+      "step": 3627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770582200564247e-05,
+      "loss": 0.5621,
+      "step": 3628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7705180815593744e-05,
+      "loss": 0.5299,
+      "step": 3629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7704539625545016e-05,
+      "loss": 0.6787,
+      "step": 3630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770389843549629e-05,
+      "loss": 0.5948,
+      "step": 3631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770325724544755e-05,
+      "loss": 0.6081,
+      "step": 3632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7702616055398825e-05,
+      "loss": 0.6024,
+      "step": 3633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770197486535009e-05,
+      "loss": 0.6333,
+      "step": 3634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770133367530136e-05,
+      "loss": 0.681,
+      "step": 3635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7700692485252626e-05,
+      "loss": 0.6487,
+      "step": 3636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7700051295203905e-05,
+      "loss": 0.5903,
+      "step": 3637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769941010515517e-05,
+      "loss": 0.5308,
+      "step": 3638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769876891510644e-05,
+      "loss": 0.5722,
+      "step": 3639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769812772505771e-05,
+      "loss": 0.5762,
+      "step": 3640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769748653500898e-05,
+      "loss": 0.6293,
+      "step": 3641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769684534496025e-05,
+      "loss": 0.6073,
+      "step": 3642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7696204154911515e-05,
+      "loss": 0.5185,
+      "step": 3643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7695562964862786e-05,
+      "loss": 0.5903,
+      "step": 3644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769492177481406e-05,
+      "loss": 0.5793,
+      "step": 3645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769428058476533e-05,
+      "loss": 0.6735,
+      "step": 3646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7693639394716595e-05,
+      "loss": 0.6043,
+      "step": 3647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7692998204667866e-05,
+      "loss": 0.5582,
+      "step": 3648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769235701461914e-05,
+      "loss": 0.5907,
+      "step": 3649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76917158245704e-05,
+      "loss": 0.5973,
+      "step": 3650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7691074634521675e-05,
+      "loss": 0.6262,
+      "step": 3651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769043344447294e-05,
+      "loss": 0.6133,
+      "step": 3652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768979225442421e-05,
+      "loss": 0.5925,
+      "step": 3653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768915106437548e-05,
+      "loss": 0.589,
+      "step": 3654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7688509874326755e-05,
+      "loss": 0.6423,
+      "step": 3655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768786868427802e-05,
+      "loss": 0.5545,
+      "step": 3656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768722749422929e-05,
+      "loss": 0.5424,
+      "step": 3657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768658630418056e-05,
+      "loss": 0.7075,
+      "step": 3658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768594511413183e-05,
+      "loss": 0.5181,
+      "step": 3659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76853039240831e-05,
+      "loss": 0.5054,
+      "step": 3660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768466273403437e-05,
+      "loss": 0.5796,
+      "step": 3661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768402154398564e-05,
+      "loss": 0.5919,
+      "step": 3662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768338035393691e-05,
+      "loss": 0.604,
+      "step": 3663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768273916388818e-05,
+      "loss": 0.6968,
+      "step": 3664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7682097973839445e-05,
+      "loss": 0.5666,
+      "step": 3665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768145678379072e-05,
+      "loss": 0.6291,
+      "step": 3666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768081559374199e-05,
+      "loss": 0.6046,
+      "step": 3667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768017440369325e-05,
+      "loss": 0.5917,
+      "step": 3668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7679533213644525e-05,
+      "loss": 0.5944,
+      "step": 3669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76788920235958e-05,
+      "loss": 0.576,
+      "step": 3670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767825083354707e-05,
+      "loss": 0.5638,
+      "step": 3671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7677609643498333e-05,
+      "loss": 0.651,
+      "step": 3672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7676968453449605e-05,
+      "loss": 0.6104,
+      "step": 3673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767632726340087e-05,
+      "loss": 0.624,
+      "step": 3674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767568607335214e-05,
+      "loss": 0.5468,
+      "step": 3675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7675044883303414e-05,
+      "loss": 0.6536,
+      "step": 3676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7674403693254685e-05,
+      "loss": 0.5978,
+      "step": 3677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767376250320596e-05,
+      "loss": 0.6453,
+      "step": 3678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767312131315722e-05,
+      "loss": 0.5991,
+      "step": 3679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7672480123108494e-05,
+      "loss": 0.5594,
+      "step": 3680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767183893305976e-05,
+      "loss": 0.6364,
+      "step": 3681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767119774301103e-05,
+      "loss": 0.6152,
+      "step": 3682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7670556552962295e-05,
+      "loss": 0.6597,
+      "step": 3683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766991536291357e-05,
+      "loss": 0.5756,
+      "step": 3684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766927417286484e-05,
+      "loss": 0.561,
+      "step": 3685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766863298281611e-05,
+      "loss": 0.5846,
+      "step": 3686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766799179276738e-05,
+      "loss": 0.5573,
+      "step": 3687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766735060271865e-05,
+      "loss": 0.5545,
+      "step": 3688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766670941266992e-05,
+      "loss": 0.6391,
+      "step": 3689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7666068222621184e-05,
+      "loss": 0.5885,
+      "step": 3690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7665427032572455e-05,
+      "loss": 0.5657,
+      "step": 3691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766478584252372e-05,
+      "loss": 0.5892,
+      "step": 3692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7664144652475e-05,
+      "loss": 0.5855,
+      "step": 3693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766350346242627e-05,
+      "loss": 0.5808,
+      "step": 3694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7662862272377535e-05,
+      "loss": 0.5733,
+      "step": 3695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766222108232881e-05,
+      "loss": 0.6601,
+      "step": 3696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766157989228007e-05,
+      "loss": 0.6468,
+      "step": 3697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7660938702231344e-05,
+      "loss": 0.598,
+      "step": 3698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766029751218261e-05,
+      "loss": 0.5414,
+      "step": 3699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765965632213388e-05,
+      "loss": 0.5842,
+      "step": 3700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765901513208515e-05,
+      "loss": 0.5421,
+      "step": 3701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7658373942036424e-05,
+      "loss": 0.5685,
+      "step": 3702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7657732751987696e-05,
+      "loss": 0.5815,
+      "step": 3703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765709156193896e-05,
+      "loss": 0.5701,
+      "step": 3704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765645037189023e-05,
+      "loss": 0.6203,
+      "step": 3705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76558091818415e-05,
+      "loss": 0.5688,
+      "step": 3706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765516799179277e-05,
+      "loss": 0.6002,
+      "step": 3707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7654526801744034e-05,
+      "loss": 0.6083,
+      "step": 3708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765388561169531e-05,
+      "loss": 0.5702,
+      "step": 3709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765324442164658e-05,
+      "loss": 0.6105,
+      "step": 3710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765260323159785e-05,
+      "loss": 0.7105,
+      "step": 3711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765196204154912e-05,
+      "loss": 0.648,
+      "step": 3712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7651320851500386e-05,
+      "loss": 0.5784,
+      "step": 3713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765067966145166e-05,
+      "loss": 0.6326,
+      "step": 3714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765003847140292e-05,
+      "loss": 0.6485,
+      "step": 3715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7649397281354194e-05,
+      "loss": 0.5842,
+      "step": 3716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7648756091305466e-05,
+      "loss": 0.5558,
+      "step": 3717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764811490125674e-05,
+      "loss": 0.634,
+      "step": 3718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7647473711208e-05,
+      "loss": 0.568,
+      "step": 3719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7646832521159274e-05,
+      "loss": 0.5813,
+      "step": 3720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7646191331110546e-05,
+      "loss": 0.6158,
+      "step": 3721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764555014106181e-05,
+      "loss": 0.6346,
+      "step": 3722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764490895101308e-05,
+      "loss": 0.6262,
+      "step": 3723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764426776096435e-05,
+      "loss": 0.5826,
+      "step": 3724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7643626570915626e-05,
+      "loss": 0.6156,
+      "step": 3725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764298538086689e-05,
+      "loss": 0.5721,
+      "step": 3726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764234419081816e-05,
+      "loss": 0.5835,
+      "step": 3727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764170300076943e-05,
+      "loss": 0.6688,
+      "step": 3728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76410618107207e-05,
+      "loss": 0.6833,
+      "step": 3729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764042062067197e-05,
+      "loss": 0.53,
+      "step": 3730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7639779430623236e-05,
+      "loss": 0.6307,
+      "step": 3731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763913824057451e-05,
+      "loss": 0.6052,
+      "step": 3732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763849705052578e-05,
+      "loss": 0.5855,
+      "step": 3733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763785586047705e-05,
+      "loss": 0.5722,
+      "step": 3734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7637214670428316e-05,
+      "loss": 0.5885,
+      "step": 3735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763657348037959e-05,
+      "loss": 0.7122,
+      "step": 3736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763593229033085e-05,
+      "loss": 0.6388,
+      "step": 3737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7635291100282124e-05,
+      "loss": 0.5595,
+      "step": 3738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7634649910233396e-05,
+      "loss": 0.6562,
+      "step": 3739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763400872018466e-05,
+      "loss": 0.5922,
+      "step": 3740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763336753013593e-05,
+      "loss": 0.6749,
+      "step": 3741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7632726340087205e-05,
+      "loss": 0.5722,
+      "step": 3742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7632085150038476e-05,
+      "loss": 0.6957,
+      "step": 3743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763144395998974e-05,
+      "loss": 0.613,
+      "step": 3744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763080276994101e-05,
+      "loss": 0.6239,
+      "step": 3745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7630161579892285e-05,
+      "loss": 0.5591,
+      "step": 3746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762952038984355e-05,
+      "loss": 0.624,
+      "step": 3747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762887919979482e-05,
+      "loss": 0.6706,
+      "step": 3748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762823800974609e-05,
+      "loss": 0.6047,
+      "step": 3749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7627596819697365e-05,
+      "loss": 0.6901,
+      "step": 3750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762695562964863e-05,
+      "loss": 0.6174,
+      "step": 3751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76263144395999e-05,
+      "loss": 0.5557,
+      "step": 3752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7625673249551166e-05,
+      "loss": 0.5845,
+      "step": 3753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762503205950244e-05,
+      "loss": 0.5389,
+      "step": 3754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762439086945371e-05,
+      "loss": 0.6193,
+      "step": 3755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7623749679404975e-05,
+      "loss": 0.6478,
+      "step": 3756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7623108489356246e-05,
+      "loss": 0.6177,
+      "step": 3757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762246729930752e-05,
+      "loss": 0.6073,
+      "step": 3758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762182610925879e-05,
+      "loss": 0.5432,
+      "step": 3759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7621184919210055e-05,
+      "loss": 0.6274,
+      "step": 3760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7620543729161326e-05,
+      "loss": 0.5794,
+      "step": 3761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761990253911259e-05,
+      "loss": 0.5988,
+      "step": 3762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761926134906386e-05,
+      "loss": 0.5003,
+      "step": 3763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7618620159015135e-05,
+      "loss": 0.5524,
+      "step": 3764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7617978968966407e-05,
+      "loss": 0.5089,
+      "step": 3765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761733777891768e-05,
+      "loss": 0.4822,
+      "step": 3766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761669658886894e-05,
+      "loss": 0.5658,
+      "step": 3767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7616055398820215e-05,
+      "loss": 0.5958,
+      "step": 3768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761541420877148e-05,
+      "loss": 0.5887,
+      "step": 3769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761477301872275e-05,
+      "loss": 0.6046,
+      "step": 3770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7614131828674017e-05,
+      "loss": 0.5866,
+      "step": 3771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761349063862529e-05,
+      "loss": 0.6789,
+      "step": 3772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761284944857656e-05,
+      "loss": 0.5824,
+      "step": 3773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761220825852783e-05,
+      "loss": 0.627,
+      "step": 3774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7611567068479103e-05,
+      "loss": 0.6081,
+      "step": 3775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761092587843037e-05,
+      "loss": 0.5903,
+      "step": 3776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761028468838164e-05,
+      "loss": 0.4826,
+      "step": 3777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7609643498332905e-05,
+      "loss": 0.5433,
+      "step": 3778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760900230828418e-05,
+      "loss": 0.6252,
+      "step": 3779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760836111823544e-05,
+      "loss": 0.818,
+      "step": 3780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760771992818672e-05,
+      "loss": 0.5378,
+      "step": 3781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760707873813799e-05,
+      "loss": 0.555,
+      "step": 3782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760643754808926e-05,
+      "loss": 0.6524,
+      "step": 3783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760579635804053e-05,
+      "loss": 0.5999,
+      "step": 3784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7605155167991793e-05,
+      "loss": 0.5832,
+      "step": 3785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7604513977943065e-05,
+      "loss": 0.5493,
+      "step": 3786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760387278789433e-05,
+      "loss": 0.6089,
+      "step": 3787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.76032315978456e-05,
+      "loss": 0.6015,
+      "step": 3788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7602590407796874e-05,
+      "loss": 0.6493,
+      "step": 3789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7601949217748145e-05,
+      "loss": 0.612,
+      "step": 3790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760130802769942e-05,
+      "loss": 0.7069,
+      "step": 3791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760066683765068e-05,
+      "loss": 0.5458,
+      "step": 3792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7600025647601954e-05,
+      "loss": 0.6337,
+      "step": 3793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759938445755322e-05,
+      "loss": 0.5398,
+      "step": 3794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759874326750449e-05,
+      "loss": 0.6164,
+      "step": 3795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7598102077455755e-05,
+      "loss": 0.4931,
+      "step": 3796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7597460887407034e-05,
+      "loss": 0.6449,
+      "step": 3797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75968196973583e-05,
+      "loss": 0.5984,
+      "step": 3798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759617850730957e-05,
+      "loss": 0.6085,
+      "step": 3799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759553731726084e-05,
+      "loss": 0.5992,
+      "step": 3800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759489612721211e-05,
+      "loss": 0.6327,
+      "step": 3801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759425493716338e-05,
+      "loss": 0.6234,
+      "step": 3802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7593613747114644e-05,
+      "loss": 0.5556,
+      "step": 3803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7592972557065915e-05,
+      "loss": 0.5112,
+      "step": 3804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759233136701719e-05,
+      "loss": 0.5835,
+      "step": 3805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759169017696846e-05,
+      "loss": 0.6742,
+      "step": 3806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7591048986919724e-05,
+      "loss": 0.5493,
+      "step": 3807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7590407796870996e-05,
+      "loss": 0.6312,
+      "step": 3808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758976660682227e-05,
+      "loss": 0.6738,
+      "step": 3809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758912541677353e-05,
+      "loss": 0.6913,
+      "step": 3810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7588484226724804e-05,
+      "loss": 0.639,
+      "step": 3811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758784303667607e-05,
+      "loss": 0.6251,
+      "step": 3812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758720184662735e-05,
+      "loss": 0.7062,
+      "step": 3813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758656065657861e-05,
+      "loss": 0.65,
+      "step": 3814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7585919466529884e-05,
+      "loss": 0.5746,
+      "step": 3815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758527827648115e-05,
+      "loss": 0.6088,
+      "step": 3816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758463708643242e-05,
+      "loss": 0.6471,
+      "step": 3817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758399589638369e-05,
+      "loss": 0.5595,
+      "step": 3818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758335470633496e-05,
+      "loss": 0.579,
+      "step": 3819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758271351628623e-05,
+      "loss": 0.6898,
+      "step": 3820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75820723262375e-05,
+      "loss": 0.5021,
+      "step": 3821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758143113618877e-05,
+      "loss": 0.6629,
+      "step": 3822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758078994614004e-05,
+      "loss": 0.6832,
+      "step": 3823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758014875609131e-05,
+      "loss": 0.5242,
+      "step": 3824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7579507566042574e-05,
+      "loss": 0.7327,
+      "step": 3825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7578866375993846e-05,
+      "loss": 0.6846,
+      "step": 3826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757822518594512e-05,
+      "loss": 0.6417,
+      "step": 3827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757758399589638e-05,
+      "loss": 0.6458,
+      "step": 3828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7576942805847654e-05,
+      "loss": 0.5937,
+      "step": 3829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7576301615798926e-05,
+      "loss": 0.6296,
+      "step": 3830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75756604257502e-05,
+      "loss": 0.5764,
+      "step": 3831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757501923570146e-05,
+      "loss": 0.5974,
+      "step": 3832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7574378045652734e-05,
+      "loss": 0.6011,
+      "step": 3833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7573736855604e-05,
+      "loss": 0.535,
+      "step": 3834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757309566555527e-05,
+      "loss": 0.6232,
+      "step": 3835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757245447550654e-05,
+      "loss": 0.6218,
+      "step": 3836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7571813285457814e-05,
+      "loss": 0.6287,
+      "step": 3837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7571172095409086e-05,
+      "loss": 0.6297,
+      "step": 3838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757053090536035e-05,
+      "loss": 0.6171,
+      "step": 3839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756988971531162e-05,
+      "loss": 0.6114,
+      "step": 3840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756924852526289e-05,
+      "loss": 0.6296,
+      "step": 3841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756860733521416e-05,
+      "loss": 0.5737,
+      "step": 3842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7567966145165424e-05,
+      "loss": 0.5536,
+      "step": 3843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7567324955116696e-05,
+      "loss": 0.6645,
+      "step": 3844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756668376506797e-05,
+      "loss": 0.7116,
+      "step": 3845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756604257501924e-05,
+      "loss": 0.5691,
+      "step": 3846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756540138497051e-05,
+      "loss": 0.6229,
+      "step": 3847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7564760194921776e-05,
+      "loss": 0.645,
+      "step": 3848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756411900487305e-05,
+      "loss": 0.5911,
+      "step": 3849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756347781482431e-05,
+      "loss": 0.5432,
+      "step": 3850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7562836624775584e-05,
+      "loss": 0.6063,
+      "step": 3851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7562195434726856e-05,
+      "loss": 0.6019,
+      "step": 3852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756155424467813e-05,
+      "loss": 0.5553,
+      "step": 3853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75609130546294e-05,
+      "loss": 0.5842,
+      "step": 3854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7560271864580665e-05,
+      "loss": 0.5967,
+      "step": 3855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7559630674531936e-05,
+      "loss": 0.5497,
+      "step": 3856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75589894844832e-05,
+      "loss": 0.618,
+      "step": 3857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755834829443447e-05,
+      "loss": 0.554,
+      "step": 3858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755770710438574e-05,
+      "loss": 0.6885,
+      "step": 3859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755706591433701e-05,
+      "loss": 0.5567,
+      "step": 3860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755642472428828e-05,
+      "loss": 0.5831,
+      "step": 3861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755578353423955e-05,
+      "loss": 0.6449,
+      "step": 3862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7555142344190825e-05,
+      "loss": 0.591,
+      "step": 3863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755450115414209e-05,
+      "loss": 0.5751,
+      "step": 3864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755385996409336e-05,
+      "loss": 0.6159,
+      "step": 3865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7553218774044626e-05,
+      "loss": 0.704,
+      "step": 3866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75525775839959e-05,
+      "loss": 0.5487,
+      "step": 3867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755193639394716e-05,
+      "loss": 0.6087,
+      "step": 3868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755129520389844e-05,
+      "loss": 0.6618,
+      "step": 3869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7550654013849706e-05,
+      "loss": 0.6749,
+      "step": 3870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755001282380098e-05,
+      "loss": 0.6147,
+      "step": 3871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754937163375225e-05,
+      "loss": 0.5972,
+      "step": 3872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7548730443703515e-05,
+      "loss": 0.6098,
+      "step": 3873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7548089253654786e-05,
+      "loss": 0.6426,
+      "step": 3874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754744806360605e-05,
+      "loss": 0.719,
+      "step": 3875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754680687355732e-05,
+      "loss": 0.6011,
+      "step": 3876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7546165683508595e-05,
+      "loss": 0.5558,
+      "step": 3877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7545524493459867e-05,
+      "loss": 0.6238,
+      "step": 3878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754488330341113e-05,
+      "loss": 0.5331,
+      "step": 3879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75442421133624e-05,
+      "loss": 0.665,
+      "step": 3880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7543600923313675e-05,
+      "loss": 0.588,
+      "step": 3881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754295973326494e-05,
+      "loss": 0.5468,
+      "step": 3882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754231854321621e-05,
+      "loss": 0.602,
+      "step": 3883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7541677353167477e-05,
+      "loss": 0.6038,
+      "step": 3884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7541036163118755e-05,
+      "loss": 0.5117,
+      "step": 3885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754039497307002e-05,
+      "loss": 0.6303,
+      "step": 3886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753975378302129e-05,
+      "loss": 0.6252,
+      "step": 3887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753911259297256e-05,
+      "loss": 0.5344,
+      "step": 3888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753847140292383e-05,
+      "loss": 0.6011,
+      "step": 3889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75378302128751e-05,
+      "loss": 0.5378,
+      "step": 3890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7537189022826365e-05,
+      "loss": 0.6966,
+      "step": 3891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753654783277764e-05,
+      "loss": 0.6463,
+      "step": 3892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753590664272891e-05,
+      "loss": 0.6041,
+      "step": 3893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753526545268018e-05,
+      "loss": 0.5829,
+      "step": 3894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7534624262631445e-05,
+      "loss": 0.59,
+      "step": 3895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753398307258272e-05,
+      "loss": 0.5822,
+      "step": 3896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753334188253398e-05,
+      "loss": 0.5913,
+      "step": 3897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7532700692485253e-05,
+      "loss": 0.6078,
+      "step": 3898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7532059502436525e-05,
+      "loss": 0.6246,
+      "step": 3899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753141831238779e-05,
+      "loss": 0.6293,
+      "step": 3900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753077712233906e-05,
+      "loss": 0.6673,
+      "step": 3901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7530135932290334e-05,
+      "loss": 0.711,
+      "step": 3902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7529494742241605e-05,
+      "loss": 0.5969,
+      "step": 3903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752885355219287e-05,
+      "loss": 0.5786,
+      "step": 3904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752821236214414e-05,
+      "loss": 0.5535,
+      "step": 3905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752757117209541e-05,
+      "loss": 0.5901,
+      "step": 3906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752692998204668e-05,
+      "loss": 0.6202,
+      "step": 3907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752628879199795e-05,
+      "loss": 0.6203,
+      "step": 3908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752564760194922e-05,
+      "loss": 0.5778,
+      "step": 3909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7525006411900494e-05,
+      "loss": 0.6614,
+      "step": 3910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752436522185176e-05,
+      "loss": 0.6459,
+      "step": 3911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752372403180303e-05,
+      "loss": 0.5804,
+      "step": 3912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7523082841754295e-05,
+      "loss": 0.5353,
+      "step": 3913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752244165170557e-05,
+      "loss": 0.465,
+      "step": 3914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752180046165683e-05,
+      "loss": 0.6009,
+      "step": 3915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7521159271608104e-05,
+      "loss": 0.6134,
+      "step": 3916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7520518081559375e-05,
+      "loss": 0.543,
+      "step": 3917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751987689151065e-05,
+      "loss": 0.7149,
+      "step": 3918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751923570146192e-05,
+      "loss": 0.5725,
+      "step": 3919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7518594511413184e-05,
+      "loss": 0.545,
+      "step": 3920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7517953321364456e-05,
+      "loss": 0.5982,
+      "step": 3921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751731213131572e-05,
+      "loss": 0.638,
+      "step": 3922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751667094126699e-05,
+      "loss": 0.5092,
+      "step": 3923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7516029751218264e-05,
+      "loss": 0.6218,
+      "step": 3924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7515388561169536e-05,
+      "loss": 0.5464,
+      "step": 3925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751474737112081e-05,
+      "loss": 0.7073,
+      "step": 3926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751410618107207e-05,
+      "loss": 0.5739,
+      "step": 3927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7513464991023344e-05,
+      "loss": 0.6355,
+      "step": 3928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751282380097461e-05,
+      "loss": 0.5994,
+      "step": 3929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751218261092588e-05,
+      "loss": 0.6739,
+      "step": 3930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7511541420877146e-05,
+      "loss": 0.6128,
+      "step": 3931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751090023082842e-05,
+      "loss": 0.6411,
+      "step": 3932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751025904077969e-05,
+      "loss": 0.5385,
+      "step": 3933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750961785073096e-05,
+      "loss": 0.5722,
+      "step": 3934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750897666068223e-05,
+      "loss": 0.6108,
+      "step": 3935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75083354706335e-05,
+      "loss": 0.6275,
+      "step": 3936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750769428058477e-05,
+      "loss": 0.4888,
+      "step": 3937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7507053090536034e-05,
+      "loss": 0.6383,
+      "step": 3938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7506411900487306e-05,
+      "loss": 0.5981,
+      "step": 3939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750577071043857e-05,
+      "loss": 0.6075,
+      "step": 3940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750512952038985e-05,
+      "loss": 0.5484,
+      "step": 3941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750448833034112e-05,
+      "loss": 0.5396,
+      "step": 3942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7503847140292386e-05,
+      "loss": 0.5477,
+      "step": 3943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750320595024366e-05,
+      "loss": 0.5923,
+      "step": 3944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750256476019492e-05,
+      "loss": 0.5817,
+      "step": 3945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7501923570146194e-05,
+      "loss": 0.5456,
+      "step": 3946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750128238009746e-05,
+      "loss": 0.5436,
+      "step": 3947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750064119004873e-05,
+      "loss": 0.5864,
+      "step": 3948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75e-05,
+      "loss": 0.5927,
+      "step": 3949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7499358809951274e-05,
+      "loss": 0.6182,
+      "step": 3950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7498717619902546e-05,
+      "loss": 0.616,
+      "step": 3951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749807642985381e-05,
+      "loss": 0.6664,
+      "step": 3952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749743523980508e-05,
+      "loss": 0.6514,
+      "step": 3953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749679404975635e-05,
+      "loss": 0.4279,
+      "step": 3954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749615285970762e-05,
+      "loss": 0.6325,
+      "step": 3955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7495511669658884e-05,
+      "loss": 0.657,
+      "step": 3956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749487047961016e-05,
+      "loss": 0.5784,
+      "step": 3957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749422928956143e-05,
+      "loss": 0.5475,
+      "step": 3958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74935880995127e-05,
+      "loss": 0.5481,
+      "step": 3959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749294690946397e-05,
+      "loss": 0.6793,
+      "step": 3960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7492305719415236e-05,
+      "loss": 0.6593,
+      "step": 3961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749166452936651e-05,
+      "loss": 0.5818,
+      "step": 3962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749102333931777e-05,
+      "loss": 0.5867,
+      "step": 3963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7490382149269044e-05,
+      "loss": 0.5461,
+      "step": 3964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7489740959220316e-05,
+      "loss": 0.6123,
+      "step": 3965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748909976917159e-05,
+      "loss": 0.6489,
+      "step": 3966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748845857912285e-05,
+      "loss": 0.6897,
+      "step": 3967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7487817389074125e-05,
+      "loss": 0.6131,
+      "step": 3968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7487176199025396e-05,
+      "loss": 0.5955,
+      "step": 3969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748653500897666e-05,
+      "loss": 0.6151,
+      "step": 3970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748589381892793e-05,
+      "loss": 0.5871,
+      "step": 3971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74852526288792e-05,
+      "loss": 0.5798,
+      "step": 3972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7484611438830476e-05,
+      "loss": 0.5538,
+      "step": 3973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748397024878174e-05,
+      "loss": 0.6438,
+      "step": 3974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748332905873301e-05,
+      "loss": 0.6546,
+      "step": 3975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748268786868428e-05,
+      "loss": 0.6099,
+      "step": 3976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748204667863555e-05,
+      "loss": 0.5803,
+      "step": 3977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748140548858682e-05,
+      "loss": 0.4538,
+      "step": 3978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7480764298538086e-05,
+      "loss": 0.6345,
+      "step": 3979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748012310848936e-05,
+      "loss": 0.6238,
+      "step": 3980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747948191844063e-05,
+      "loss": 0.5935,
+      "step": 3981
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74788407283919e-05,
+      "loss": 0.6254,
+      "step": 3982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7478199538343166e-05,
+      "loss": 0.6507,
+      "step": 3983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747755834829444e-05,
+      "loss": 0.5997,
+      "step": 3984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74769171582457e-05,
+      "loss": 0.606,
+      "step": 3985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7476275968196975e-05,
+      "loss": 0.5106,
+      "step": 3986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7475634778148247e-05,
+      "loss": 0.6172,
+      "step": 3987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747499358809951e-05,
+      "loss": 0.6329,
+      "step": 3988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747435239805078e-05,
+      "loss": 0.5764,
+      "step": 3989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7473711208002055e-05,
+      "loss": 0.6322,
+      "step": 3990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747307001795333e-05,
+      "loss": 0.5565,
+      "step": 3991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747242882790459e-05,
+      "loss": 0.5419,
+      "step": 3992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747178763785586e-05,
+      "loss": 0.6003,
+      "step": 3993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747114644780713e-05,
+      "loss": 0.6074,
+      "step": 3994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74705052577584e-05,
+      "loss": 0.5389,
+      "step": 3995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746986406770967e-05,
+      "loss": 0.6593,
+      "step": 3996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746922287766094e-05,
+      "loss": 0.5913,
+      "step": 3997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7468581687612215e-05,
+      "loss": 0.5442,
+      "step": 3998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746794049756348e-05,
+      "loss": 0.6488,
+      "step": 3999
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746729930751475e-05,
+      "loss": 0.5802,
+      "step": 4000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746665811746602e-05,
+      "loss": 0.5482,
+      "step": 4001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746601692741729e-05,
+      "loss": 0.6009,
+      "step": 4002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746537573736855e-05,
+      "loss": 0.5869,
+      "step": 4003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7464734547319825e-05,
+      "loss": 0.5968,
+      "step": 4004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74640933572711e-05,
+      "loss": 0.627,
+      "step": 4005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746345216722237e-05,
+      "loss": 0.5893,
+      "step": 4006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746281097717364e-05,
+      "loss": 0.6585,
+      "step": 4007
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7462169787124905e-05,
+      "loss": 0.7039,
+      "step": 4008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746152859707618e-05,
+      "loss": 0.6968,
+      "step": 4009
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746088740702744e-05,
+      "loss": 0.6535,
+      "step": 4010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7460246216978714e-05,
+      "loss": 0.515,
+      "step": 4011
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7459605026929985e-05,
+      "loss": 0.5498,
+      "step": 4012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745896383688126e-05,
+      "loss": 0.6346,
+      "step": 4013
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745832264683253e-05,
+      "loss": 0.5708,
+      "step": 4014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7457681456783794e-05,
+      "loss": 0.5836,
+      "step": 4015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7457040266735065e-05,
+      "loss": 0.5924,
+      "step": 4016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745639907668633e-05,
+      "loss": 0.613,
+      "step": 4017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74557578866376e-05,
+      "loss": 0.6645,
+      "step": 4018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745511669658887e-05,
+      "loss": 0.6082,
+      "step": 4019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745447550654014e-05,
+      "loss": 0.6651,
+      "step": 4020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745383431649141e-05,
+      "loss": 0.6497,
+      "step": 4021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745319312644268e-05,
+      "loss": 0.6069,
+      "step": 4022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7452551936393954e-05,
+      "loss": 0.6437,
+      "step": 4023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745191074634522e-05,
+      "loss": 0.6566,
+      "step": 4024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.745126955629649e-05,
+      "loss": 0.6697,
+      "step": 4025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7450628366247755e-05,
+      "loss": 0.594,
+      "step": 4026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744998717619903e-05,
+      "loss": 0.6124,
+      "step": 4027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744934598615029e-05,
+      "loss": 0.6516,
+      "step": 4028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744870479610157e-05,
+      "loss": 0.5962,
+      "step": 4029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7448063606052835e-05,
+      "loss": 0.6069,
+      "step": 4030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744742241600411e-05,
+      "loss": 0.5814,
+      "step": 4031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744678122595538e-05,
+      "loss": 0.5511,
+      "step": 4032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7446140035906644e-05,
+      "loss": 0.6392,
+      "step": 4033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7445498845857916e-05,
+      "loss": 0.6821,
+      "step": 4034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744485765580918e-05,
+      "loss": 0.6107,
+      "step": 4035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744421646576045e-05,
+      "loss": 0.5188,
+      "step": 4036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7443575275711724e-05,
+      "loss": 0.5203,
+      "step": 4037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7442934085662996e-05,
+      "loss": 0.5857,
+      "step": 4038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744229289561426e-05,
+      "loss": 0.586,
+      "step": 4039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744165170556553e-05,
+      "loss": 0.5953,
+      "step": 4040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7441010515516804e-05,
+      "loss": 0.4162,
+      "step": 4041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744036932546807e-05,
+      "loss": 0.6111,
+      "step": 4042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743972813541934e-05,
+      "loss": 0.5897,
+      "step": 4043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7439086945370606e-05,
+      "loss": 0.5772,
+      "step": 4044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7438445755321884e-05,
+      "loss": 0.6951,
+      "step": 4045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743780456527315e-05,
+      "loss": 0.5878,
+      "step": 4046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743716337522442e-05,
+      "loss": 0.6627,
+      "step": 4047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7436522185175686e-05,
+      "loss": 0.6808,
+      "step": 4048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743588099512696e-05,
+      "loss": 0.5384,
+      "step": 4049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743523980507823e-05,
+      "loss": 0.7647,
+      "step": 4050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7434598615029494e-05,
+      "loss": 0.6147,
+      "step": 4051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7433957424980766e-05,
+      "loss": 0.6266,
+      "step": 4052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743331623493204e-05,
+      "loss": 0.6134,
+      "step": 4053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743267504488331e-05,
+      "loss": 0.5298,
+      "step": 4054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7432033854834574e-05,
+      "loss": 0.6126,
+      "step": 4055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7431392664785846e-05,
+      "loss": 0.6648,
+      "step": 4056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743075147473711e-05,
+      "loss": 0.6442,
+      "step": 4057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743011028468838e-05,
+      "loss": 0.6127,
+      "step": 4058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7429469094639654e-05,
+      "loss": 0.5449,
+      "step": 4059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742882790459092e-05,
+      "loss": 0.5629,
+      "step": 4060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742818671454219e-05,
+      "loss": 0.6286,
+      "step": 4061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742754552449346e-05,
+      "loss": 0.5815,
+      "step": 4062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7426904334444734e-05,
+      "loss": 0.5803,
+      "step": 4063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7426263144396e-05,
+      "loss": 0.5774,
+      "step": 4064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742562195434727e-05,
+      "loss": 0.5923,
+      "step": 4065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7424980764298536e-05,
+      "loss": 0.5347,
+      "step": 4066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742433957424981e-05,
+      "loss": 0.6638,
+      "step": 4067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742369838420108e-05,
+      "loss": 0.5596,
+      "step": 4068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742305719415235e-05,
+      "loss": 0.6044,
+      "step": 4069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742241600410362e-05,
+      "loss": 0.5785,
+      "step": 4070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742177481405489e-05,
+      "loss": 0.57,
+      "step": 4071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742113362400616e-05,
+      "loss": 0.5985,
+      "step": 4072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7420492433957424e-05,
+      "loss": 0.5622,
+      "step": 4073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7419851243908696e-05,
+      "loss": 0.6076,
+      "step": 4074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741921005385996e-05,
+      "loss": 0.6279,
+      "step": 4075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741856886381123e-05,
+      "loss": 0.7098,
+      "step": 4076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7417927673762505e-05,
+      "loss": 0.6854,
+      "step": 4077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7417286483713776e-05,
+      "loss": 0.5057,
+      "step": 4078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741664529366505e-05,
+      "loss": 0.6143,
+      "step": 4079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741600410361631e-05,
+      "loss": 0.5703,
+      "step": 4080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7415362913567585e-05,
+      "loss": 0.6024,
+      "step": 4081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741472172351885e-05,
+      "loss": 0.6894,
+      "step": 4082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741408053347012e-05,
+      "loss": 0.5623,
+      "step": 4083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741343934342139e-05,
+      "loss": 0.6266,
+      "step": 4084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7412798153372665e-05,
+      "loss": 0.7227,
+      "step": 4085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7412156963323936e-05,
+      "loss": 0.6827,
+      "step": 4086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.74115157732752e-05,
+      "loss": 0.6023,
+      "step": 4087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741087458322647e-05,
+      "loss": 0.5969,
+      "step": 4088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741023339317774e-05,
+      "loss": 0.6502,
+      "step": 4089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740959220312901e-05,
+      "loss": 0.6992,
+      "step": 4090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7408951013080275e-05,
+      "loss": 0.5491,
+      "step": 4091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7408309823031546e-05,
+      "loss": 0.5786,
+      "step": 4092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740766863298282e-05,
+      "loss": 0.5578,
+      "step": 4093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740702744293409e-05,
+      "loss": 0.6036,
+      "step": 4094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740638625288536e-05,
+      "loss": 0.56,
+      "step": 4095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7405745062836626e-05,
+      "loss": 0.5823,
+      "step": 4096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.74051038727879e-05,
+      "loss": 0.5731,
+      "step": 4097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740446268273916e-05,
+      "loss": 0.55,
+      "step": 4098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7403821492690435e-05,
+      "loss": 0.6844,
+      "step": 4099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7403180302641707e-05,
+      "loss": 0.7929,
+      "step": 4100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740253911259298e-05,
+      "loss": 0.5819,
+      "step": 4101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740189792254424e-05,
+      "loss": 0.6348,
+      "step": 4102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7401256732495515e-05,
+      "loss": 0.5941,
+      "step": 4103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740061554244679e-05,
+      "loss": 0.6127,
+      "step": 4104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739997435239805e-05,
+      "loss": 0.6879,
+      "step": 4105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739933316234932e-05,
+      "loss": 0.6338,
+      "step": 4106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739869197230059e-05,
+      "loss": 0.5703,
+      "step": 4107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739805078225186e-05,
+      "loss": 0.6163,
+      "step": 4108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739740959220313e-05,
+      "loss": 0.7145,
+      "step": 4109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7396768402154403e-05,
+      "loss": 0.5573,
+      "step": 4110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739612721210567e-05,
+      "loss": 0.5751,
+      "step": 4111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739548602205694e-05,
+      "loss": 0.5254,
+      "step": 4112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739484483200821e-05,
+      "loss": 0.5169,
+      "step": 4113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739420364195948e-05,
+      "loss": 0.5678,
+      "step": 4114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739356245191075e-05,
+      "loss": 0.5401,
+      "step": 4115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739292126186201e-05,
+      "loss": 0.5987,
+      "step": 4116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739228007181329e-05,
+      "loss": 0.636,
+      "step": 4117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739163888176456e-05,
+      "loss": 0.7071,
+      "step": 4118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739099769171583e-05,
+      "loss": 0.5031,
+      "step": 4119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73903565016671e-05,
+      "loss": 0.6484,
+      "step": 4120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7389715311618365e-05,
+      "loss": 0.659,
+      "step": 4121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738907412156964e-05,
+      "loss": 0.5576,
+      "step": 4122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73884329315209e-05,
+      "loss": 0.6205,
+      "step": 4123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7387791741472174e-05,
+      "loss": 0.5314,
+      "step": 4124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7387150551423445e-05,
+      "loss": 0.5895,
+      "step": 4125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738650936137472e-05,
+      "loss": 0.5128,
+      "step": 4126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738586817132598e-05,
+      "loss": 0.6646,
+      "step": 4127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7385226981277254e-05,
+      "loss": 0.5162,
+      "step": 4128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7384585791228525e-05,
+      "loss": 0.593,
+      "step": 4129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738394460117979e-05,
+      "loss": 0.6501,
+      "step": 4130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738330341113106e-05,
+      "loss": 0.544,
+      "step": 4131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738266222108233e-05,
+      "loss": 0.6178,
+      "step": 4132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7382021031033605e-05,
+      "loss": 0.685,
+      "step": 4133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738137984098487e-05,
+      "loss": 0.6316,
+      "step": 4134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738073865093614e-05,
+      "loss": 0.5262,
+      "step": 4135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738009746088741e-05,
+      "loss": 0.5688,
+      "step": 4136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737945627083868e-05,
+      "loss": 0.6067,
+      "step": 4137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737881508078995e-05,
+      "loss": 0.5979,
+      "step": 4138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7378173890741215e-05,
+      "loss": 0.6443,
+      "step": 4139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737753270069249e-05,
+      "loss": 0.6695,
+      "step": 4140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737689151064376e-05,
+      "loss": 0.6689,
+      "step": 4141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737625032059503e-05,
+      "loss": 0.6188,
+      "step": 4142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7375609130546296e-05,
+      "loss": 0.6006,
+      "step": 4143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737496794049757e-05,
+      "loss": 0.6139,
+      "step": 4144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737432675044883e-05,
+      "loss": 0.5224,
+      "step": 4145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7373685560400104e-05,
+      "loss": 0.5723,
+      "step": 4146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7373044370351376e-05,
+      "loss": 0.6039,
+      "step": 4147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737240318030264e-05,
+      "loss": 0.6052,
+      "step": 4148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737176199025391e-05,
+      "loss": 0.5522,
+      "step": 4149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7371120800205184e-05,
+      "loss": 0.5577,
+      "step": 4150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7370479610156456e-05,
+      "loss": 0.5577,
+      "step": 4151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736983842010772e-05,
+      "loss": 0.6519,
+      "step": 4152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736919723005899e-05,
+      "loss": 0.5373,
+      "step": 4153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736855604001026e-05,
+      "loss": 0.5967,
+      "step": 4154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736791484996153e-05,
+      "loss": 0.6394,
+      "step": 4155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73672736599128e-05,
+      "loss": 0.5378,
+      "step": 4156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736663246986407e-05,
+      "loss": 0.641,
+      "step": 4157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7365991279815344e-05,
+      "loss": 0.601,
+      "step": 4158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736535008976661e-05,
+      "loss": 0.5341,
+      "step": 4159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736470889971788e-05,
+      "loss": 0.5343,
+      "step": 4160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7364067709669146e-05,
+      "loss": 0.5517,
+      "step": 4161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736342651962042e-05,
+      "loss": 0.6189,
+      "step": 4162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736278532957168e-05,
+      "loss": 0.587,
+      "step": 4163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7362144139522954e-05,
+      "loss": 0.519,
+      "step": 4164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7361502949474226e-05,
+      "loss": 0.5811,
+      "step": 4165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73608617594255e-05,
+      "loss": 0.7612,
+      "step": 4166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736022056937677e-05,
+      "loss": 0.5558,
+      "step": 4167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7359579379328034e-05,
+      "loss": 0.6152,
+      "step": 4168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7358938189279306e-05,
+      "loss": 0.5495,
+      "step": 4169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735829699923057e-05,
+      "loss": 0.5819,
+      "step": 4170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735765580918184e-05,
+      "loss": 0.6288,
+      "step": 4171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7357014619133114e-05,
+      "loss": 0.64,
+      "step": 4172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7356373429084386e-05,
+      "loss": 0.5339,
+      "step": 4173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735573223903566e-05,
+      "loss": 0.5011,
+      "step": 4174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735509104898692e-05,
+      "loss": 0.5719,
+      "step": 4175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7354449858938194e-05,
+      "loss": 0.5658,
+      "step": 4176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735380866888946e-05,
+      "loss": 0.6811,
+      "step": 4177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735316747884073e-05,
+      "loss": 0.5507,
+      "step": 4178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7352526288791996e-05,
+      "loss": 0.5182,
+      "step": 4179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735188509874327e-05,
+      "loss": 0.5754,
+      "step": 4180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735124390869454e-05,
+      "loss": 0.593,
+      "step": 4181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.735060271864581e-05,
+      "loss": 0.6133,
+      "step": 4182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734996152859708e-05,
+      "loss": 0.6719,
+      "step": 4183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734932033854835e-05,
+      "loss": 0.5422,
+      "step": 4184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734867914849962e-05,
+      "loss": 0.586,
+      "step": 4185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7348037958450884e-05,
+      "loss": 0.6016,
+      "step": 4186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7347396768402156e-05,
+      "loss": 0.5867,
+      "step": 4187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734675557835342e-05,
+      "loss": 0.5893,
+      "step": 4188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73461143883047e-05,
+      "loss": 0.7295,
+      "step": 4189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7345473198255965e-05,
+      "loss": 0.6951,
+      "step": 4190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7344832008207236e-05,
+      "loss": 0.4994,
+      "step": 4191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734419081815851e-05,
+      "loss": 0.5626,
+      "step": 4192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734354962810977e-05,
+      "loss": 0.5926,
+      "step": 4193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7342908438061045e-05,
+      "loss": 0.6074,
+      "step": 4194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734226724801231e-05,
+      "loss": 0.5161,
+      "step": 4195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734162605796358e-05,
+      "loss": 0.6213,
+      "step": 4196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.734098486791485e-05,
+      "loss": 0.5689,
+      "step": 4197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7340343677866125e-05,
+      "loss": 0.6532,
+      "step": 4198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733970248781739e-05,
+      "loss": 0.6541,
+      "step": 4199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733906129776866e-05,
+      "loss": 0.6666,
+      "step": 4200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733842010771993e-05,
+      "loss": 0.5653,
+      "step": 4201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73377789176712e-05,
+      "loss": 0.6187,
+      "step": 4202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733713772762247e-05,
+      "loss": 0.5413,
+      "step": 4203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7336496537573735e-05,
+      "loss": 0.5973,
+      "step": 4204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733585534752501e-05,
+      "loss": 0.5687,
+      "step": 4205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733521415747628e-05,
+      "loss": 0.5754,
+      "step": 4206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733457296742755e-05,
+      "loss": 0.5346,
+      "step": 4207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7333931777378815e-05,
+      "loss": 0.6761,
+      "step": 4208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7333290587330087e-05,
+      "loss": 0.5468,
+      "step": 4209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733264939728136e-05,
+      "loss": 0.5869,
+      "step": 4210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733200820723262e-05,
+      "loss": 0.5588,
+      "step": 4211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7331367017183895e-05,
+      "loss": 0.7361,
+      "step": 4212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7330725827135167e-05,
+      "loss": 0.582,
+      "step": 4213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733008463708644e-05,
+      "loss": 0.6724,
+      "step": 4214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73294434470377e-05,
+      "loss": 0.5832,
+      "step": 4215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7328802256988975e-05,
+      "loss": 0.6042,
+      "step": 4216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732816106694024e-05,
+      "loss": 0.5739,
+      "step": 4217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732751987689151e-05,
+      "loss": 0.581,
+      "step": 4218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732687868684278e-05,
+      "loss": 0.6092,
+      "step": 4219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732623749679405e-05,
+      "loss": 0.6667,
+      "step": 4220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732559630674533e-05,
+      "loss": 0.6068,
+      "step": 4221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732495511669659e-05,
+      "loss": 0.5781,
+      "step": 4222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7324313926647863e-05,
+      "loss": 0.5475,
+      "step": 4223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732367273659913e-05,
+      "loss": 0.5369,
+      "step": 4224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73230315465504e-05,
+      "loss": 0.5848,
+      "step": 4225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7322390356501665e-05,
+      "loss": 0.5299,
+      "step": 4226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732174916645294e-05,
+      "loss": 0.5572,
+      "step": 4227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732110797640421e-05,
+      "loss": 0.5812,
+      "step": 4228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732046678635548e-05,
+      "loss": 0.5972,
+      "step": 4229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731982559630675e-05,
+      "loss": 0.5851,
+      "step": 4230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731918440625802e-05,
+      "loss": 0.6487,
+      "step": 4231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731854321620929e-05,
+      "loss": 0.5647,
+      "step": 4232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7317902026160553e-05,
+      "loss": 0.6963,
+      "step": 4233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7317260836111825e-05,
+      "loss": 0.5146,
+      "step": 4234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731661964606309e-05,
+      "loss": 0.5376,
+      "step": 4235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731597845601436e-05,
+      "loss": 0.599,
+      "step": 4236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7315337265965634e-05,
+      "loss": 0.5922,
+      "step": 4237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7314696075916905e-05,
+      "loss": 0.5554,
+      "step": 4238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731405488586818e-05,
+      "loss": 0.5823,
+      "step": 4239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731341369581944e-05,
+      "loss": 0.6816,
+      "step": 4240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7312772505770714e-05,
+      "loss": 0.6758,
+      "step": 4241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731213131572198e-05,
+      "loss": 0.621,
+      "step": 4242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731149012567325e-05,
+      "loss": 0.5426,
+      "step": 4243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.731084893562452e-05,
+      "loss": 0.6371,
+      "step": 4244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7310207745575794e-05,
+      "loss": 0.6473,
+      "step": 4245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7309566555527065e-05,
+      "loss": 0.5433,
+      "step": 4246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.730892536547833e-05,
+      "loss": 0.4775,
+      "step": 4247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73082841754296e-05,
+      "loss": 0.6058,
+      "step": 4248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.730764298538087e-05,
+      "loss": 0.6015,
+      "step": 4249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.730700179533214e-05,
+      "loss": 0.5933,
+      "step": 4250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7306360605283404e-05,
+      "loss": 0.6277,
+      "step": 4251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7305719415234675e-05,
+      "loss": 0.682,
+      "step": 4252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.730507822518595e-05,
+      "loss": 0.5526,
+      "step": 4253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.730443703513722e-05,
+      "loss": 0.5945,
+      "step": 4254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.730379584508849e-05,
+      "loss": 0.5704,
+      "step": 4255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7303154655039756e-05,
+      "loss": 0.6004,
+      "step": 4256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.730251346499103e-05,
+      "loss": 0.5936,
+      "step": 4257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.730187227494229e-05,
+      "loss": 0.4962,
+      "step": 4258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7301231084893564e-05,
+      "loss": 0.5997,
+      "step": 4259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7300589894844836e-05,
+      "loss": 0.511,
+      "step": 4260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729994870479611e-05,
+      "loss": 0.6747,
+      "step": 4261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729930751474737e-05,
+      "loss": 0.5489,
+      "step": 4262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7298666324698644e-05,
+      "loss": 0.5643,
+      "step": 4263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7298025134649916e-05,
+      "loss": 0.5665,
+      "step": 4264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729738394460118e-05,
+      "loss": 0.5816,
+      "step": 4265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729674275455245e-05,
+      "loss": 0.6546,
+      "step": 4266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729610156450372e-05,
+      "loss": 0.7327,
+      "step": 4267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729546037445499e-05,
+      "loss": 0.6058,
+      "step": 4268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729481918440626e-05,
+      "loss": 0.6188,
+      "step": 4269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729417799435753e-05,
+      "loss": 0.5511,
+      "step": 4270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.72935368043088e-05,
+      "loss": 0.6726,
+      "step": 4271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729289561426007e-05,
+      "loss": 0.6898,
+      "step": 4272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729225442421134e-05,
+      "loss": 0.609,
+      "step": 4273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7291613234162606e-05,
+      "loss": 0.6202,
+      "step": 4274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729097204411388e-05,
+      "loss": 0.5142,
+      "step": 4275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.729033085406514e-05,
+      "loss": 0.5956,
+      "step": 4276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728968966401642e-05,
+      "loss": 0.6762,
+      "step": 4277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7289048473967686e-05,
+      "loss": 0.5116,
+      "step": 4278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728840728391896e-05,
+      "loss": 0.55,
+      "step": 4279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728776609387022e-05,
+      "loss": 0.6122,
+      "step": 4280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7287124903821494e-05,
+      "loss": 0.6114,
+      "step": 4281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7286483713772766e-05,
+      "loss": 0.5619,
+      "step": 4282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728584252372403e-05,
+      "loss": 0.6289,
+      "step": 4283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.72852013336753e-05,
+      "loss": 0.6274,
+      "step": 4284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7284560143626574e-05,
+      "loss": 0.5807,
+      "step": 4285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7283918953577846e-05,
+      "loss": 0.6433,
+      "step": 4286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728327776352911e-05,
+      "loss": 0.5195,
+      "step": 4287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728263657348038e-05,
+      "loss": 0.5658,
+      "step": 4288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728199538343165e-05,
+      "loss": 0.5749,
+      "step": 4289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728135419338292e-05,
+      "loss": 0.5844,
+      "step": 4290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728071300333419e-05,
+      "loss": 0.5808,
+      "step": 4291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7280071813285456e-05,
+      "loss": 0.6023,
+      "step": 4292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7279430623236735e-05,
+      "loss": 0.6188,
+      "step": 4293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7278789433188e-05,
+      "loss": 0.6388,
+      "step": 4294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727814824313927e-05,
+      "loss": 0.6272,
+      "step": 4295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7277507053090536e-05,
+      "loss": 0.6737,
+      "step": 4296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727686586304181e-05,
+      "loss": 0.6318,
+      "step": 4297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727622467299307e-05,
+      "loss": 0.559,
+      "step": 4298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7275583482944344e-05,
+      "loss": 0.6313,
+      "step": 4299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7274942292895616e-05,
+      "loss": 0.6255,
+      "step": 4300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727430110284689e-05,
+      "loss": 0.6021,
+      "step": 4301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727365991279816e-05,
+      "loss": 0.5523,
+      "step": 4302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7273018722749425e-05,
+      "loss": 0.5353,
+      "step": 4303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7272377532700696e-05,
+      "loss": 0.5184,
+      "step": 4304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727173634265196e-05,
+      "loss": 0.6335,
+      "step": 4305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727109515260323e-05,
+      "loss": 0.5952,
+      "step": 4306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72704539625545e-05,
+      "loss": 0.5842,
+      "step": 4307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726981277250577e-05,
+      "loss": 0.5088,
+      "step": 4308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726917158245704e-05,
+      "loss": 0.6022,
+      "step": 4309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726853039240831e-05,
+      "loss": 0.5433,
+      "step": 4310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7267889202359585e-05,
+      "loss": 0.6969,
+      "step": 4311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726724801231085e-05,
+      "loss": 0.5898,
+      "step": 4312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726660682226212e-05,
+      "loss": 0.5819,
+      "step": 4313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7265965632213386e-05,
+      "loss": 0.555,
+      "step": 4314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726532444216466e-05,
+      "loss": 0.5732,
+      "step": 4315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726468325211593e-05,
+      "loss": 0.5966,
+      "step": 4316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72640420620672e-05,
+      "loss": 0.512,
+      "step": 4317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726340087201847e-05,
+      "loss": 0.6257,
+      "step": 4318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726275968196974e-05,
+      "loss": 0.6199,
+      "step": 4319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726211849192101e-05,
+      "loss": 0.6007,
+      "step": 4320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7261477301872275e-05,
+      "loss": 0.5328,
+      "step": 4321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7260836111823547e-05,
+      "loss": 0.5671,
+      "step": 4322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726019492177481e-05,
+      "loss": 0.5665,
+      "step": 4323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725955373172608e-05,
+      "loss": 0.6894,
+      "step": 4324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7258912541677355e-05,
+      "loss": 0.6036,
+      "step": 4325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725827135162863e-05,
+      "loss": 0.6526,
+      "step": 4326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72576301615799e-05,
+      "loss": 0.6847,
+      "step": 4327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725698897153116e-05,
+      "loss": 0.5221,
+      "step": 4328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7256347781482435e-05,
+      "loss": 0.5431,
+      "step": 4329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72557065914337e-05,
+      "loss": 0.5796,
+      "step": 4330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725506540138497e-05,
+      "loss": 0.6393,
+      "step": 4331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725442421133624e-05,
+      "loss": 0.6054,
+      "step": 4332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7253783021287515e-05,
+      "loss": 0.6018,
+      "step": 4333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725314183123879e-05,
+      "loss": 0.6504,
+      "step": 4334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725250064119005e-05,
+      "loss": 0.5705,
+      "step": 4335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7251859451141323e-05,
+      "loss": 0.6197,
+      "step": 4336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725121826109259e-05,
+      "loss": 0.5678,
+      "step": 4337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725057707104386e-05,
+      "loss": 0.6352,
+      "step": 4338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7249935880995125e-05,
+      "loss": 0.6888,
+      "step": 4339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72492946909464e-05,
+      "loss": 0.6692,
+      "step": 4340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724865350089767e-05,
+      "loss": 0.667,
+      "step": 4341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724801231084894e-05,
+      "loss": 0.5836,
+      "step": 4342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724737112080021e-05,
+      "loss": 0.6215,
+      "step": 4343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724672993075148e-05,
+      "loss": 0.5184,
+      "step": 4344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724608874070275e-05,
+      "loss": 0.5639,
+      "step": 4345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7245447550654014e-05,
+      "loss": 0.6161,
+      "step": 4346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7244806360605285e-05,
+      "loss": 0.6319,
+      "step": 4347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724416517055655e-05,
+      "loss": 0.584,
+      "step": 4348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724352398050783e-05,
+      "loss": 0.631,
+      "step": 4349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7242882790459094e-05,
+      "loss": 0.5528,
+      "step": 4350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7242241600410365e-05,
+      "loss": 0.5391,
+      "step": 4351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724160041036164e-05,
+      "loss": 0.553,
+      "step": 4352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72409592203129e-05,
+      "loss": 0.6193,
+      "step": 4353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7240318030264174e-05,
+      "loss": 0.6741,
+      "step": 4354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723967684021544e-05,
+      "loss": 0.5969,
+      "step": 4355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723903565016671e-05,
+      "loss": 0.5991,
+      "step": 4356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723839446011798e-05,
+      "loss": 0.5992,
+      "step": 4357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7237753270069254e-05,
+      "loss": 0.5096,
+      "step": 4358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723711208002052e-05,
+      "loss": 0.5208,
+      "step": 4359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723647088997179e-05,
+      "loss": 0.5423,
+      "step": 4360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723582969992306e-05,
+      "loss": 0.6968,
+      "step": 4361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723518850987433e-05,
+      "loss": 0.5837,
+      "step": 4362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72345473198256e-05,
+      "loss": 0.5838,
+      "step": 4363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7233906129776864e-05,
+      "loss": 0.6247,
+      "step": 4364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723326493972814e-05,
+      "loss": 0.603,
+      "step": 4365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723262374967941e-05,
+      "loss": 0.6424,
+      "step": 4366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723198255963068e-05,
+      "loss": 0.4916,
+      "step": 4367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7231341369581944e-05,
+      "loss": 0.6064,
+      "step": 4368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7230700179533216e-05,
+      "loss": 0.5808,
+      "step": 4369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723005898948449e-05,
+      "loss": 0.5807,
+      "step": 4370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722941779943575e-05,
+      "loss": 0.5025,
+      "step": 4371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7228776609387024e-05,
+      "loss": 0.6418,
+      "step": 4372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7228135419338296e-05,
+      "loss": 0.6518,
+      "step": 4373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722749422928957e-05,
+      "loss": 0.655,
+      "step": 4374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722685303924083e-05,
+      "loss": 0.5849,
+      "step": 4375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7226211849192104e-05,
+      "loss": 0.6447,
+      "step": 4376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722557065914337e-05,
+      "loss": 0.6257,
+      "step": 4377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722492946909464e-05,
+      "loss": 0.6365,
+      "step": 4378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722428827904591e-05,
+      "loss": 0.6047,
+      "step": 4379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722364708899718e-05,
+      "loss": 0.56,
+      "step": 4380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7223005898948456e-05,
+      "loss": 0.5992,
+      "step": 4381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722236470889972e-05,
+      "loss": 0.5559,
+      "step": 4382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722172351885099e-05,
+      "loss": 0.5763,
+      "step": 4383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722108232880226e-05,
+      "loss": 0.6112,
+      "step": 4384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722044113875353e-05,
+      "loss": 0.586,
+      "step": 4385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7219799948704794e-05,
+      "loss": 0.55,
+      "step": 4386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7219158758656066e-05,
+      "loss": 0.5546,
+      "step": 4387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721851756860734e-05,
+      "loss": 0.6137,
+      "step": 4388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721787637855861e-05,
+      "loss": 0.6472,
+      "step": 4389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721723518850988e-05,
+      "loss": 0.5558,
+      "step": 4390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7216593998461146e-05,
+      "loss": 0.574,
+      "step": 4391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721595280841242e-05,
+      "loss": 0.6059,
+      "step": 4392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721531161836368e-05,
+      "loss": 0.6378,
+      "step": 4393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7214670428314954e-05,
+      "loss": 0.5148,
+      "step": 4394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721402923826622e-05,
+      "loss": 0.6649,
+      "step": 4395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721338804821749e-05,
+      "loss": 0.5216,
+      "step": 4396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721274685816876e-05,
+      "loss": 0.676,
+      "step": 4397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7212105668120034e-05,
+      "loss": 0.5587,
+      "step": 4398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7211464478071306e-05,
+      "loss": 0.5982,
+      "step": 4399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721082328802257e-05,
+      "loss": 0.5004,
+      "step": 4400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721018209797384e-05,
+      "loss": 0.6243,
+      "step": 4401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720954090792511e-05,
+      "loss": 0.5149,
+      "step": 4402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720889971787638e-05,
+      "loss": 0.5144,
+      "step": 4403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720825852782765e-05,
+      "loss": 0.5816,
+      "step": 4404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720761733777892e-05,
+      "loss": 0.546,
+      "step": 4405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7206976147730195e-05,
+      "loss": 0.6192,
+      "step": 4406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720633495768146e-05,
+      "loss": 0.6243,
+      "step": 4407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720569376763273e-05,
+      "loss": 0.54,
+      "step": 4408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7205052577583996e-05,
+      "loss": 0.6031,
+      "step": 4409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720441138753527e-05,
+      "loss": 0.6214,
+      "step": 4410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720377019748653e-05,
+      "loss": 0.6286,
+      "step": 4411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7203129007437805e-05,
+      "loss": 0.516,
+      "step": 4412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7202487817389076e-05,
+      "loss": 0.6451,
+      "step": 4413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720184662734035e-05,
+      "loss": 0.486,
+      "step": 4414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720120543729162e-05,
+      "loss": 0.6477,
+      "step": 4415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7200564247242885e-05,
+      "loss": 0.6287,
+      "step": 4416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7199923057194156e-05,
+      "loss": 0.5775,
+      "step": 4417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719928186714542e-05,
+      "loss": 0.6089,
+      "step": 4418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719864067709669e-05,
+      "loss": 0.5793,
+      "step": 4419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7197999487047965e-05,
+      "loss": 0.5389,
+      "step": 4420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7197358296999236e-05,
+      "loss": 0.6258,
+      "step": 4421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71967171069505e-05,
+      "loss": 0.6146,
+      "step": 4422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719607591690177e-05,
+      "loss": 0.632,
+      "step": 4423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7195434726853045e-05,
+      "loss": 0.7046,
+      "step": 4424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719479353680431e-05,
+      "loss": 0.5813,
+      "step": 4425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719415234675558e-05,
+      "loss": 0.6155,
+      "step": 4426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7193511156706846e-05,
+      "loss": 0.675,
+      "step": 4427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719286996665812e-05,
+      "loss": 0.6442,
+      "step": 4428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719222877660939e-05,
+      "loss": 0.5733,
+      "step": 4429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719158758656066e-05,
+      "loss": 0.5476,
+      "step": 4430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7190946396511926e-05,
+      "loss": 0.6014,
+      "step": 4431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71903052064632e-05,
+      "loss": 0.6407,
+      "step": 4432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718966401641447e-05,
+      "loss": 0.636,
+      "step": 4433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7189022826365735e-05,
+      "loss": 0.5967,
+      "step": 4434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7188381636317007e-05,
+      "loss": 0.6432,
+      "step": 4435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718774044626827e-05,
+      "loss": 0.6058,
+      "step": 4436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718709925621955e-05,
+      "loss": 0.6089,
+      "step": 4437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7186458066170815e-05,
+      "loss": 0.5694,
+      "step": 4438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718581687612209e-05,
+      "loss": 0.652,
+      "step": 4439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718517568607335e-05,
+      "loss": 0.5875,
+      "step": 4440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718453449602462e-05,
+      "loss": 0.5649,
+      "step": 4441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7183893305975895e-05,
+      "loss": 0.5983,
+      "step": 4442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718325211592716e-05,
+      "loss": 0.5683,
+      "step": 4443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718261092587843e-05,
+      "loss": 0.626,
+      "step": 4444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7181969735829703e-05,
+      "loss": 0.6125,
+      "step": 4445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7181328545780975e-05,
+      "loss": 0.6418,
+      "step": 4446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718068735573224e-05,
+      "loss": 0.5969,
+      "step": 4447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718004616568351e-05,
+      "loss": 0.5605,
+      "step": 4448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.717940497563478e-05,
+      "loss": 0.5404,
+      "step": 4449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.717876378558605e-05,
+      "loss": 0.6189,
+      "step": 4450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.717812259553732e-05,
+      "loss": 0.6192,
+      "step": 4451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7177481405488585e-05,
+      "loss": 0.6922,
+      "step": 4452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7176840215439864e-05,
+      "loss": 0.6333,
+      "step": 4453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.717619902539113e-05,
+      "loss": 0.5805,
+      "step": 4454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71755578353424e-05,
+      "loss": 0.6553,
+      "step": 4455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7174916645293665e-05,
+      "loss": 0.5568,
+      "step": 4456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.717427545524494e-05,
+      "loss": 0.5854,
+      "step": 4457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71736342651962e-05,
+      "loss": 0.6353,
+      "step": 4458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7172993075147474e-05,
+      "loss": 0.5605,
+      "step": 4459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7172351885098745e-05,
+      "loss": 0.5943,
+      "step": 4460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.717171069505002e-05,
+      "loss": 0.5739,
+      "step": 4461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.717106950500129e-05,
+      "loss": 0.6443,
+      "step": 4462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7170428314952554e-05,
+      "loss": 0.6452,
+      "step": 4463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7169787124903825e-05,
+      "loss": 0.6716,
+      "step": 4464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716914593485509e-05,
+      "loss": 0.5987,
+      "step": 4465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716850474480636e-05,
+      "loss": 0.5885,
+      "step": 4466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716786355475763e-05,
+      "loss": 0.6257,
+      "step": 4467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71672223647089e-05,
+      "loss": 0.7254,
+      "step": 4468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716658117466017e-05,
+      "loss": 0.6247,
+      "step": 4469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716593998461144e-05,
+      "loss": 0.6141,
+      "step": 4470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7165298794562714e-05,
+      "loss": 0.6759,
+      "step": 4471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716465760451398e-05,
+      "loss": 0.6168,
+      "step": 4472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716401641446525e-05,
+      "loss": 0.6348,
+      "step": 4473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7163375224416515e-05,
+      "loss": 0.6005,
+      "step": 4474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716273403436779e-05,
+      "loss": 0.5263,
+      "step": 4475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716209284431906e-05,
+      "loss": 0.6194,
+      "step": 4476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716145165427033e-05,
+      "loss": 0.6562,
+      "step": 4477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71608104642216e-05,
+      "loss": 0.6216,
+      "step": 4478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716016927417287e-05,
+      "loss": 0.4861,
+      "step": 4479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715952808412414e-05,
+      "loss": 0.6238,
+      "step": 4480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7158886894075404e-05,
+      "loss": 0.5524,
+      "step": 4481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7158245704026676e-05,
+      "loss": 0.4684,
+      "step": 4482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715760451397794e-05,
+      "loss": 0.597,
+      "step": 4483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715696332392921e-05,
+      "loss": 0.5535,
+      "step": 4484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7156322133880484e-05,
+      "loss": 0.5654,
+      "step": 4485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7155680943831756e-05,
+      "loss": 0.6848,
+      "step": 4486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715503975378303e-05,
+      "loss": 0.5146,
+      "step": 4487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715439856373429e-05,
+      "loss": 0.5172,
+      "step": 4488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7153757373685564e-05,
+      "loss": 0.6836,
+      "step": 4489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715311618363683e-05,
+      "loss": 0.5864,
+      "step": 4490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71524749935881e-05,
+      "loss": 0.6238,
+      "step": 4491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715183380353937e-05,
+      "loss": 0.6605,
+      "step": 4492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7151192613490644e-05,
+      "loss": 0.525,
+      "step": 4493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715055142344191e-05,
+      "loss": 0.6957,
+      "step": 4494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714991023339318e-05,
+      "loss": 0.5468,
+      "step": 4495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714926904334445e-05,
+      "loss": 0.5329,
+      "step": 4496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714862785329572e-05,
+      "loss": 0.6059,
+      "step": 4497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714798666324699e-05,
+      "loss": 0.5616,
+      "step": 4498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7147345473198254e-05,
+      "loss": 0.6077,
+      "step": 4499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7146704283149526e-05,
+      "loss": 0.6465,
+      "step": 4500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71460630931008e-05,
+      "loss": 0.6803,
+      "step": 4501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714542190305207e-05,
+      "loss": 0.6392,
+      "step": 4502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714478071300334e-05,
+      "loss": 0.5412,
+      "step": 4503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7144139522954606e-05,
+      "loss": 0.5646,
+      "step": 4504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714349833290588e-05,
+      "loss": 0.7062,
+      "step": 4505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714285714285714e-05,
+      "loss": 0.6141,
+      "step": 4506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7142215952808414e-05,
+      "loss": 0.5637,
+      "step": 4507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7141574762759686e-05,
+      "loss": 0.5713,
+      "step": 4508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714093357271096e-05,
+      "loss": 0.6184,
+      "step": 4509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714029238266222e-05,
+      "loss": 0.5896,
+      "step": 4510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7139651192613494e-05,
+      "loss": 0.5266,
+      "step": 4511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7139010002564766e-05,
+      "loss": 0.579,
+      "step": 4512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713836881251603e-05,
+      "loss": 0.5981,
+      "step": 4513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71377276224673e-05,
+      "loss": 0.621,
+      "step": 4514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713708643241857e-05,
+      "loss": 0.5521,
+      "step": 4515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713644524236984e-05,
+      "loss": 0.5932,
+      "step": 4516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713580405232111e-05,
+      "loss": 0.5831,
+      "step": 4517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713516286227238e-05,
+      "loss": 0.5828,
+      "step": 4518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713452167222365e-05,
+      "loss": 0.5041,
+      "step": 4519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713388048217492e-05,
+      "loss": 0.5171,
+      "step": 4520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713323929212619e-05,
+      "loss": 0.6281,
+      "step": 4521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7132598102077456e-05,
+      "loss": 0.6741,
+      "step": 4522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713195691202873e-05,
+      "loss": 0.5554,
+      "step": 4523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713131572197999e-05,
+      "loss": 0.6356,
+      "step": 4524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713067453193127e-05,
+      "loss": 0.6057,
+      "step": 4525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7130033341882536e-05,
+      "loss": 0.649,
+      "step": 4526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.712939215183381e-05,
+      "loss": 0.5724,
+      "step": 4527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.712875096178507e-05,
+      "loss": 0.5407,
+      "step": 4528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7128109771736345e-05,
+      "loss": 0.5135,
+      "step": 4529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7127468581687616e-05,
+      "loss": 0.5733,
+      "step": 4530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.712682739163888e-05,
+      "loss": 0.5076,
+      "step": 4531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.712618620159015e-05,
+      "loss": 0.6183,
+      "step": 4532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7125545011541425e-05,
+      "loss": 0.6041,
+      "step": 4533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7124903821492696e-05,
+      "loss": 0.5179,
+      "step": 4534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.712426263144396e-05,
+      "loss": 0.6142,
+      "step": 4535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.712362144139523e-05,
+      "loss": 0.5965,
+      "step": 4536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.71229802513465e-05,
+      "loss": 0.6869,
+      "step": 4537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.712233906129777e-05,
+      "loss": 0.5551,
+      "step": 4538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.712169787124904e-05,
+      "loss": 0.6074,
+      "step": 4539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7121056681200306e-05,
+      "loss": 0.5822,
+      "step": 4540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7120415491151585e-05,
+      "loss": 0.5924,
+      "step": 4541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.711977430110285e-05,
+      "loss": 0.6158,
+      "step": 4542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.711913311105412e-05,
+      "loss": 0.5584,
+      "step": 4543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7118491921005387e-05,
+      "loss": 0.5599,
+      "step": 4544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.711785073095666e-05,
+      "loss": 0.5744,
+      "step": 4545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.711720954090792e-05,
+      "loss": 0.596,
+      "step": 4546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7116568350859195e-05,
+      "loss": 0.63,
+      "step": 4547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7115927160810467e-05,
+      "loss": 0.5674,
+      "step": 4548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.711528597076174e-05,
+      "loss": 0.6929,
+      "step": 4549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.711464478071301e-05,
+      "loss": 0.6787,
+      "step": 4550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7114003590664275e-05,
+      "loss": 0.6127,
+      "step": 4551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711336240061555e-05,
+      "loss": 0.5965,
+      "step": 4552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711272121056681e-05,
+      "loss": 0.5857,
+      "step": 4553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711208002051808e-05,
+      "loss": 0.5624,
+      "step": 4554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711143883046935e-05,
+      "loss": 0.5467,
+      "step": 4555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711079764042062e-05,
+      "loss": 0.5811,
+      "step": 4556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711015645037189e-05,
+      "loss": 0.6517,
+      "step": 4557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7109515260323163e-05,
+      "loss": 0.5305,
+      "step": 4558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7108874070274435e-05,
+      "loss": 0.6101,
+      "step": 4559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.71082328802257e-05,
+      "loss": 0.5271,
+      "step": 4560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710759169017697e-05,
+      "loss": 0.5738,
+      "step": 4561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710695050012824e-05,
+      "loss": 0.5694,
+      "step": 4562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710630931007951e-05,
+      "loss": 0.6748,
+      "step": 4563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710566812003078e-05,
+      "loss": 0.5259,
+      "step": 4564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710502692998205e-05,
+      "loss": 0.487,
+      "step": 4565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7104385739933324e-05,
+      "loss": 0.7207,
+      "step": 4566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710374454988459e-05,
+      "loss": 0.6225,
+      "step": 4567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710310335983586e-05,
+      "loss": 0.5753,
+      "step": 4568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7102462169787125e-05,
+      "loss": 0.6096,
+      "step": 4569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.71018209797384e-05,
+      "loss": 0.575,
+      "step": 4570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710117978968966e-05,
+      "loss": 0.612,
+      "step": 4571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7100538599640934e-05,
+      "loss": 0.5788,
+      "step": 4572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7099897409592205e-05,
+      "loss": 0.5608,
+      "step": 4573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709925621954348e-05,
+      "loss": 0.648,
+      "step": 4574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709861502949475e-05,
+      "loss": 0.5482,
+      "step": 4575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7097973839446014e-05,
+      "loss": 0.6526,
+      "step": 4576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7097332649397285e-05,
+      "loss": 0.5263,
+      "step": 4577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709669145934855e-05,
+      "loss": 0.6628,
+      "step": 4578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709605026929982e-05,
+      "loss": 0.6053,
+      "step": 4579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7095409079251094e-05,
+      "loss": 0.5606,
+      "step": 4580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7094767889202365e-05,
+      "loss": 0.6684,
+      "step": 4581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709412669915363e-05,
+      "loss": 0.6127,
+      "step": 4582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70934855091049e-05,
+      "loss": 0.5833,
+      "step": 4583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7092844319056174e-05,
+      "loss": 0.5706,
+      "step": 4584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709220312900744e-05,
+      "loss": 0.5664,
+      "step": 4585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709156193895871e-05,
+      "loss": 0.6148,
+      "step": 4586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7090920748909975e-05,
+      "loss": 0.5983,
+      "step": 4587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709027955886125e-05,
+      "loss": 0.4693,
+      "step": 4588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708963836881252e-05,
+      "loss": 0.5842,
+      "step": 4589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708899717876379e-05,
+      "loss": 0.6348,
+      "step": 4590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7088355988715056e-05,
+      "loss": 0.4883,
+      "step": 4591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708771479866633e-05,
+      "loss": 0.5917,
+      "step": 4592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70870736086176e-05,
+      "loss": 0.5566,
+      "step": 4593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7086432418568864e-05,
+      "loss": 0.6369,
+      "step": 4594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7085791228520136e-05,
+      "loss": 0.5627,
+      "step": 4595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70851500384714e-05,
+      "loss": 0.6069,
+      "step": 4596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708450884842268e-05,
+      "loss": 0.5539,
+      "step": 4597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7083867658373944e-05,
+      "loss": 0.5298,
+      "step": 4598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7083226468325216e-05,
+      "loss": 0.6629,
+      "step": 4599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708258527827648e-05,
+      "loss": 0.6146,
+      "step": 4600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708194408822775e-05,
+      "loss": 0.5739,
+      "step": 4601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7081302898179024e-05,
+      "loss": 0.6416,
+      "step": 4602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708066170813029e-05,
+      "loss": 0.5726,
+      "step": 4603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708002051808156e-05,
+      "loss": 0.5224,
+      "step": 4604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707937932803283e-05,
+      "loss": 0.5656,
+      "step": 4605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7078738137984104e-05,
+      "loss": 0.5647,
+      "step": 4606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707809694793537e-05,
+      "loss": 0.6905,
+      "step": 4607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707745575788664e-05,
+      "loss": 0.6759,
+      "step": 4608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7076814567837906e-05,
+      "loss": 0.5656,
+      "step": 4609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707617337778918e-05,
+      "loss": 0.5792,
+      "step": 4610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707553218774045e-05,
+      "loss": 0.6676,
+      "step": 4611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7074890997691714e-05,
+      "loss": 0.6103,
+      "step": 4612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707424980764299e-05,
+      "loss": 0.5438,
+      "step": 4613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707360861759426e-05,
+      "loss": 0.6458,
+      "step": 4614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707296742754553e-05,
+      "loss": 0.563,
+      "step": 4615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7072326237496794e-05,
+      "loss": 0.6546,
+      "step": 4616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7071685047448066e-05,
+      "loss": 0.5847,
+      "step": 4617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707104385739933e-05,
+      "loss": 0.6131,
+      "step": 4618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70704026673506e-05,
+      "loss": 0.6495,
+      "step": 4619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7069761477301874e-05,
+      "loss": 0.559,
+      "step": 4620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7069120287253146e-05,
+      "loss": 0.5554,
+      "step": 4621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706847909720442e-05,
+      "loss": 0.7412,
+      "step": 4622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706783790715568e-05,
+      "loss": 0.6236,
+      "step": 4623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7067196717106954e-05,
+      "loss": 0.5672,
+      "step": 4624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706655552705822e-05,
+      "loss": 0.569,
+      "step": 4625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706591433700949e-05,
+      "loss": 0.5707,
+      "step": 4626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7065273146960756e-05,
+      "loss": 0.5467,
+      "step": 4627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706463195691203e-05,
+      "loss": 0.6011,
+      "step": 4628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7063990766863306e-05,
+      "loss": 0.6088,
+      "step": 4629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706334957681457e-05,
+      "loss": 0.5818,
+      "step": 4630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706270838676584e-05,
+      "loss": 0.5538,
+      "step": 4631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706206719671711e-05,
+      "loss": 0.7274,
+      "step": 4632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706142600666838e-05,
+      "loss": 0.6135,
+      "step": 4633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7060784816619645e-05,
+      "loss": 0.6194,
+      "step": 4634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7060143626570916e-05,
+      "loss": 0.5702,
+      "step": 4635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705950243652219e-05,
+      "loss": 0.5176,
+      "step": 4636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705886124647346e-05,
+      "loss": 0.6437,
+      "step": 4637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705822005642473e-05,
+      "loss": 0.5749,
+      "step": 4638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7057578866375996e-05,
+      "loss": 0.6498,
+      "step": 4639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705693767632727e-05,
+      "loss": 0.5808,
+      "step": 4640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705629648627853e-05,
+      "loss": 0.5708,
+      "step": 4641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7055655296229805e-05,
+      "loss": 0.6483,
+      "step": 4642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705501410618107e-05,
+      "loss": 0.6187,
+      "step": 4643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705437291613234e-05,
+      "loss": 0.5926,
+      "step": 4644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705373172608361e-05,
+      "loss": 0.539,
+      "step": 4645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7053090536034885e-05,
+      "loss": 0.6166,
+      "step": 4646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7052449345986156e-05,
+      "loss": 0.5641,
+      "step": 4647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705180815593742e-05,
+      "loss": 0.5428,
+      "step": 4648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705116696588869e-05,
+      "loss": 0.4762,
+      "step": 4649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705052577583996e-05,
+      "loss": 0.5417,
+      "step": 4650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704988458579123e-05,
+      "loss": 0.5945,
+      "step": 4651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70492433957425e-05,
+      "loss": 0.5832,
+      "step": 4652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704860220569377e-05,
+      "loss": 0.6842,
+      "step": 4653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704796101564504e-05,
+      "loss": 0.5595,
+      "step": 4654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704731982559631e-05,
+      "loss": 0.6015,
+      "step": 4655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704667863554758e-05,
+      "loss": 0.6772,
+      "step": 4656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7046037445498847e-05,
+      "loss": 0.5907,
+      "step": 4657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704539625545012e-05,
+      "loss": 0.6288,
+      "step": 4658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704475506540138e-05,
+      "loss": 0.5985,
+      "step": 4659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7044113875352655e-05,
+      "loss": 0.4647,
+      "step": 4660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704347268530393e-05,
+      "loss": 0.4758,
+      "step": 4661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70428314952552e-05,
+      "loss": 0.5328,
+      "step": 4662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704219030520646e-05,
+      "loss": 0.5929,
+      "step": 4663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7041549115157735e-05,
+      "loss": 0.5704,
+      "step": 4664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704090792510901e-05,
+      "loss": 0.5964,
+      "step": 4665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704026673506027e-05,
+      "loss": 0.5821,
+      "step": 4666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703962554501154e-05,
+      "loss": 0.5052,
+      "step": 4667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7038984354962815e-05,
+      "loss": 0.4712,
+      "step": 4668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703834316491409e-05,
+      "loss": 0.6604,
+      "step": 4669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703770197486535e-05,
+      "loss": 0.6457,
+      "step": 4670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7037060784816623e-05,
+      "loss": 0.6597,
+      "step": 4671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703641959476789e-05,
+      "loss": 0.547,
+      "step": 4672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703577840471916e-05,
+      "loss": 0.5944,
+      "step": 4673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703513721467043e-05,
+      "loss": 0.7664,
+      "step": 4674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70344960246217e-05,
+      "loss": 0.6026,
+      "step": 4675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703385483457297e-05,
+      "loss": 0.7211,
+      "step": 4676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703321364452424e-05,
+      "loss": 0.693,
+      "step": 4677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703257245447551e-05,
+      "loss": 0.5438,
+      "step": 4678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703193126442678e-05,
+      "loss": 0.6277,
+      "step": 4679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703129007437805e-05,
+      "loss": 0.6268,
+      "step": 4680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7030648884329314e-05,
+      "loss": 0.5416,
+      "step": 4681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7030007694280585e-05,
+      "loss": 0.6283,
+      "step": 4682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702936650423186e-05,
+      "loss": 0.6028,
+      "step": 4683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702872531418312e-05,
+      "loss": 0.6546,
+      "step": 4684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70280841241344e-05,
+      "loss": 0.5768,
+      "step": 4685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7027442934085665e-05,
+      "loss": 0.5909,
+      "step": 4686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702680174403694e-05,
+      "loss": 0.5863,
+      "step": 4687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70261605539882e-05,
+      "loss": 0.5708,
+      "step": 4688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7025519363939474e-05,
+      "loss": 0.6255,
+      "step": 4689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7024878173890745e-05,
+      "loss": 0.6281,
+      "step": 4690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702423698384201e-05,
+      "loss": 0.6446,
+      "step": 4691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702359579379328e-05,
+      "loss": 0.6442,
+      "step": 4692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7022954603744554e-05,
+      "loss": 0.5469,
+      "step": 4693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7022313413695826e-05,
+      "loss": 0.686,
+      "step": 4694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702167222364709e-05,
+      "loss": 0.5075,
+      "step": 4695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702103103359836e-05,
+      "loss": 0.5293,
+      "step": 4696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702038984354963e-05,
+      "loss": 0.566,
+      "step": 4697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70197486535009e-05,
+      "loss": 0.5746,
+      "step": 4698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701910746345217e-05,
+      "loss": 0.6299,
+      "step": 4699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7018466273403435e-05,
+      "loss": 0.6351,
+      "step": 4700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7017825083354714e-05,
+      "loss": 0.5723,
+      "step": 4701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701718389330598e-05,
+      "loss": 0.5745,
+      "step": 4702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701654270325725e-05,
+      "loss": 0.6462,
+      "step": 4703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7015901513208516e-05,
+      "loss": 0.4602,
+      "step": 4704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701526032315979e-05,
+      "loss": 0.6417,
+      "step": 4705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701461913311105e-05,
+      "loss": 0.6252,
+      "step": 4706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7013977943062324e-05,
+      "loss": 0.5382,
+      "step": 4707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7013336753013596e-05,
+      "loss": 0.5269,
+      "step": 4708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701269556296487e-05,
+      "loss": 0.6131,
+      "step": 4709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701205437291614e-05,
+      "loss": 0.63,
+      "step": 4710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7011413182867404e-05,
+      "loss": 0.6798,
+      "step": 4711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7010771992818676e-05,
+      "loss": 0.5975,
+      "step": 4712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.701013080276994e-05,
+      "loss": 0.5996,
+      "step": 4713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700948961272121e-05,
+      "loss": 0.5979,
+      "step": 4714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700884842267248e-05,
+      "loss": 0.6607,
+      "step": 4715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700820723262375e-05,
+      "loss": 0.56,
+      "step": 4716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700756604257502e-05,
+      "loss": 0.6063,
+      "step": 4717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700692485252629e-05,
+      "loss": 0.6724,
+      "step": 4718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7006283662477564e-05,
+      "loss": 0.5757,
+      "step": 4719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700564247242883e-05,
+      "loss": 0.6443,
+      "step": 4720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70050012823801e-05,
+      "loss": 0.6196,
+      "step": 4721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7004360092331366e-05,
+      "loss": 0.5893,
+      "step": 4722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700371890228264e-05,
+      "loss": 0.601,
+      "step": 4723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700307771223391e-05,
+      "loss": 0.6022,
+      "step": 4724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700243652218518e-05,
+      "loss": 0.6509,
+      "step": 4725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700179533213645e-05,
+      "loss": 0.6407,
+      "step": 4726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700115414208772e-05,
+      "loss": 0.5816,
+      "step": 4727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.700051295203899e-05,
+      "loss": 0.6475,
+      "step": 4728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6999871761990254e-05,
+      "loss": 0.6023,
+      "step": 4729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6999230571941526e-05,
+      "loss": 0.6682,
+      "step": 4730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699858938189279e-05,
+      "loss": 0.6494,
+      "step": 4731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699794819184406e-05,
+      "loss": 0.5818,
+      "step": 4732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6997307001795334e-05,
+      "loss": 0.6182,
+      "step": 4733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6996665811746606e-05,
+      "loss": 0.5798,
+      "step": 4734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699602462169788e-05,
+      "loss": 0.6428,
+      "step": 4735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699538343164914e-05,
+      "loss": 0.6252,
+      "step": 4736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6994742241600414e-05,
+      "loss": 0.6074,
+      "step": 4737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699410105155168e-05,
+      "loss": 0.5529,
+      "step": 4738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699345986150295e-05,
+      "loss": 0.6913,
+      "step": 4739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699281867145422e-05,
+      "loss": 0.6416,
+      "step": 4740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6992177481405495e-05,
+      "loss": 0.5797,
+      "step": 4741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699153629135676e-05,
+      "loss": 0.6534,
+      "step": 4742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699089510130803e-05,
+      "loss": 0.6146,
+      "step": 4743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69902539112593e-05,
+      "loss": 0.5675,
+      "step": 4744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698961272121057e-05,
+      "loss": 0.6391,
+      "step": 4745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698897153116184e-05,
+      "loss": 0.5494,
+      "step": 4746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6988330341113105e-05,
+      "loss": 0.5758,
+      "step": 4747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6987689151064376e-05,
+      "loss": 0.6139,
+      "step": 4748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698704796101565e-05,
+      "loss": 0.6301,
+      "step": 4749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698640677096692e-05,
+      "loss": 0.5869,
+      "step": 4750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6985765580918185e-05,
+      "loss": 0.6063,
+      "step": 4751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6985124390869456e-05,
+      "loss": 0.6585,
+      "step": 4752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698448320082073e-05,
+      "loss": 0.5885,
+      "step": 4753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698384201077199e-05,
+      "loss": 0.6407,
+      "step": 4754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6983200820723265e-05,
+      "loss": 0.5817,
+      "step": 4755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698255963067453e-05,
+      "loss": 0.5083,
+      "step": 4756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698191844062581e-05,
+      "loss": 0.609,
+      "step": 4757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698127725057707e-05,
+      "loss": 0.6464,
+      "step": 4758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6980636060528345e-05,
+      "loss": 0.5647,
+      "step": 4759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697999487047961e-05,
+      "loss": 0.6387,
+      "step": 4760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697935368043088e-05,
+      "loss": 0.654,
+      "step": 4761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697871249038215e-05,
+      "loss": 0.6366,
+      "step": 4762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697807130033342e-05,
+      "loss": 0.5495,
+      "step": 4763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697743011028469e-05,
+      "loss": 0.587,
+      "step": 4764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697678892023596e-05,
+      "loss": 0.6368,
+      "step": 4765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697614773018723e-05,
+      "loss": 0.6009,
+      "step": 4766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69755065401385e-05,
+      "loss": 0.5486,
+      "step": 4767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697486535008977e-05,
+      "loss": 0.5708,
+      "step": 4768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6974224160041035e-05,
+      "loss": 0.6205,
+      "step": 4769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6973582969992307e-05,
+      "loss": 0.5895,
+      "step": 4770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697294177994358e-05,
+      "loss": 0.6426,
+      "step": 4771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697230058989484e-05,
+      "loss": 0.6063,
+      "step": 4772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697165939984612e-05,
+      "loss": 0.5654,
+      "step": 4773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697101820979739e-05,
+      "loss": 0.5334,
+      "step": 4774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697037701974866e-05,
+      "loss": 0.559,
+      "step": 4775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696973582969992e-05,
+      "loss": 0.5443,
+      "step": 4776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6969094639651195e-05,
+      "loss": 0.5247,
+      "step": 4777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696845344960246e-05,
+      "loss": 0.6458,
+      "step": 4778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696781225955373e-05,
+      "loss": 0.6179,
+      "step": 4779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6967171069505003e-05,
+      "loss": 0.5594,
+      "step": 4780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6966529879456275e-05,
+      "loss": 0.6443,
+      "step": 4781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696588868940755e-05,
+      "loss": 0.6364,
+      "step": 4782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696524749935881e-05,
+      "loss": 0.5997,
+      "step": 4783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6964606309310084e-05,
+      "loss": 0.6246,
+      "step": 4784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696396511926135e-05,
+      "loss": 0.562,
+      "step": 4785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696332392921262e-05,
+      "loss": 0.6093,
+      "step": 4786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6962682739163885e-05,
+      "loss": 0.6156,
+      "step": 4787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696204154911516e-05,
+      "loss": 0.5873,
+      "step": 4788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6961400359066435e-05,
+      "loss": 0.7025,
+      "step": 4789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69607591690177e-05,
+      "loss": 0.5956,
+      "step": 4790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.696011797896897e-05,
+      "loss": 0.5442,
+      "step": 4791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695947678892024e-05,
+      "loss": 0.571,
+      "step": 4792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695883559887151e-05,
+      "loss": 0.5825,
+      "step": 4793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6958194408822774e-05,
+      "loss": 0.6007,
+      "step": 4794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6957553218774045e-05,
+      "loss": 0.6088,
+      "step": 4795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695691202872532e-05,
+      "loss": 0.6915,
+      "step": 4796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695627083867659e-05,
+      "loss": 0.5742,
+      "step": 4797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695562964862786e-05,
+      "loss": 0.5361,
+      "step": 4798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6954988458579125e-05,
+      "loss": 0.5461,
+      "step": 4799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69543472685304e-05,
+      "loss": 0.6443,
+      "step": 4800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695370607848166e-05,
+      "loss": 0.5922,
+      "step": 4801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6953064888432934e-05,
+      "loss": 0.5705,
+      "step": 4802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69524236983842e-05,
+      "loss": 0.526,
+      "step": 4803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695178250833547e-05,
+      "loss": 0.5989,
+      "step": 4804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.695114131828674e-05,
+      "loss": 0.6113,
+      "step": 4805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6950500128238014e-05,
+      "loss": 0.6128,
+      "step": 4806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6949858938189286e-05,
+      "loss": 0.661,
+      "step": 4807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694921774814055e-05,
+      "loss": 0.5885,
+      "step": 4808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694857655809182e-05,
+      "loss": 0.6166,
+      "step": 4809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694793536804309e-05,
+      "loss": 0.6241,
+      "step": 4810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694729417799436e-05,
+      "loss": 0.5964,
+      "step": 4811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694665298794563e-05,
+      "loss": 0.5769,
+      "step": 4812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69460117978969e-05,
+      "loss": 0.6301,
+      "step": 4813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694537060784817e-05,
+      "loss": 0.6181,
+      "step": 4814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694472941779944e-05,
+      "loss": 0.6564,
+      "step": 4815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694408822775071e-05,
+      "loss": 0.5523,
+      "step": 4816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6943447037701976e-05,
+      "loss": 0.5761,
+      "step": 4817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694280584765325e-05,
+      "loss": 0.5884,
+      "step": 4818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694216465760451e-05,
+      "loss": 0.6512,
+      "step": 4819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6941523467555784e-05,
+      "loss": 0.5287,
+      "step": 4820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6940882277507056e-05,
+      "loss": 0.5834,
+      "step": 4821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694024108745833e-05,
+      "loss": 0.5967,
+      "step": 4822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693959989740959e-05,
+      "loss": 0.5779,
+      "step": 4823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6938958707360864e-05,
+      "loss": 0.5309,
+      "step": 4824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6938317517312136e-05,
+      "loss": 0.5955,
+      "step": 4825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69376763272634e-05,
+      "loss": 0.4966,
+      "step": 4826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693703513721467e-05,
+      "loss": 0.5817,
+      "step": 4827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6936393947165944e-05,
+      "loss": 0.6193,
+      "step": 4828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6935752757117216e-05,
+      "loss": 0.5906,
+      "step": 4829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693511156706848e-05,
+      "loss": 0.6704,
+      "step": 4830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693447037701975e-05,
+      "loss": 0.5177,
+      "step": 4831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693382918697102e-05,
+      "loss": 0.549,
+      "step": 4832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693318799692229e-05,
+      "loss": 0.6151,
+      "step": 4833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693254680687356e-05,
+      "loss": 0.5557,
+      "step": 4834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6931905616824826e-05,
+      "loss": 0.5102,
+      "step": 4835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69312644267761e-05,
+      "loss": 0.4801,
+      "step": 4836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693062323672737e-05,
+      "loss": 0.5567,
+      "step": 4837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692998204667864e-05,
+      "loss": 0.5381,
+      "step": 4838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6929340856629906e-05,
+      "loss": 0.571,
+      "step": 4839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692869966658118e-05,
+      "loss": 0.5207,
+      "step": 4840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692805847653244e-05,
+      "loss": 0.5807,
+      "step": 4841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6927417286483714e-05,
+      "loss": 0.6093,
+      "step": 4842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6926776096434986e-05,
+      "loss": 0.6211,
+      "step": 4843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692613490638625e-05,
+      "loss": 0.6314,
+      "step": 4844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692549371633753e-05,
+      "loss": 0.6207,
+      "step": 4845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6924852526288794e-05,
+      "loss": 0.6268,
+      "step": 4846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6924211336240066e-05,
+      "loss": 0.6785,
+      "step": 4847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692357014619133e-05,
+      "loss": 0.5979,
+      "step": 4848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69229289561426e-05,
+      "loss": 0.6942,
+      "step": 4849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692228776609387e-05,
+      "loss": 0.5929,
+      "step": 4850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692164657604514e-05,
+      "loss": 0.5063,
+      "step": 4851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692100538599641e-05,
+      "loss": 0.5387,
+      "step": 4852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.692036419594768e-05,
+      "loss": 0.6208,
+      "step": 4853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6919723005898955e-05,
+      "loss": 0.6011,
+      "step": 4854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691908181585022e-05,
+      "loss": 0.5025,
+      "step": 4855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691844062580149e-05,
+      "loss": 0.638,
+      "step": 4856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6917799435752756e-05,
+      "loss": 0.5437,
+      "step": 4857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691715824570403e-05,
+      "loss": 0.5296,
+      "step": 4858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691651705565529e-05,
+      "loss": 0.6477,
+      "step": 4859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6915875865606565e-05,
+      "loss": 0.5679,
+      "step": 4860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691523467555784e-05,
+      "loss": 0.6199,
+      "step": 4861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691459348550911e-05,
+      "loss": 0.566,
+      "step": 4862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691395229546038e-05,
+      "loss": 0.6112,
+      "step": 4863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6913311105411645e-05,
+      "loss": 0.5627,
+      "step": 4864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6912669915362916e-05,
+      "loss": 0.561,
+      "step": 4865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691202872531418e-05,
+      "loss": 0.6431,
+      "step": 4866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691138753526545e-05,
+      "loss": 0.5931,
+      "step": 4867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6910746345216725e-05,
+      "loss": 0.5877,
+      "step": 4868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6910105155167996e-05,
+      "loss": 0.5925,
+      "step": 4869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690946396511927e-05,
+      "loss": 0.5967,
+      "step": 4870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690882277507053e-05,
+      "loss": 0.503,
+      "step": 4871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6908181585021805e-05,
+      "loss": 0.5351,
+      "step": 4872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690754039497307e-05,
+      "loss": 0.53,
+      "step": 4873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690689920492434e-05,
+      "loss": 0.5574,
+      "step": 4874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6906258014875606e-05,
+      "loss": 0.5128,
+      "step": 4875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690561682482688e-05,
+      "loss": 0.6416,
+      "step": 4876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690497563477816e-05,
+      "loss": 0.5695,
+      "step": 4877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690433444472942e-05,
+      "loss": 0.5595,
+      "step": 4878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690369325468069e-05,
+      "loss": 0.5277,
+      "step": 4879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690305206463196e-05,
+      "loss": 0.5877,
+      "step": 4880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690241087458323e-05,
+      "loss": 0.5838,
+      "step": 4881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6901769684534495e-05,
+      "loss": 0.6062,
+      "step": 4882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690112849448577e-05,
+      "loss": 0.6655,
+      "step": 4883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690048730443704e-05,
+      "loss": 0.5386,
+      "step": 4884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689984611438831e-05,
+      "loss": 0.6526,
+      "step": 4885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689920492433958e-05,
+      "loss": 0.5867,
+      "step": 4886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689856373429085e-05,
+      "loss": 0.6654,
+      "step": 4887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689792254424212e-05,
+      "loss": 0.6673,
+      "step": 4888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689728135419338e-05,
+      "loss": 0.4851,
+      "step": 4889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6896640164144655e-05,
+      "loss": 0.6594,
+      "step": 4890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689599897409592e-05,
+      "loss": 0.5233,
+      "step": 4891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689535778404719e-05,
+      "loss": 0.6422,
+      "step": 4892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6894716593998463e-05,
+      "loss": 0.6083,
+      "step": 4893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6894075403949735e-05,
+      "loss": 0.4848,
+      "step": 4894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689343421390101e-05,
+      "loss": 0.5765,
+      "step": 4895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689279302385227e-05,
+      "loss": 0.5261,
+      "step": 4896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6892151833803544e-05,
+      "loss": 0.679,
+      "step": 4897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689151064375481e-05,
+      "loss": 0.5814,
+      "step": 4898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689086945370608e-05,
+      "loss": 0.5189,
+      "step": 4899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689022826365735e-05,
+      "loss": 0.5318,
+      "step": 4900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6889587073608624e-05,
+      "loss": 0.606,
+      "step": 4901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688894588355989e-05,
+      "loss": 0.6135,
+      "step": 4902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688830469351116e-05,
+      "loss": 0.6067,
+      "step": 4903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688766350346243e-05,
+      "loss": 0.5142,
+      "step": 4904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68870223134137e-05,
+      "loss": 0.6499,
+      "step": 4905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688638112336497e-05,
+      "loss": 0.5905,
+      "step": 4906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6885739933316234e-05,
+      "loss": 0.6693,
+      "step": 4907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6885098743267505e-05,
+      "loss": 0.5773,
+      "step": 4908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688445755321878e-05,
+      "loss": 0.6536,
+      "step": 4909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688381636317005e-05,
+      "loss": 0.6502,
+      "step": 4910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6883175173121314e-05,
+      "loss": 0.6023,
+      "step": 4911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6882533983072585e-05,
+      "loss": 0.5624,
+      "step": 4912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688189279302386e-05,
+      "loss": 0.587,
+      "step": 4913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688125160297512e-05,
+      "loss": 0.5743,
+      "step": 4914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6880610412926394e-05,
+      "loss": 0.5296,
+      "step": 4915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6879969222877666e-05,
+      "loss": 0.5974,
+      "step": 4916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687932803282894e-05,
+      "loss": 0.6777,
+      "step": 4917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68786868427802e-05,
+      "loss": 0.566,
+      "step": 4918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6878045652731474e-05,
+      "loss": 0.6246,
+      "step": 4919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687740446268274e-05,
+      "loss": 0.5538,
+      "step": 4920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687676327263401e-05,
+      "loss": 0.5939,
+      "step": 4921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687612208258528e-05,
+      "loss": 0.6284,
+      "step": 4922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687548089253655e-05,
+      "loss": 0.6213,
+      "step": 4923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687483970248782e-05,
+      "loss": 0.6166,
+      "step": 4924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687419851243909e-05,
+      "loss": 0.5709,
+      "step": 4925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687355732239036e-05,
+      "loss": 0.6103,
+      "step": 4926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687291613234163e-05,
+      "loss": 0.5543,
+      "step": 4927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68722749422929e-05,
+      "loss": 0.6566,
+      "step": 4928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6871633752244164e-05,
+      "loss": 0.5986,
+      "step": 4929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6870992562195436e-05,
+      "loss": 0.6072,
+      "step": 4930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687035137214671e-05,
+      "loss": 0.5912,
+      "step": 4931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686971018209797e-05,
+      "loss": 0.5631,
+      "step": 4932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686906899204925e-05,
+      "loss": 0.6023,
+      "step": 4933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6868427802000516e-05,
+      "loss": 0.5959,
+      "step": 4934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686778661195179e-05,
+      "loss": 0.5347,
+      "step": 4935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686714542190305e-05,
+      "loss": 0.6858,
+      "step": 4936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6866504231854324e-05,
+      "loss": 0.5683,
+      "step": 4937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686586304180559e-05,
+      "loss": 0.5273,
+      "step": 4938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686522185175686e-05,
+      "loss": 0.5746,
+      "step": 4939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686458066170813e-05,
+      "loss": 0.7113,
+      "step": 4940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6863939471659404e-05,
+      "loss": 0.6121,
+      "step": 4941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6863298281610676e-05,
+      "loss": 0.572,
+      "step": 4942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686265709156194e-05,
+      "loss": 0.5806,
+      "step": 4943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686201590151321e-05,
+      "loss": 0.5361,
+      "step": 4944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686137471146448e-05,
+      "loss": 0.5706,
+      "step": 4945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686073352141575e-05,
+      "loss": 0.5275,
+      "step": 4946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6860092331367014e-05,
+      "loss": 0.5798,
+      "step": 4947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6859451141318286e-05,
+      "loss": 0.5655,
+      "step": 4948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6858809951269564e-05,
+      "loss": 0.576,
+      "step": 4949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685816876122083e-05,
+      "loss": 0.6035,
+      "step": 4950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68575275711721e-05,
+      "loss": 0.616,
+      "step": 4951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6856886381123366e-05,
+      "loss": 0.5336,
+      "step": 4952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685624519107464e-05,
+      "loss": 0.5261,
+      "step": 4953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68556040010259e-05,
+      "loss": 0.623,
+      "step": 4954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6854962810977174e-05,
+      "loss": 0.5751,
+      "step": 4955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6854321620928446e-05,
+      "loss": 0.5982,
+      "step": 4956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685368043087972e-05,
+      "loss": 0.5458,
+      "step": 4957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685303924083099e-05,
+      "loss": 0.7074,
+      "step": 4958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6852398050782254e-05,
+      "loss": 0.7939,
+      "step": 4959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6851756860733526e-05,
+      "loss": 0.6492,
+      "step": 4960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685111567068479e-05,
+      "loss": 0.6039,
+      "step": 4961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685047448063606e-05,
+      "loss": 0.5248,
+      "step": 4962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684983329058733e-05,
+      "loss": 0.5626,
+      "step": 4963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68491921005386e-05,
+      "loss": 0.5988,
+      "step": 4964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684855091048987e-05,
+      "loss": 0.6549,
+      "step": 4965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684790972044114e-05,
+      "loss": 0.634,
+      "step": 4966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6847268530392415e-05,
+      "loss": 0.6306,
+      "step": 4967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684662734034368e-05,
+      "loss": 0.6658,
+      "step": 4968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684598615029495e-05,
+      "loss": 0.5903,
+      "step": 4969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6845344960246216e-05,
+      "loss": 0.6242,
+      "step": 4970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684470377019749e-05,
+      "loss": 0.5466,
+      "step": 4971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684406258014876e-05,
+      "loss": 0.6745,
+      "step": 4972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684342139010003e-05,
+      "loss": 0.5747,
+      "step": 4973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6842780200051296e-05,
+      "loss": 0.6543,
+      "step": 4974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684213901000257e-05,
+      "loss": 0.6193,
+      "step": 4975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684149781995384e-05,
+      "loss": 0.5615,
+      "step": 4976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6840856629905105e-05,
+      "loss": 0.6949,
+      "step": 4977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6840215439856376e-05,
+      "loss": 0.5501,
+      "step": 4978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683957424980764e-05,
+      "loss": 0.6365,
+      "step": 4979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683893305975891e-05,
+      "loss": 0.6164,
+      "step": 4980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6838291869710185e-05,
+      "loss": 0.6008,
+      "step": 4981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6837650679661456e-05,
+      "loss": 0.5691,
+      "step": 4982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683700948961272e-05,
+      "loss": 0.5886,
+      "step": 4983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683636829956399e-05,
+      "loss": 0.6141,
+      "step": 4984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6835727109515265e-05,
+      "loss": 0.6997,
+      "step": 4985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683508591946653e-05,
+      "loss": 0.5545,
+      "step": 4986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68344447294178e-05,
+      "loss": 0.6489,
+      "step": 4987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683380353936907e-05,
+      "loss": 0.5465,
+      "step": 4988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6833162349320345e-05,
+      "loss": 0.6494,
+      "step": 4989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683252115927161e-05,
+      "loss": 0.629,
+      "step": 4990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683187996922288e-05,
+      "loss": 0.5672,
+      "step": 4991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6831238779174147e-05,
+      "loss": 0.6275,
+      "step": 4992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.683059758912542e-05,
+      "loss": 0.5966,
+      "step": 4993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682995639907669e-05,
+      "loss": 0.5555,
+      "step": 4994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6829315209027955e-05,
+      "loss": 0.6053,
+      "step": 4995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682867401897923e-05,
+      "loss": 0.6393,
+      "step": 4996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68280328289305e-05,
+      "loss": 0.6314,
+      "step": 4997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682739163888177e-05,
+      "loss": 0.5826,
+      "step": 4998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6826750448833035e-05,
+      "loss": 0.5955,
+      "step": 4999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682610925878431e-05,
+      "loss": 0.5731,
+      "step": 5000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682546806873557e-05,
+      "loss": 0.6276,
+      "step": 5001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6824826878686843e-05,
+      "loss": 0.5248,
+      "step": 5002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6824185688638115e-05,
+      "loss": 0.6036,
+      "step": 5003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682354449858938e-05,
+      "loss": 0.5736,
+      "step": 5004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682290330854066e-05,
+      "loss": 0.6041,
+      "step": 5005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6822262118491923e-05,
+      "loss": 0.6786,
+      "step": 5006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6821620928443195e-05,
+      "loss": 0.5703,
+      "step": 5007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682097973839446e-05,
+      "loss": 0.6433,
+      "step": 5008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.682033854834573e-05,
+      "loss": 0.5022,
+      "step": 5009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6819697358297e-05,
+      "loss": 0.637,
+      "step": 5010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681905616824827e-05,
+      "loss": 0.6253,
+      "step": 5011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681841497819954e-05,
+      "loss": 0.6766,
+      "step": 5012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681777378815081e-05,
+      "loss": 0.5623,
+      "step": 5013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6817132598102084e-05,
+      "loss": 0.5592,
+      "step": 5014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681649140805335e-05,
+      "loss": 0.5491,
+      "step": 5015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681585021800462e-05,
+      "loss": 0.5824,
+      "step": 5016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6815209027955885e-05,
+      "loss": 0.5621,
+      "step": 5017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681456783790716e-05,
+      "loss": 0.5155,
+      "step": 5018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681392664785842e-05,
+      "loss": 0.5783,
+      "step": 5019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6813285457809694e-05,
+      "loss": 0.5397,
+      "step": 5020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681264426776097e-05,
+      "loss": 0.5839,
+      "step": 5021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681200307771224e-05,
+      "loss": 0.6245,
+      "step": 5022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681136188766351e-05,
+      "loss": 0.6327,
+      "step": 5023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6810720697614774e-05,
+      "loss": 0.5642,
+      "step": 5024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6810079507566045e-05,
+      "loss": 0.5198,
+      "step": 5025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680943831751731e-05,
+      "loss": 0.58,
+      "step": 5026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680879712746858e-05,
+      "loss": 0.6232,
+      "step": 5027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6808155937419854e-05,
+      "loss": 0.6513,
+      "step": 5028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6807514747371126e-05,
+      "loss": 0.6342,
+      "step": 5029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68068735573224e-05,
+      "loss": 0.6206,
+      "step": 5030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680623236727366e-05,
+      "loss": 0.585,
+      "step": 5031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6805591177224934e-05,
+      "loss": 0.6504,
+      "step": 5032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68049499871762e-05,
+      "loss": 0.6458,
+      "step": 5033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680430879712747e-05,
+      "loss": 0.5137,
+      "step": 5034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6803667607078736e-05,
+      "loss": 0.5859,
+      "step": 5035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680302641703001e-05,
+      "loss": 0.554,
+      "step": 5036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680238522698128e-05,
+      "loss": 0.5356,
+      "step": 5037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680174403693255e-05,
+      "loss": 0.6332,
+      "step": 5038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680110284688382e-05,
+      "loss": 0.5437,
+      "step": 5039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.680046165683509e-05,
+      "loss": 0.6007,
+      "step": 5040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679982046678636e-05,
+      "loss": 0.5872,
+      "step": 5041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6799179276737624e-05,
+      "loss": 0.6196,
+      "step": 5042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6798538086688896e-05,
+      "loss": 0.6961,
+      "step": 5043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679789689664017e-05,
+      "loss": 0.5139,
+      "step": 5044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679725570659144e-05,
+      "loss": 0.4955,
+      "step": 5045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6796614516542704e-05,
+      "loss": 0.5502,
+      "step": 5046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6795973326493976e-05,
+      "loss": 0.577,
+      "step": 5047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679533213644525e-05,
+      "loss": 0.5314,
+      "step": 5048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679469094639651e-05,
+      "loss": 0.5513,
+      "step": 5049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6794049756347784e-05,
+      "loss": 0.5778,
+      "step": 5050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679340856629905e-05,
+      "loss": 0.6601,
+      "step": 5051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679276737625032e-05,
+      "loss": 0.5792,
+      "step": 5052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679212618620159e-05,
+      "loss": 0.5804,
+      "step": 5053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6791484996152864e-05,
+      "loss": 0.6209,
+      "step": 5054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679084380610413e-05,
+      "loss": 0.6258,
+      "step": 5055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.67902026160554e-05,
+      "loss": 0.6085,
+      "step": 5056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678956142600667e-05,
+      "loss": 0.6191,
+      "step": 5057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678892023595794e-05,
+      "loss": 0.5227,
+      "step": 5058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678827904590921e-05,
+      "loss": 0.5953,
+      "step": 5059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678763785586048e-05,
+      "loss": 0.5491,
+      "step": 5060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678699666581175e-05,
+      "loss": 0.5294,
+      "step": 5061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678635547576302e-05,
+      "loss": 0.6189,
+      "step": 5062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678571428571429e-05,
+      "loss": 0.5825,
+      "step": 5063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6785073095665554e-05,
+      "loss": 0.677,
+      "step": 5064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6784431905616826e-05,
+      "loss": 0.5888,
+      "step": 5065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.67837907155681e-05,
+      "loss": 0.6305,
+      "step": 5066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678314952551936e-05,
+      "loss": 0.588,
+      "step": 5067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6782508335470634e-05,
+      "loss": 0.6132,
+      "step": 5068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6781867145421906e-05,
+      "loss": 0.6411,
+      "step": 5069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678122595537318e-05,
+      "loss": 0.5999,
+      "step": 5070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678058476532444e-05,
+      "loss": 0.7271,
+      "step": 5071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6779943575275714e-05,
+      "loss": 0.573,
+      "step": 5072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6779302385226986e-05,
+      "loss": 0.5538,
+      "step": 5073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677866119517825e-05,
+      "loss": 0.654,
+      "step": 5074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677802000512952e-05,
+      "loss": 0.5867,
+      "step": 5075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6777378815080795e-05,
+      "loss": 0.5143,
+      "step": 5076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6776737625032066e-05,
+      "loss": 0.5868,
+      "step": 5077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677609643498333e-05,
+      "loss": 0.5615,
+      "step": 5078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67754552449346e-05,
+      "loss": 0.5795,
+      "step": 5079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677481405488587e-05,
+      "loss": 0.6157,
+      "step": 5080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677417286483714e-05,
+      "loss": 0.6026,
+      "step": 5081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677353167478841e-05,
+      "loss": 0.5639,
+      "step": 5082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6772890484739676e-05,
+      "loss": 0.5639,
+      "step": 5083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677224929469095e-05,
+      "loss": 0.567,
+      "step": 5084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677160810464222e-05,
+      "loss": 0.6273,
+      "step": 5085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677096691459349e-05,
+      "loss": 0.5433,
+      "step": 5086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6770325724544756e-05,
+      "loss": 0.5786,
+      "step": 5087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676968453449603e-05,
+      "loss": 0.6174,
+      "step": 5088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676904334444729e-05,
+      "loss": 0.6047,
+      "step": 5089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6768402154398565e-05,
+      "loss": 0.5787,
+      "step": 5090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6767760964349836e-05,
+      "loss": 0.6221,
+      "step": 5091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67671197743011e-05,
+      "loss": 0.6555,
+      "step": 5092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676647858425238e-05,
+      "loss": 0.6033,
+      "step": 5093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6765837394203645e-05,
+      "loss": 0.5402,
+      "step": 5094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6765196204154917e-05,
+      "loss": 0.5503,
+      "step": 5095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676455501410618e-05,
+      "loss": 0.5423,
+      "step": 5096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676391382405745e-05,
+      "loss": 0.6089,
+      "step": 5097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676327263400872e-05,
+      "loss": 0.596,
+      "step": 5098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676263144395999e-05,
+      "loss": 0.6027,
+      "step": 5099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676199025391126e-05,
+      "loss": 0.6439,
+      "step": 5100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676134906386253e-05,
+      "loss": 0.5456,
+      "step": 5101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6760707873813805e-05,
+      "loss": 0.5536,
+      "step": 5102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676006668376507e-05,
+      "loss": 0.5809,
+      "step": 5103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675942549371634e-05,
+      "loss": 0.5418,
+      "step": 5104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6758784303667607e-05,
+      "loss": 0.5793,
+      "step": 5105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675814311361888e-05,
+      "loss": 0.6781,
+      "step": 5106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675750192357014e-05,
+      "loss": 0.5759,
+      "step": 5107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6756860733521415e-05,
+      "loss": 0.5901,
+      "step": 5108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6756219543472693e-05,
+      "loss": 0.6193,
+      "step": 5109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675557835342396e-05,
+      "loss": 0.488,
+      "step": 5110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675493716337523e-05,
+      "loss": 0.6391,
+      "step": 5111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6754295973326495e-05,
+      "loss": 0.5785,
+      "step": 5112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675365478327777e-05,
+      "loss": 0.5507,
+      "step": 5113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675301359322903e-05,
+      "loss": 0.6194,
+      "step": 5114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6752372403180303e-05,
+      "loss": 0.6225,
+      "step": 5115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6751731213131575e-05,
+      "loss": 0.7444,
+      "step": 5116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675109002308285e-05,
+      "loss": 0.557,
+      "step": 5117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675044883303412e-05,
+      "loss": 0.6107,
+      "step": 5118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6749807642985384e-05,
+      "loss": 0.5304,
+      "step": 5119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6749166452936655e-05,
+      "loss": 0.6371,
+      "step": 5120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674852526288792e-05,
+      "loss": 0.6386,
+      "step": 5121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674788407283919e-05,
+      "loss": 0.6243,
+      "step": 5122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674724288279046e-05,
+      "loss": 0.5942,
+      "step": 5123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674660169274173e-05,
+      "loss": 0.6242,
+      "step": 5124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6745960502693e-05,
+      "loss": 0.5982,
+      "step": 5125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674531931264427e-05,
+      "loss": 0.5626,
+      "step": 5126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6744678122595544e-05,
+      "loss": 0.5969,
+      "step": 5127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674403693254681e-05,
+      "loss": 0.5378,
+      "step": 5128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674339574249808e-05,
+      "loss": 0.6174,
+      "step": 5129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6742754552449345e-05,
+      "loss": 0.5376,
+      "step": 5130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674211336240062e-05,
+      "loss": 0.5881,
+      "step": 5131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674147217235189e-05,
+      "loss": 0.5697,
+      "step": 5132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674083098230316e-05,
+      "loss": 0.6,
+      "step": 5133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6740189792254425e-05,
+      "loss": 0.5468,
+      "step": 5134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67395486022057e-05,
+      "loss": 0.546,
+      "step": 5135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673890741215697e-05,
+      "loss": 0.4769,
+      "step": 5136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6738266222108234e-05,
+      "loss": 0.5646,
+      "step": 5137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6737625032059505e-05,
+      "loss": 0.5868,
+      "step": 5138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673698384201077e-05,
+      "loss": 0.5825,
+      "step": 5139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673634265196204e-05,
+      "loss": 0.4626,
+      "step": 5140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6735701461913314e-05,
+      "loss": 0.628,
+      "step": 5141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6735060271864586e-05,
+      "loss": 0.5928,
+      "step": 5142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673441908181585e-05,
+      "loss": 0.5738,
+      "step": 5143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673377789176712e-05,
+      "loss": 0.568,
+      "step": 5144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6733136701718394e-05,
+      "loss": 0.5169,
+      "step": 5145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673249551166966e-05,
+      "loss": 0.512,
+      "step": 5146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673185432162093e-05,
+      "loss": 0.634,
+      "step": 5147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67312131315722e-05,
+      "loss": 0.6587,
+      "step": 5148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6730571941523474e-05,
+      "loss": 0.5889,
+      "step": 5149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672993075147474e-05,
+      "loss": 0.6714,
+      "step": 5150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672928956142601e-05,
+      "loss": 0.5582,
+      "step": 5151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6728648371377276e-05,
+      "loss": 0.5783,
+      "step": 5152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672800718132855e-05,
+      "loss": 0.6975,
+      "step": 5153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672736599127982e-05,
+      "loss": 0.5943,
+      "step": 5154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6726724801231084e-05,
+      "loss": 0.5627,
+      "step": 5155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6726083611182356e-05,
+      "loss": 0.5018,
+      "step": 5156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672544242113363e-05,
+      "loss": 0.6527,
+      "step": 5157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67248012310849e-05,
+      "loss": 0.6742,
+      "step": 5158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6724160041036164e-05,
+      "loss": 0.5349,
+      "step": 5159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6723518850987436e-05,
+      "loss": 0.6326,
+      "step": 5160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67228776609387e-05,
+      "loss": 0.6218,
+      "step": 5161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672223647088997e-05,
+      "loss": 0.6196,
+      "step": 5162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6721595280841244e-05,
+      "loss": 0.6017,
+      "step": 5163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6720954090792516e-05,
+      "loss": 0.5623,
+      "step": 5164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672031290074379e-05,
+      "loss": 0.5695,
+      "step": 5165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671967171069505e-05,
+      "loss": 0.6177,
+      "step": 5166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6719030520646324e-05,
+      "loss": 0.5574,
+      "step": 5167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671838933059759e-05,
+      "loss": 0.5576,
+      "step": 5168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671774814054886e-05,
+      "loss": 0.58,
+      "step": 5169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6717106950500126e-05,
+      "loss": 0.5317,
+      "step": 5170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67164657604514e-05,
+      "loss": 0.5148,
+      "step": 5171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671582457040267e-05,
+      "loss": 0.6196,
+      "step": 5172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671518338035394e-05,
+      "loss": 0.6438,
+      "step": 5173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671454219030521e-05,
+      "loss": 0.6089,
+      "step": 5174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671390100025648e-05,
+      "loss": 0.6583,
+      "step": 5175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671325981020775e-05,
+      "loss": 0.5577,
+      "step": 5176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6712618620159014e-05,
+      "loss": 0.6854,
+      "step": 5177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6711977430110286e-05,
+      "loss": 0.5523,
+      "step": 5178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671133624006155e-05,
+      "loss": 0.698,
+      "step": 5179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671069505001282e-05,
+      "loss": 0.6223,
+      "step": 5180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67100538599641e-05,
+      "loss": 0.5072,
+      "step": 5181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6709412669915366e-05,
+      "loss": 0.546,
+      "step": 5182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670877147986664e-05,
+      "loss": 0.5211,
+      "step": 5183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67081302898179e-05,
+      "loss": 0.5804,
+      "step": 5184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6707489099769175e-05,
+      "loss": 0.5962,
+      "step": 5185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670684790972044e-05,
+      "loss": 0.6097,
+      "step": 5186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670620671967171e-05,
+      "loss": 0.5772,
+      "step": 5187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670556552962298e-05,
+      "loss": 0.5714,
+      "step": 5188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6704924339574255e-05,
+      "loss": 0.6043,
+      "step": 5189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6704283149525526e-05,
+      "loss": 0.639,
+      "step": 5190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670364195947679e-05,
+      "loss": 0.6737,
+      "step": 5191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670300076942806e-05,
+      "loss": 0.6327,
+      "step": 5192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670235957937933e-05,
+      "loss": 0.6581,
+      "step": 5193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67017183893306e-05,
+      "loss": 0.5306,
+      "step": 5194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6701077199281865e-05,
+      "loss": 0.6201,
+      "step": 5195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6700436009233136e-05,
+      "loss": 0.5382,
+      "step": 5196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669979481918441e-05,
+      "loss": 0.6625,
+      "step": 5197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669915362913568e-05,
+      "loss": 0.6208,
+      "step": 5198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669851243908695e-05,
+      "loss": 0.6254,
+      "step": 5199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6697871249038216e-05,
+      "loss": 0.624,
+      "step": 5200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669723005898949e-05,
+      "loss": 0.5401,
+      "step": 5201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669658886894075e-05,
+      "loss": 0.5562,
+      "step": 5202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6695947678892025e-05,
+      "loss": 0.5306,
+      "step": 5203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6695306488843296e-05,
+      "loss": 0.528,
+      "step": 5204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669466529879457e-05,
+      "loss": 0.5939,
+      "step": 5205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669402410874583e-05,
+      "loss": 0.6019,
+      "step": 5206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6693382918697105e-05,
+      "loss": 0.5868,
+      "step": 5207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6692741728648377e-05,
+      "loss": 0.6379,
+      "step": 5208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669210053859964e-05,
+      "loss": 0.5879,
+      "step": 5209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669145934855091e-05,
+      "loss": 0.6178,
+      "step": 5210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669081815850218e-05,
+      "loss": 0.5871,
+      "step": 5211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669017696845345e-05,
+      "loss": 0.5746,
+      "step": 5212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668953577840472e-05,
+      "loss": 0.5611,
+      "step": 5213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668889458835599e-05,
+      "loss": 0.5741,
+      "step": 5214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668825339830726e-05,
+      "loss": 0.5771,
+      "step": 5215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668761220825853e-05,
+      "loss": 0.5523,
+      "step": 5216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.66869710182098e-05,
+      "loss": 0.5295,
+      "step": 5217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668632982816107e-05,
+      "loss": 0.5763,
+      "step": 5218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668568863811234e-05,
+      "loss": 0.6377,
+      "step": 5219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668504744806361e-05,
+      "loss": 0.5776,
+      "step": 5220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668440625801488e-05,
+      "loss": 0.5157,
+      "step": 5221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668376506796615e-05,
+      "loss": 0.5782,
+      "step": 5222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668312387791742e-05,
+      "loss": 0.5513,
+      "step": 5223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668248268786868e-05,
+      "loss": 0.5229,
+      "step": 5224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6681841497819955e-05,
+      "loss": 0.6252,
+      "step": 5225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668120030777123e-05,
+      "loss": 0.655,
+      "step": 5226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668055911772249e-05,
+      "loss": 0.6757,
+      "step": 5227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6679917927673763e-05,
+      "loss": 0.6246,
+      "step": 5228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6679276737625035e-05,
+      "loss": 0.5907,
+      "step": 5229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667863554757631e-05,
+      "loss": 0.6046,
+      "step": 5230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667799435752757e-05,
+      "loss": 0.5686,
+      "step": 5231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6677353167478844e-05,
+      "loss": 0.5821,
+      "step": 5232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667671197743011e-05,
+      "loss": 0.6616,
+      "step": 5233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667607078738138e-05,
+      "loss": 0.5814,
+      "step": 5234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667542959733265e-05,
+      "loss": 0.5914,
+      "step": 5235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6674788407283924e-05,
+      "loss": 0.581,
+      "step": 5236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6674147217235195e-05,
+      "loss": 0.5701,
+      "step": 5237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667350602718646e-05,
+      "loss": 0.5772,
+      "step": 5238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667286483713773e-05,
+      "loss": 0.6093,
+      "step": 5239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6672223647089e-05,
+      "loss": 0.5635,
+      "step": 5240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667158245704027e-05,
+      "loss": 0.5625,
+      "step": 5241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6670941266991534e-05,
+      "loss": 0.5888,
+      "step": 5242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6670300076942805e-05,
+      "loss": 0.6004,
+      "step": 5243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666965888689408e-05,
+      "loss": 0.6015,
+      "step": 5244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666901769684535e-05,
+      "loss": 0.5768,
+      "step": 5245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666837650679662e-05,
+      "loss": 0.6725,
+      "step": 5246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6667735316747885e-05,
+      "loss": 0.4963,
+      "step": 5247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666709412669916e-05,
+      "loss": 0.5127,
+      "step": 5248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666645293665042e-05,
+      "loss": 0.6378,
+      "step": 5249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6665811746601694e-05,
+      "loss": 0.7011,
+      "step": 5250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666517055655296e-05,
+      "loss": 0.6631,
+      "step": 5251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666452936650423e-05,
+      "loss": 0.6314,
+      "step": 5252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666388817645551e-05,
+      "loss": 0.7052,
+      "step": 5253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6663246986406774e-05,
+      "loss": 0.5302,
+      "step": 5254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6662605796358046e-05,
+      "loss": 0.5898,
+      "step": 5255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666196460630931e-05,
+      "loss": 0.5882,
+      "step": 5256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666132341626058e-05,
+      "loss": 0.5771,
+      "step": 5257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666068222621185e-05,
+      "loss": 0.5704,
+      "step": 5258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666004103616312e-05,
+      "loss": 0.6902,
+      "step": 5259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665939984611439e-05,
+      "loss": 0.6392,
+      "step": 5260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665875865606566e-05,
+      "loss": 0.6004,
+      "step": 5261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6658117466016934e-05,
+      "loss": 0.5555,
+      "step": 5262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.66574762759682e-05,
+      "loss": 0.4959,
+      "step": 5263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665683508591947e-05,
+      "loss": 0.6183,
+      "step": 5264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6656193895870736e-05,
+      "loss": 0.5951,
+      "step": 5265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665555270582201e-05,
+      "loss": 0.5081,
+      "step": 5266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665491151577327e-05,
+      "loss": 0.5022,
+      "step": 5267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6654270325724544e-05,
+      "loss": 0.5994,
+      "step": 5268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665362913567582e-05,
+      "loss": 0.5952,
+      "step": 5269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665298794562709e-05,
+      "loss": 0.5023,
+      "step": 5270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665234675557836e-05,
+      "loss": 0.6063,
+      "step": 5271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6651705565529624e-05,
+      "loss": 0.5252,
+      "step": 5272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6651064375480896e-05,
+      "loss": 0.6505,
+      "step": 5273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665042318543216e-05,
+      "loss": 0.5819,
+      "step": 5274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664978199538343e-05,
+      "loss": 0.6907,
+      "step": 5275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6649140805334704e-05,
+      "loss": 0.7126,
+      "step": 5276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6648499615285976e-05,
+      "loss": 0.574,
+      "step": 5277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664785842523725e-05,
+      "loss": 0.59,
+      "step": 5278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664721723518851e-05,
+      "loss": 0.5663,
+      "step": 5279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6646576045139784e-05,
+      "loss": 0.6479,
+      "step": 5280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664593485509105e-05,
+      "loss": 0.5896,
+      "step": 5281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664529366504232e-05,
+      "loss": 0.6409,
+      "step": 5282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6644652474993586e-05,
+      "loss": 0.647,
+      "step": 5283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664401128494486e-05,
+      "loss": 0.6376,
+      "step": 5284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664337009489613e-05,
+      "loss": 0.5604,
+      "step": 5285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.66427289048474e-05,
+      "loss": 0.6132,
+      "step": 5286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664208771479867e-05,
+      "loss": 0.5876,
+      "step": 5287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664144652474994e-05,
+      "loss": 0.5175,
+      "step": 5288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664080533470121e-05,
+      "loss": 0.6641,
+      "step": 5289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6640164144652474e-05,
+      "loss": 0.6356,
+      "step": 5290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6639522954603746e-05,
+      "loss": 0.6144,
+      "step": 5291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663888176455502e-05,
+      "loss": 0.5879,
+      "step": 5292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663824057450629e-05,
+      "loss": 0.6119,
+      "step": 5293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6637599384457554e-05,
+      "loss": 0.5954,
+      "step": 5294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6636958194408826e-05,
+      "loss": 0.5908,
+      "step": 5295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.66363170043601e-05,
+      "loss": 0.6174,
+      "step": 5296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663567581431136e-05,
+      "loss": 0.6002,
+      "step": 5297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6635034624262635e-05,
+      "loss": 0.6199,
+      "step": 5298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.66343934342139e-05,
+      "loss": 0.6316,
+      "step": 5299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663375224416517e-05,
+      "loss": 0.5353,
+      "step": 5300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663311105411644e-05,
+      "loss": 0.5862,
+      "step": 5301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6632469864067715e-05,
+      "loss": 0.543,
+      "step": 5302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663182867401898e-05,
+      "loss": 0.5369,
+      "step": 5303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663118748397025e-05,
+      "loss": 0.6287,
+      "step": 5304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663054629392152e-05,
+      "loss": 0.6805,
+      "step": 5305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662990510387279e-05,
+      "loss": 0.6093,
+      "step": 5306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662926391382406e-05,
+      "loss": 0.5704,
+      "step": 5307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662862272377533e-05,
+      "loss": 0.5802,
+      "step": 5308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.66279815337266e-05,
+      "loss": 0.5539,
+      "step": 5309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662734034367787e-05,
+      "loss": 0.5823,
+      "step": 5310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662669915362914e-05,
+      "loss": 0.5331,
+      "step": 5311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6626057963580405e-05,
+      "loss": 0.5336,
+      "step": 5312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6625416773531676e-05,
+      "loss": 0.617,
+      "step": 5313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662477558348295e-05,
+      "loss": 0.5859,
+      "step": 5314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662413439343421e-05,
+      "loss": 0.6401,
+      "step": 5315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6623493203385485e-05,
+      "loss": 0.7276,
+      "step": 5316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6622852013336757e-05,
+      "loss": 0.568,
+      "step": 5317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662221082328803e-05,
+      "loss": 0.5588,
+      "step": 5318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662156963323929e-05,
+      "loss": 0.5681,
+      "step": 5319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6620928443190565e-05,
+      "loss": 0.53,
+      "step": 5320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662028725314183e-05,
+      "loss": 0.5828,
+      "step": 5321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.66196460630931e-05,
+      "loss": 0.73,
+      "step": 5322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.661900487304437e-05,
+      "loss": 0.6132,
+      "step": 5323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6618363682995645e-05,
+      "loss": 0.6479,
+      "step": 5324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.661772249294692e-05,
+      "loss": 0.6019,
+      "step": 5325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.661708130289818e-05,
+      "loss": 0.7477,
+      "step": 5326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.661644011284945e-05,
+      "loss": 0.6018,
+      "step": 5327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.661579892280072e-05,
+      "loss": 0.6638,
+      "step": 5328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.661515773275199e-05,
+      "loss": 0.5634,
+      "step": 5329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6614516542703255e-05,
+      "loss": 0.5584,
+      "step": 5330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.661387535265453e-05,
+      "loss": 0.6544,
+      "step": 5331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.66132341626058e-05,
+      "loss": 0.5249,
+      "step": 5332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.661259297255707e-05,
+      "loss": 0.6973,
+      "step": 5333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.661195178250834e-05,
+      "loss": 0.5914,
+      "step": 5334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.661131059245961e-05,
+      "loss": 0.5386,
+      "step": 5335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.661066940241088e-05,
+      "loss": 0.5955,
+      "step": 5336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6610028212362143e-05,
+      "loss": 0.5445,
+      "step": 5337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6609387022313415e-05,
+      "loss": 0.5693,
+      "step": 5338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660874583226468e-05,
+      "loss": 0.5038,
+      "step": 5339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660810464221595e-05,
+      "loss": 0.6442,
+      "step": 5340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660746345216723e-05,
+      "loss": 0.7669,
+      "step": 5341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6606822262118495e-05,
+      "loss": 0.519,
+      "step": 5342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660618107206977e-05,
+      "loss": 0.6603,
+      "step": 5343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660553988202103e-05,
+      "loss": 0.5469,
+      "step": 5344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6604898691972304e-05,
+      "loss": 0.6075,
+      "step": 5345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660425750192357e-05,
+      "loss": 0.566,
+      "step": 5346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660361631187484e-05,
+      "loss": 0.5811,
+      "step": 5347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660297512182611e-05,
+      "loss": 0.5052,
+      "step": 5348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6602333931777384e-05,
+      "loss": 0.6377,
+      "step": 5349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6601692741728655e-05,
+      "loss": 0.6139,
+      "step": 5350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660105155167992e-05,
+      "loss": 0.615,
+      "step": 5351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660041036163119e-05,
+      "loss": 0.6027,
+      "step": 5352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659976917158246e-05,
+      "loss": 0.574,
+      "step": 5353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659912798153373e-05,
+      "loss": 0.5316,
+      "step": 5354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6598486791484994e-05,
+      "loss": 0.591,
+      "step": 5355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6597845601436265e-05,
+      "loss": 0.543,
+      "step": 5356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659720441138754e-05,
+      "loss": 0.6273,
+      "step": 5357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659656322133881e-05,
+      "loss": 0.5608,
+      "step": 5358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659592203129008e-05,
+      "loss": 0.5836,
+      "step": 5359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6595280841241345e-05,
+      "loss": 0.6234,
+      "step": 5360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659463965119262e-05,
+      "loss": 0.5188,
+      "step": 5361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659399846114388e-05,
+      "loss": 0.4952,
+      "step": 5362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6593357271095154e-05,
+      "loss": 0.5285,
+      "step": 5363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6592716081046426e-05,
+      "loss": 0.5491,
+      "step": 5364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65920748909977e-05,
+      "loss": 0.5964,
+      "step": 5365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659143370094896e-05,
+      "loss": 0.6806,
+      "step": 5366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6590792510900234e-05,
+      "loss": 0.6277,
+      "step": 5367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6590151320851506e-05,
+      "loss": 0.5767,
+      "step": 5368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658951013080277e-05,
+      "loss": 0.454,
+      "step": 5369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658886894075404e-05,
+      "loss": 0.6661,
+      "step": 5370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658822775070531e-05,
+      "loss": 0.6588,
+      "step": 5371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658758656065658e-05,
+      "loss": 0.5507,
+      "step": 5372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658694537060785e-05,
+      "loss": 0.5832,
+      "step": 5373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658630418055912e-05,
+      "loss": 0.5431,
+      "step": 5374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658566299051039e-05,
+      "loss": 0.5797,
+      "step": 5375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658502180046166e-05,
+      "loss": 0.5506,
+      "step": 5376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658438061041293e-05,
+      "loss": 0.6467,
+      "step": 5377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6583739420364196e-05,
+      "loss": 0.8012,
+      "step": 5378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658309823031547e-05,
+      "loss": 0.5685,
+      "step": 5379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658245704026674e-05,
+      "loss": 0.5922,
+      "step": 5380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658181585021801e-05,
+      "loss": 0.5853,
+      "step": 5381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6581174660169276e-05,
+      "loss": 0.5345,
+      "step": 5382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658053347012055e-05,
+      "loss": 0.5917,
+      "step": 5383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657989228007181e-05,
+      "loss": 0.6386,
+      "step": 5384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6579251090023084e-05,
+      "loss": 0.6291,
+      "step": 5385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6578609899974356e-05,
+      "loss": 0.5439,
+      "step": 5386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657796870992562e-05,
+      "loss": 0.572,
+      "step": 5387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657732751987689e-05,
+      "loss": 0.6382,
+      "step": 5388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6576686329828164e-05,
+      "loss": 0.5574,
+      "step": 5389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6576045139779436e-05,
+      "loss": 0.5943,
+      "step": 5390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65754039497307e-05,
+      "loss": 0.4964,
+      "step": 5391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657476275968197e-05,
+      "loss": 0.6869,
+      "step": 5392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657412156963324e-05,
+      "loss": 0.5958,
+      "step": 5393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657348037958451e-05,
+      "loss": 0.5491,
+      "step": 5394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657283918953578e-05,
+      "loss": 0.5732,
+      "step": 5395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657219799948705e-05,
+      "loss": 0.5806,
+      "step": 5396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6571556809438324e-05,
+      "loss": 0.6242,
+      "step": 5397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657091561938959e-05,
+      "loss": 0.5502,
+      "step": 5398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657027442934086e-05,
+      "loss": 0.5814,
+      "step": 5399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6569633239292126e-05,
+      "loss": 0.6112,
+      "step": 5400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65689920492434e-05,
+      "loss": 0.6705,
+      "step": 5401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656835085919466e-05,
+      "loss": 0.5535,
+      "step": 5402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6567709669145934e-05,
+      "loss": 0.625,
+      "step": 5403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6567068479097206e-05,
+      "loss": 0.6209,
+      "step": 5404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656642728904848e-05,
+      "loss": 0.5386,
+      "step": 5405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656578609899975e-05,
+      "loss": 0.5421,
+      "step": 5406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6565144908951014e-05,
+      "loss": 0.5467,
+      "step": 5407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6564503718902286e-05,
+      "loss": 0.5501,
+      "step": 5408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656386252885355e-05,
+      "loss": 0.5158,
+      "step": 5409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656322133880482e-05,
+      "loss": 0.5581,
+      "step": 5410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656258014875609e-05,
+      "loss": 0.5994,
+      "step": 5411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656193895870736e-05,
+      "loss": 0.577,
+      "step": 5412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656129776865864e-05,
+      "loss": 0.5983,
+      "step": 5413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65606565786099e-05,
+      "loss": 0.5565,
+      "step": 5414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6560015388561175e-05,
+      "loss": 0.5526,
+      "step": 5415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655937419851244e-05,
+      "loss": 0.5496,
+      "step": 5416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655873300846371e-05,
+      "loss": 0.6581,
+      "step": 5417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6558091818414976e-05,
+      "loss": 0.5592,
+      "step": 5418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655745062836625e-05,
+      "loss": 0.5274,
+      "step": 5419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655680943831752e-05,
+      "loss": 0.7018,
+      "step": 5420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655616824826879e-05,
+      "loss": 0.5182,
+      "step": 5421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655552705822006e-05,
+      "loss": 0.5293,
+      "step": 5422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655488586817133e-05,
+      "loss": 0.5827,
+      "step": 5423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65542446781226e-05,
+      "loss": 0.7101,
+      "step": 5424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6553603488073865e-05,
+      "loss": 0.6239,
+      "step": 5425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6552962298025136e-05,
+      "loss": 0.6252,
+      "step": 5426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65523211079764e-05,
+      "loss": 0.6311,
+      "step": 5427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655167991792767e-05,
+      "loss": 0.6025,
+      "step": 5428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6551038727878945e-05,
+      "loss": 0.5547,
+      "step": 5429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6550397537830217e-05,
+      "loss": 0.6314,
+      "step": 5430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654975634778149e-05,
+      "loss": 0.632,
+      "step": 5431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654911515773275e-05,
+      "loss": 0.5568,
+      "step": 5432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6548473967684025e-05,
+      "loss": 0.5001,
+      "step": 5433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654783277763529e-05,
+      "loss": 0.6588,
+      "step": 5434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654719158758656e-05,
+      "loss": 0.5753,
+      "step": 5435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654655039753783e-05,
+      "loss": 0.6104,
+      "step": 5436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6545909207489105e-05,
+      "loss": 0.6258,
+      "step": 5437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654526801744037e-05,
+      "loss": 0.5521,
+      "step": 5438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654462682739164e-05,
+      "loss": 0.59,
+      "step": 5439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654398563734291e-05,
+      "loss": 0.5617,
+      "step": 5440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654334444729418e-05,
+      "loss": 0.5689,
+      "step": 5441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654270325724545e-05,
+      "loss": 0.6689,
+      "step": 5442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6542062067196715e-05,
+      "loss": 0.5564,
+      "step": 5443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654142087714799e-05,
+      "loss": 0.5268,
+      "step": 5444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654077968709926e-05,
+      "loss": 0.5685,
+      "step": 5445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654013849705053e-05,
+      "loss": 0.6208,
+      "step": 5446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65394973070018e-05,
+      "loss": 0.7016,
+      "step": 5447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653885611695307e-05,
+      "loss": 0.6366,
+      "step": 5448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653821492690434e-05,
+      "loss": 0.5983,
+      "step": 5449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6537573736855603e-05,
+      "loss": 0.5775,
+      "step": 5450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6536932546806875e-05,
+      "loss": 0.515,
+      "step": 5451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653629135675815e-05,
+      "loss": 0.5865,
+      "step": 5452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653565016670942e-05,
+      "loss": 0.6692,
+      "step": 5453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6535008976660684e-05,
+      "loss": 0.6176,
+      "step": 5454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6534367786611955e-05,
+      "loss": 0.6461,
+      "step": 5455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653372659656323e-05,
+      "loss": 0.5905,
+      "step": 5456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653308540651449e-05,
+      "loss": 0.5898,
+      "step": 5457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6532444216465764e-05,
+      "loss": 0.5898,
+      "step": 5458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653180302641703e-05,
+      "loss": 0.6145,
+      "step": 5459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65311618363683e-05,
+      "loss": 0.554,
+      "step": 5460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653052064631957e-05,
+      "loss": 0.6214,
+      "step": 5461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6529879456270844e-05,
+      "loss": 0.6342,
+      "step": 5462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652923826622211e-05,
+      "loss": 0.647,
+      "step": 5463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652859707617338e-05,
+      "loss": 0.5719,
+      "step": 5464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652795588612465e-05,
+      "loss": 0.59,
+      "step": 5465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652731469607592e-05,
+      "loss": 0.5942,
+      "step": 5466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652667350602719e-05,
+      "loss": 0.7153,
+      "step": 5467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652603231597846e-05,
+      "loss": 0.5711,
+      "step": 5468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652539112592973e-05,
+      "loss": 0.6442,
+      "step": 5469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6524749935881e-05,
+      "loss": 0.5762,
+      "step": 5470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652410874583227e-05,
+      "loss": 0.5912,
+      "step": 5471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6523467555783534e-05,
+      "loss": 0.5215,
+      "step": 5472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6522826365734805e-05,
+      "loss": 0.666,
+      "step": 5473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652218517568608e-05,
+      "loss": 0.5662,
+      "step": 5474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652154398563734e-05,
+      "loss": 0.551,
+      "step": 5475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6520902795588614e-05,
+      "loss": 0.5074,
+      "step": 5476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6520261605539886e-05,
+      "loss": 0.6114,
+      "step": 5477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651962041549116e-05,
+      "loss": 0.625,
+      "step": 5478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651897922544242e-05,
+      "loss": 0.6639,
+      "step": 5479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6518338035393694e-05,
+      "loss": 0.6172,
+      "step": 5480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651769684534496e-05,
+      "loss": 0.4844,
+      "step": 5481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651705565529623e-05,
+      "loss": 0.5828,
+      "step": 5482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65164144652475e-05,
+      "loss": 0.5234,
+      "step": 5483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6515773275198774e-05,
+      "loss": 0.5537,
+      "step": 5484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6515132085150046e-05,
+      "loss": 0.6986,
+      "step": 5485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651449089510131e-05,
+      "loss": 0.5683,
+      "step": 5486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651384970505258e-05,
+      "loss": 0.5353,
+      "step": 5487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651320851500385e-05,
+      "loss": 0.516,
+      "step": 5488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651256732495512e-05,
+      "loss": 0.5547,
+      "step": 5489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6511926134906384e-05,
+      "loss": 0.5918,
+      "step": 5490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6511284944857656e-05,
+      "loss": 0.6052,
+      "step": 5491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.651064375480893e-05,
+      "loss": 0.5407,
+      "step": 5492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65100025647602e-05,
+      "loss": 0.5987,
+      "step": 5493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650936137471147e-05,
+      "loss": 0.5224,
+      "step": 5494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6508720184662736e-05,
+      "loss": 0.5782,
+      "step": 5495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650807899461401e-05,
+      "loss": 0.5489,
+      "step": 5496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650743780456527e-05,
+      "loss": 0.6225,
+      "step": 5497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6506796614516544e-05,
+      "loss": 0.614,
+      "step": 5498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650615542446781e-05,
+      "loss": 0.5313,
+      "step": 5499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650551423441908e-05,
+      "loss": 0.681,
+      "step": 5500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650487304437036e-05,
+      "loss": 0.6095,
+      "step": 5501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6504231854321624e-05,
+      "loss": 0.5847,
+      "step": 5502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6503590664272896e-05,
+      "loss": 0.6103,
+      "step": 5503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650294947422416e-05,
+      "loss": 0.5991,
+      "step": 5504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650230828417543e-05,
+      "loss": 0.5298,
+      "step": 5505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65016670941267e-05,
+      "loss": 0.5801,
+      "step": 5506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650102590407797e-05,
+      "loss": 0.62,
+      "step": 5507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650038471402924e-05,
+      "loss": 0.6038,
+      "step": 5508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649974352398051e-05,
+      "loss": 0.5885,
+      "step": 5509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6499102333931784e-05,
+      "loss": 0.5524,
+      "step": 5510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649846114388305e-05,
+      "loss": 0.5932,
+      "step": 5511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649781995383432e-05,
+      "loss": 0.6138,
+      "step": 5512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6497178763785586e-05,
+      "loss": 0.7358,
+      "step": 5513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649653757373686e-05,
+      "loss": 0.5857,
+      "step": 5514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649589638368812e-05,
+      "loss": 0.567,
+      "step": 5515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6495255193639394e-05,
+      "loss": 0.551,
+      "step": 5516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6494614003590666e-05,
+      "loss": 0.6159,
+      "step": 5517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649397281354194e-05,
+      "loss": 0.5931,
+      "step": 5518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649333162349321e-05,
+      "loss": 0.5431,
+      "step": 5519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6492690433444475e-05,
+      "loss": 0.5148,
+      "step": 5520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6492049243395746e-05,
+      "loss": 0.5871,
+      "step": 5521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649140805334701e-05,
+      "loss": 0.5439,
+      "step": 5522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.649076686329828e-05,
+      "loss": 0.6247,
+      "step": 5523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6490125673249555e-05,
+      "loss": 0.6191,
+      "step": 5524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6489484483200826e-05,
+      "loss": 0.649,
+      "step": 5525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648884329315209e-05,
+      "loss": 0.5219,
+      "step": 5526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648820210310336e-05,
+      "loss": 0.648,
+      "step": 5527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6487560913054635e-05,
+      "loss": 0.6092,
+      "step": 5528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.64869197230059e-05,
+      "loss": 0.6472,
+      "step": 5529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648627853295717e-05,
+      "loss": 0.6038,
+      "step": 5530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6485637342908436e-05,
+      "loss": 0.5968,
+      "step": 5531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648499615285971e-05,
+      "loss": 0.6111,
+      "step": 5532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648435496281098e-05,
+      "loss": 0.6421,
+      "step": 5533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648371377276225e-05,
+      "loss": 0.5402,
+      "step": 5534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6483072582713516e-05,
+      "loss": 0.6076,
+      "step": 5535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648243139266479e-05,
+      "loss": 0.5935,
+      "step": 5536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648179020261606e-05,
+      "loss": 0.6121,
+      "step": 5537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6481149012567325e-05,
+      "loss": 0.565,
+      "step": 5538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6480507822518596e-05,
+      "loss": 0.6663,
+      "step": 5539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647986663246987e-05,
+      "loss": 0.5856,
+      "step": 5540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647922544242114e-05,
+      "loss": 0.5321,
+      "step": 5541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6478584252372405e-05,
+      "loss": 0.6067,
+      "step": 5542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6477943062323677e-05,
+      "loss": 0.6012,
+      "step": 5543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647730187227494e-05,
+      "loss": 0.5555,
+      "step": 5544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647666068222621e-05,
+      "loss": 0.5968,
+      "step": 5545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6476019492177485e-05,
+      "loss": 0.7153,
+      "step": 5546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647537830212875e-05,
+      "loss": 0.6138,
+      "step": 5547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647473711208002e-05,
+      "loss": 0.5491,
+      "step": 5548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647409592203129e-05,
+      "loss": 0.5321,
+      "step": 5549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6473454731982565e-05,
+      "loss": 0.5436,
+      "step": 5550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647281354193383e-05,
+      "loss": 0.5639,
+      "step": 5551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.64721723518851e-05,
+      "loss": 0.7412,
+      "step": 5552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647153116183637e-05,
+      "loss": 0.5087,
+      "step": 5553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647088997178764e-05,
+      "loss": 0.5832,
+      "step": 5554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.647024878173891e-05,
+      "loss": 0.6689,
+      "step": 5555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646960759169018e-05,
+      "loss": 0.596,
+      "step": 5556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6468966401641453e-05,
+      "loss": 0.5114,
+      "step": 5557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646832521159272e-05,
+      "loss": 0.5777,
+      "step": 5558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646768402154399e-05,
+      "loss": 0.5803,
+      "step": 5559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6467042831495255e-05,
+      "loss": 0.644,
+      "step": 5560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646640164144653e-05,
+      "loss": 0.6441,
+      "step": 5561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646576045139779e-05,
+      "loss": 0.5715,
+      "step": 5562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6465119261349063e-05,
+      "loss": 0.5279,
+      "step": 5563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6464478071300335e-05,
+      "loss": 0.5682,
+      "step": 5564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646383688125161e-05,
+      "loss": 0.6644,
+      "step": 5565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646319569120288e-05,
+      "loss": 0.6803,
+      "step": 5566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6462554501154144e-05,
+      "loss": 0.5186,
+      "step": 5567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6461913311105415e-05,
+      "loss": 0.6546,
+      "step": 5568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646127212105668e-05,
+      "loss": 0.5576,
+      "step": 5569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.646063093100795e-05,
+      "loss": 0.582,
+      "step": 5570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645998974095922e-05,
+      "loss": 0.6048,
+      "step": 5571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6459348550910495e-05,
+      "loss": 0.5945,
+      "step": 5572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645870736086177e-05,
+      "loss": 0.6932,
+      "step": 5573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645806617081303e-05,
+      "loss": 0.5361,
+      "step": 5574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6457424980764304e-05,
+      "loss": 0.5175,
+      "step": 5575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645678379071557e-05,
+      "loss": 0.5148,
+      "step": 5576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645614260066684e-05,
+      "loss": 0.6955,
+      "step": 5577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6455501410618105e-05,
+      "loss": 0.5867,
+      "step": 5578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645486022056938e-05,
+      "loss": 0.6531,
+      "step": 5579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645421903052065e-05,
+      "loss": 0.5469,
+      "step": 5580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645357784047192e-05,
+      "loss": 0.5953,
+      "step": 5581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645293665042319e-05,
+      "loss": 0.6051,
+      "step": 5582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645229546037446e-05,
+      "loss": 0.551,
+      "step": 5583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645165427032573e-05,
+      "loss": 0.735,
+      "step": 5584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6451013080276994e-05,
+      "loss": 0.6505,
+      "step": 5585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6450371890228266e-05,
+      "loss": 0.5006,
+      "step": 5586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.644973070017953e-05,
+      "loss": 0.5265,
+      "step": 5587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.64490895101308e-05,
+      "loss": 0.5776,
+      "step": 5588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6448448320082074e-05,
+      "loss": 0.5305,
+      "step": 5589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6447807130033346e-05,
+      "loss": 0.5099,
+      "step": 5590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.644716593998462e-05,
+      "loss": 0.6556,
+      "step": 5591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.644652474993588e-05,
+      "loss": 0.5326,
+      "step": 5592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6445883559887154e-05,
+      "loss": 0.5802,
+      "step": 5593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644524236983842e-05,
+      "loss": 0.6122,
+      "step": 5594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644460117978969e-05,
+      "loss": 0.5594,
+      "step": 5595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644395998974096e-05,
+      "loss": 0.5501,
+      "step": 5596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6443318799692234e-05,
+      "loss": 0.6115,
+      "step": 5597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64426776096435e-05,
+      "loss": 0.6331,
+      "step": 5598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644203641959477e-05,
+      "loss": 0.5185,
+      "step": 5599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644139522954604e-05,
+      "loss": 0.5254,
+      "step": 5600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644075403949731e-05,
+      "loss": 0.6304,
+      "step": 5601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644011284944858e-05,
+      "loss": 0.6405,
+      "step": 5602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6439471659399844e-05,
+      "loss": 0.5492,
+      "step": 5603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6438830469351116e-05,
+      "loss": 0.5974,
+      "step": 5604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643818927930239e-05,
+      "loss": 0.5956,
+      "step": 5605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643754808925366e-05,
+      "loss": 0.5482,
+      "step": 5606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6436906899204924e-05,
+      "loss": 0.5237,
+      "step": 5607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6436265709156196e-05,
+      "loss": 0.7228,
+      "step": 5608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643562451910747e-05,
+      "loss": 0.7752,
+      "step": 5609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643498332905873e-05,
+      "loss": 0.5768,
+      "step": 5610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6434342139010004e-05,
+      "loss": 0.6226,
+      "step": 5611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6433700948961276e-05,
+      "loss": 0.5911,
+      "step": 5612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643305975891255e-05,
+      "loss": 0.5261,
+      "step": 5613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643241856886381e-05,
+      "loss": 0.5392,
+      "step": 5614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6431777378815084e-05,
+      "loss": 0.5181,
+      "step": 5615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643113618876635e-05,
+      "loss": 0.5861,
+      "step": 5616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643049499871762e-05,
+      "loss": 0.5705,
+      "step": 5617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642985380866889e-05,
+      "loss": 0.5978,
+      "step": 5618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642921261862016e-05,
+      "loss": 0.5561,
+      "step": 5619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 0.5816,
+      "step": 5620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64279302385227e-05,
+      "loss": 0.5757,
+      "step": 5621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642728904847397e-05,
+      "loss": 0.6273,
+      "step": 5622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642664785842524e-05,
+      "loss": 0.5953,
+      "step": 5623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642600666837651e-05,
+      "loss": 0.5772,
+      "step": 5624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6425365478327774e-05,
+      "loss": 0.5418,
+      "step": 5625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6424724288279046e-05,
+      "loss": 0.5462,
+      "step": 5626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642408309823032e-05,
+      "loss": 0.5593,
+      "step": 5627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642344190818159e-05,
+      "loss": 0.5794,
+      "step": 5628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642280071813286e-05,
+      "loss": 0.5286,
+      "step": 5629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6422159528084126e-05,
+      "loss": 0.6122,
+      "step": 5630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64215183380354e-05,
+      "loss": 0.5273,
+      "step": 5631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642087714798666e-05,
+      "loss": 0.6219,
+      "step": 5632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6420235957937935e-05,
+      "loss": 0.5335,
+      "step": 5633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6419594767889206e-05,
+      "loss": 0.5406,
+      "step": 5634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641895357784047e-05,
+      "loss": 0.6527,
+      "step": 5635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641831238779174e-05,
+      "loss": 0.5785,
+      "step": 5636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6417671197743015e-05,
+      "loss": 0.6429,
+      "step": 5637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6417030007694286e-05,
+      "loss": 0.5776,
+      "step": 5638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641638881764555e-05,
+      "loss": 0.5597,
+      "step": 5639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641574762759682e-05,
+      "loss": 0.6294,
+      "step": 5640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641510643754809e-05,
+      "loss": 0.6494,
+      "step": 5641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641446524749936e-05,
+      "loss": 0.673,
+      "step": 5642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641382405745063e-05,
+      "loss": 0.5786,
+      "step": 5643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64131828674019e-05,
+      "loss": 0.6436,
+      "step": 5644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6412541677353175e-05,
+      "loss": 0.6521,
+      "step": 5645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641190048730444e-05,
+      "loss": 0.5838,
+      "step": 5646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641125929725571e-05,
+      "loss": 0.6263,
+      "step": 5647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6410618107206976e-05,
+      "loss": 0.5954,
+      "step": 5648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640997691715825e-05,
+      "loss": 0.6075,
+      "step": 5649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640933572710951e-05,
+      "loss": 0.5736,
+      "step": 5650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6408694537060785e-05,
+      "loss": 0.5989,
+      "step": 5651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6408053347012057e-05,
+      "loss": 0.5866,
+      "step": 5652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640741215696333e-05,
+      "loss": 0.6521,
+      "step": 5653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64067709669146e-05,
+      "loss": 0.6085,
+      "step": 5654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6406129776865865e-05,
+      "loss": 0.5742,
+      "step": 5655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6405488586817137e-05,
+      "loss": 0.5399,
+      "step": 5656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64048473967684e-05,
+      "loss": 0.5232,
+      "step": 5657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640420620671967e-05,
+      "loss": 0.5699,
+      "step": 5658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640356501667094e-05,
+      "loss": 0.5641,
+      "step": 5659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640292382662221e-05,
+      "loss": 0.6098,
+      "step": 5660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640228263657349e-05,
+      "loss": 0.6014,
+      "step": 5661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640164144652475e-05,
+      "loss": 0.5692,
+      "step": 5662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6401000256476025e-05,
+      "loss": 0.53,
+      "step": 5663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640035906642729e-05,
+      "loss": 0.5979,
+      "step": 5664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639971787637856e-05,
+      "loss": 0.5783,
+      "step": 5665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639907668632983e-05,
+      "loss": 0.6796,
+      "step": 5666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63984354962811e-05,
+      "loss": 0.5388,
+      "step": 5667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639779430623237e-05,
+      "loss": 0.5698,
+      "step": 5668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639715311618364e-05,
+      "loss": 0.6168,
+      "step": 5669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6396511926134914e-05,
+      "loss": 0.6915,
+      "step": 5670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639587073608618e-05,
+      "loss": 0.6028,
+      "step": 5671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639522954603745e-05,
+      "loss": 0.5695,
+      "step": 5672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6394588355988715e-05,
+      "loss": 0.5594,
+      "step": 5673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639394716593999e-05,
+      "loss": 0.5375,
+      "step": 5674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639330597589125e-05,
+      "loss": 0.5695,
+      "step": 5675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6392664785842524e-05,
+      "loss": 0.5203,
+      "step": 5676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6392023595793795e-05,
+      "loss": 0.7212,
+      "step": 5677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639138240574507e-05,
+      "loss": 0.5858,
+      "step": 5678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639074121569634e-05,
+      "loss": 0.5615,
+      "step": 5679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6390100025647604e-05,
+      "loss": 0.5463,
+      "step": 5680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6389458835598875e-05,
+      "loss": 0.5192,
+      "step": 5681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638881764555014e-05,
+      "loss": 0.584,
+      "step": 5682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638817645550141e-05,
+      "loss": 0.5931,
+      "step": 5683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6387535265452684e-05,
+      "loss": 0.7229,
+      "step": 5684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6386894075403955e-05,
+      "loss": 0.508,
+      "step": 5685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638625288535522e-05,
+      "loss": 0.664,
+      "step": 5686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638561169530649e-05,
+      "loss": 0.53,
+      "step": 5687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6384970505257764e-05,
+      "loss": 0.6235,
+      "step": 5688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638432931520903e-05,
+      "loss": 0.5581,
+      "step": 5689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63836881251603e-05,
+      "loss": 0.5821,
+      "step": 5690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6383046935111565e-05,
+      "loss": 0.6749,
+      "step": 5691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638240574506284e-05,
+      "loss": 0.6073,
+      "step": 5692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638176455501411e-05,
+      "loss": 0.5436,
+      "step": 5693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638112336496538e-05,
+      "loss": 0.614,
+      "step": 5694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6380482174916645e-05,
+      "loss": 0.5721,
+      "step": 5695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637984098486792e-05,
+      "loss": 0.5342,
+      "step": 5696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637919979481919e-05,
+      "loss": 0.5067,
+      "step": 5697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6378558604770454e-05,
+      "loss": 0.6698,
+      "step": 5698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6377917414721726e-05,
+      "loss": 0.5963,
+      "step": 5699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6377276224673e-05,
+      "loss": 0.5558,
+      "step": 5700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637663503462427e-05,
+      "loss": 0.5459,
+      "step": 5701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6375993844575534e-05,
+      "loss": 0.6113,
+      "step": 5702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6375352654526806e-05,
+      "loss": 0.617,
+      "step": 5703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637471146447807e-05,
+      "loss": 0.6173,
+      "step": 5704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637407027442934e-05,
+      "loss": 0.5912,
+      "step": 5705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6373429084380614e-05,
+      "loss": 0.4969,
+      "step": 5706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637278789433188e-05,
+      "loss": 0.5408,
+      "step": 5707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637214670428315e-05,
+      "loss": 0.5964,
+      "step": 5708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637150551423442e-05,
+      "loss": 0.553,
+      "step": 5709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6370864324185694e-05,
+      "loss": 0.5779,
+      "step": 5710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637022313413696e-05,
+      "loss": 0.5708,
+      "step": 5711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636958194408823e-05,
+      "loss": 0.5281,
+      "step": 5712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6368940754039496e-05,
+      "loss": 0.5291,
+      "step": 5713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636829956399077e-05,
+      "loss": 0.5461,
+      "step": 5714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636765837394204e-05,
+      "loss": 0.5757,
+      "step": 5715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636701718389331e-05,
+      "loss": 0.5553,
+      "step": 5716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636637599384458e-05,
+      "loss": 0.5481,
+      "step": 5717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636573480379585e-05,
+      "loss": 0.5963,
+      "step": 5718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636509361374712e-05,
+      "loss": 0.4933,
+      "step": 5719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6364452423698384e-05,
+      "loss": 0.6487,
+      "step": 5720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6363811233649656e-05,
+      "loss": 0.6875,
+      "step": 5721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636317004360092e-05,
+      "loss": 0.5185,
+      "step": 5722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636252885355219e-05,
+      "loss": 0.6193,
+      "step": 5723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6361887663503464e-05,
+      "loss": 0.5585,
+      "step": 5724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6361246473454736e-05,
+      "loss": 0.6047,
+      "step": 5725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636060528340601e-05,
+      "loss": 0.6999,
+      "step": 5726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635996409335727e-05,
+      "loss": 0.5637,
+      "step": 5727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6359322903308544e-05,
+      "loss": 0.5981,
+      "step": 5728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635868171325981e-05,
+      "loss": 0.6999,
+      "step": 5729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635804052321108e-05,
+      "loss": 0.5431,
+      "step": 5730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6357399333162346e-05,
+      "loss": 0.6336,
+      "step": 5731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6356758143113624e-05,
+      "loss": 0.5632,
+      "step": 5732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6356116953064896e-05,
+      "loss": 0.5572,
+      "step": 5733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635547576301616e-05,
+      "loss": 0.5725,
+      "step": 5734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635483457296743e-05,
+      "loss": 0.559,
+      "step": 5735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63541933829187e-05,
+      "loss": 0.6181,
+      "step": 5736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635355219286997e-05,
+      "loss": 0.6611,
+      "step": 5737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6352911002821234e-05,
+      "loss": 0.5649,
+      "step": 5738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6352269812772506e-05,
+      "loss": 0.5417,
+      "step": 5739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635162862272378e-05,
+      "loss": 0.5809,
+      "step": 5740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635098743267505e-05,
+      "loss": 0.5768,
+      "step": 5741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.635034624262632e-05,
+      "loss": 0.4475,
+      "step": 5742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6349705052577586e-05,
+      "loss": 0.5636,
+      "step": 5743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634906386252886e-05,
+      "loss": 0.615,
+      "step": 5744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634842267248012e-05,
+      "loss": 0.4388,
+      "step": 5745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6347781482431395e-05,
+      "loss": 0.5449,
+      "step": 5746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634714029238266e-05,
+      "loss": 0.5511,
+      "step": 5747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634649910233393e-05,
+      "loss": 0.5524,
+      "step": 5748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63458579122852e-05,
+      "loss": 0.5933,
+      "step": 5749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6345216722236475e-05,
+      "loss": 0.5959,
+      "step": 5750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6344575532187746e-05,
+      "loss": 0.5411,
+      "step": 5751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634393434213901e-05,
+      "loss": 0.6129,
+      "step": 5752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634329315209028e-05,
+      "loss": 0.5237,
+      "step": 5753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634265196204155e-05,
+      "loss": 0.5533,
+      "step": 5754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634201077199282e-05,
+      "loss": 0.5495,
+      "step": 5755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634136958194409e-05,
+      "loss": 0.5745,
+      "step": 5756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634072839189536e-05,
+      "loss": 0.5485,
+      "step": 5757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634008720184663e-05,
+      "loss": 0.5561,
+      "step": 5758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63394460117979e-05,
+      "loss": 0.5496,
+      "step": 5759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633880482174917e-05,
+      "loss": 0.5064,
+      "step": 5760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6338163631700436e-05,
+      "loss": 0.6714,
+      "step": 5761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633752244165171e-05,
+      "loss": 0.5971,
+      "step": 5762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633688125160297e-05,
+      "loss": 0.5829,
+      "step": 5763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6336240061554245e-05,
+      "loss": 0.5244,
+      "step": 5764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6335598871505517e-05,
+      "loss": 0.6446,
+      "step": 5765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633495768145679e-05,
+      "loss": 0.6204,
+      "step": 5766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633431649140805e-05,
+      "loss": 0.5605,
+      "step": 5767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6333675301359325e-05,
+      "loss": 0.5209,
+      "step": 5768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63330341113106e-05,
+      "loss": 0.6049,
+      "step": 5769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633239292126186e-05,
+      "loss": 0.5841,
+      "step": 5770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633175173121313e-05,
+      "loss": 0.6319,
+      "step": 5771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6331110541164405e-05,
+      "loss": 0.5427,
+      "step": 5772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633046935111568e-05,
+      "loss": 0.6518,
+      "step": 5773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632982816106694e-05,
+      "loss": 0.6227,
+      "step": 5774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632918697101821e-05,
+      "loss": 0.5731,
+      "step": 5775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632854578096948e-05,
+      "loss": 0.5664,
+      "step": 5776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632790459092075e-05,
+      "loss": 0.6197,
+      "step": 5777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632726340087202e-05,
+      "loss": 0.5783,
+      "step": 5778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632662221082329e-05,
+      "loss": 0.5997,
+      "step": 5779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632598102077456e-05,
+      "loss": 0.4902,
+      "step": 5780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632533983072583e-05,
+      "loss": 0.6295,
+      "step": 5781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63246986406771e-05,
+      "loss": 0.5251,
+      "step": 5782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632405745062837e-05,
+      "loss": 0.5685,
+      "step": 5783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632341626057964e-05,
+      "loss": 0.5813,
+      "step": 5784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6322775070530903e-05,
+      "loss": 0.5529,
+      "step": 5785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6322133880482175e-05,
+      "loss": 0.6171,
+      "step": 5786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632149269043345e-05,
+      "loss": 0.6308,
+      "step": 5787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632085150038472e-05,
+      "loss": 0.5517,
+      "step": 5788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632021031033599e-05,
+      "loss": 0.6332,
+      "step": 5789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6319569120287255e-05,
+      "loss": 0.6257,
+      "step": 5790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631892793023853e-05,
+      "loss": 0.547,
+      "step": 5791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631828674018979e-05,
+      "loss": 0.6428,
+      "step": 5792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6317645550141064e-05,
+      "loss": 0.5865,
+      "step": 5793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631700436009233e-05,
+      "loss": 0.5675,
+      "step": 5794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63163631700436e-05,
+      "loss": 0.5927,
+      "step": 5795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631572197999487e-05,
+      "loss": 0.5974,
+      "step": 5796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6315080789946144e-05,
+      "loss": 0.6075,
+      "step": 5797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6314439599897415e-05,
+      "loss": 0.5486,
+      "step": 5798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631379840984868e-05,
+      "loss": 0.6037,
+      "step": 5799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631315721979995e-05,
+      "loss": 0.597,
+      "step": 5800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631251602975122e-05,
+      "loss": 0.6399,
+      "step": 5801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631187483970249e-05,
+      "loss": 0.6056,
+      "step": 5802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6311233649653754e-05,
+      "loss": 0.6122,
+      "step": 5803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631059245960503e-05,
+      "loss": 0.6072,
+      "step": 5804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6309951269556304e-05,
+      "loss": 0.5792,
+      "step": 5805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630931007950757e-05,
+      "loss": 0.6876,
+      "step": 5806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630866888945884e-05,
+      "loss": 0.53,
+      "step": 5807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6308027699410105e-05,
+      "loss": 0.5565,
+      "step": 5808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630738650936138e-05,
+      "loss": 0.56,
+      "step": 5809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630674531931264e-05,
+      "loss": 0.6542,
+      "step": 5810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6306104129263914e-05,
+      "loss": 0.5303,
+      "step": 5811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6305462939215186e-05,
+      "loss": 0.6102,
+      "step": 5812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630482174916646e-05,
+      "loss": 0.4873,
+      "step": 5813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630418055911773e-05,
+      "loss": 0.6638,
+      "step": 5814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6303539369068994e-05,
+      "loss": 0.5966,
+      "step": 5815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6302898179020266e-05,
+      "loss": 0.585,
+      "step": 5816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630225698897153e-05,
+      "loss": 0.5656,
+      "step": 5817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63016157989228e-05,
+      "loss": 0.6232,
+      "step": 5818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630097460887407e-05,
+      "loss": 0.5896,
+      "step": 5819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.630033341882534e-05,
+      "loss": 0.5808,
+      "step": 5820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629969222877661e-05,
+      "loss": 0.5159,
+      "step": 5821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629905103872788e-05,
+      "loss": 0.6708,
+      "step": 5822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6298409848679154e-05,
+      "loss": 0.4733,
+      "step": 5823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629776865863042e-05,
+      "loss": 0.5394,
+      "step": 5824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629712746858169e-05,
+      "loss": 0.6082,
+      "step": 5825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6296486278532956e-05,
+      "loss": 0.4752,
+      "step": 5826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629584508848423e-05,
+      "loss": 0.5838,
+      "step": 5827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.62952038984355e-05,
+      "loss": 0.5687,
+      "step": 5828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629456270838677e-05,
+      "loss": 0.5728,
+      "step": 5829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629392151833804e-05,
+      "loss": 0.4755,
+      "step": 5830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629328032828931e-05,
+      "loss": 0.4989,
+      "step": 5831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629263913824058e-05,
+      "loss": 0.6057,
+      "step": 5832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6291997948191844e-05,
+      "loss": 0.5516,
+      "step": 5833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6291356758143116e-05,
+      "loss": 0.6755,
+      "step": 5834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629071556809438e-05,
+      "loss": 0.6486,
+      "step": 5835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629007437804565e-05,
+      "loss": 0.5444,
+      "step": 5836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6289433187996924e-05,
+      "loss": 0.6927,
+      "step": 5837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6288791997948196e-05,
+      "loss": 0.5554,
+      "step": 5838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628815080789947e-05,
+      "loss": 0.5573,
+      "step": 5839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628750961785073e-05,
+      "loss": 0.7242,
+      "step": 5840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6286868427802004e-05,
+      "loss": 0.6407,
+      "step": 5841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628622723775327e-05,
+      "loss": 0.4915,
+      "step": 5842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628558604770454e-05,
+      "loss": 0.6218,
+      "step": 5843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628494485765581e-05,
+      "loss": 0.5854,
+      "step": 5844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6284303667607084e-05,
+      "loss": 0.7167,
+      "step": 5845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628366247755835e-05,
+      "loss": 0.5403,
+      "step": 5846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628302128750962e-05,
+      "loss": 0.6191,
+      "step": 5847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628238009746089e-05,
+      "loss": 0.6486,
+      "step": 5848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628173890741216e-05,
+      "loss": 0.6275,
+      "step": 5849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.628109771736343e-05,
+      "loss": 0.5982,
+      "step": 5850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6280456527314694e-05,
+      "loss": 0.6111,
+      "step": 5851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6279815337265966e-05,
+      "loss": 0.5598,
+      "step": 5852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627917414721724e-05,
+      "loss": 0.5768,
+      "step": 5853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627853295716851e-05,
+      "loss": 0.524,
+      "step": 5854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6277891767119775e-05,
+      "loss": 0.5483,
+      "step": 5855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6277250577071046e-05,
+      "loss": 0.5656,
+      "step": 5856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627660938702232e-05,
+      "loss": 0.5493,
+      "step": 5857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627596819697358e-05,
+      "loss": 0.5621,
+      "step": 5858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6275327006924855e-05,
+      "loss": 0.4702,
+      "step": 5859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6274685816876126e-05,
+      "loss": 0.5802,
+      "step": 5860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62740446268274e-05,
+      "loss": 0.5735,
+      "step": 5861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627340343677866e-05,
+      "loss": 0.6663,
+      "step": 5862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6272762246729935e-05,
+      "loss": 0.5396,
+      "step": 5863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62721210566812e-05,
+      "loss": 0.5909,
+      "step": 5864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627147986663247e-05,
+      "loss": 0.6109,
+      "step": 5865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627083867658374e-05,
+      "loss": 0.4909,
+      "step": 5866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627019748653501e-05,
+      "loss": 0.6462,
+      "step": 5867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626955629648628e-05,
+      "loss": 0.6522,
+      "step": 5868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626891510643755e-05,
+      "loss": 0.6736,
+      "step": 5869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626827391638882e-05,
+      "loss": 0.5949,
+      "step": 5870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626763272634009e-05,
+      "loss": 0.6469,
+      "step": 5871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626699153629136e-05,
+      "loss": 0.5804,
+      "step": 5872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6266350346242625e-05,
+      "loss": 0.6585,
+      "step": 5873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6265709156193896e-05,
+      "loss": 0.6039,
+      "step": 5874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626506796614517e-05,
+      "loss": 0.586,
+      "step": 5875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626442677609644e-05,
+      "loss": 0.5606,
+      "step": 5876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626378558604771e-05,
+      "loss": 0.5511,
+      "step": 5877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6263144395998977e-05,
+      "loss": 0.6224,
+      "step": 5878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626250320595025e-05,
+      "loss": 0.5954,
+      "step": 5879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626186201590151e-05,
+      "loss": 0.6125,
+      "step": 5880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6261220825852785e-05,
+      "loss": 0.5921,
+      "step": 5881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626057963580405e-05,
+      "loss": 0.5984,
+      "step": 5882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625993844575532e-05,
+      "loss": 0.5462,
+      "step": 5883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625929725570659e-05,
+      "loss": 0.5477,
+      "step": 5884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6258656065657865e-05,
+      "loss": 0.6276,
+      "step": 5885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625801487560914e-05,
+      "loss": 0.5431,
+      "step": 5886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62573736855604e-05,
+      "loss": 0.6324,
+      "step": 5887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6256732495511673e-05,
+      "loss": 0.4963,
+      "step": 5888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625609130546294e-05,
+      "loss": 0.5515,
+      "step": 5889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625545011541421e-05,
+      "loss": 0.5586,
+      "step": 5890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6254808925365475e-05,
+      "loss": 0.5662,
+      "step": 5891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6254167735316754e-05,
+      "loss": 0.6435,
+      "step": 5892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6253526545268025e-05,
+      "loss": 0.5003,
+      "step": 5893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625288535521929e-05,
+      "loss": 0.5494,
+      "step": 5894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625224416517056e-05,
+      "loss": 0.5056,
+      "step": 5895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625160297512183e-05,
+      "loss": 0.6057,
+      "step": 5896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62509617850731e-05,
+      "loss": 0.5023,
+      "step": 5897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6250320595024363e-05,
+      "loss": 0.6655,
+      "step": 5898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6249679404975635e-05,
+      "loss": 0.6478,
+      "step": 5899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624903821492691e-05,
+      "loss": 0.5099,
+      "step": 5900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624839702487818e-05,
+      "loss": 0.6,
+      "step": 5901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624775583482945e-05,
+      "loss": 0.522,
+      "step": 5902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6247114644780715e-05,
+      "loss": 0.5412,
+      "step": 5903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624647345473199e-05,
+      "loss": 0.4974,
+      "step": 5904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624583226468325e-05,
+      "loss": 0.572,
+      "step": 5905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6245191074634524e-05,
+      "loss": 0.6212,
+      "step": 5906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624454988458579e-05,
+      "loss": 0.6175,
+      "step": 5907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624390869453706e-05,
+      "loss": 0.6581,
+      "step": 5908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624326750448833e-05,
+      "loss": 0.5977,
+      "step": 5909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6242626314439604e-05,
+      "loss": 0.4987,
+      "step": 5910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6241985124390875e-05,
+      "loss": 0.5882,
+      "step": 5911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624134393434214e-05,
+      "loss": 0.6158,
+      "step": 5912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624070274429341e-05,
+      "loss": 0.6415,
+      "step": 5913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624006155424468e-05,
+      "loss": 0.4937,
+      "step": 5914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623942036419595e-05,
+      "loss": 0.495,
+      "step": 5915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623877917414722e-05,
+      "loss": 0.5685,
+      "step": 5916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623813798409849e-05,
+      "loss": 0.4408,
+      "step": 5917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623749679404976e-05,
+      "loss": 0.5566,
+      "step": 5918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623685560400103e-05,
+      "loss": 0.6394,
+      "step": 5919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62362144139523e-05,
+      "loss": 0.6567,
+      "step": 5920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6235573223903566e-05,
+      "loss": 0.5668,
+      "step": 5921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623493203385484e-05,
+      "loss": 0.4959,
+      "step": 5922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62342908438061e-05,
+      "loss": 0.6902,
+      "step": 5923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6233649653757374e-05,
+      "loss": 0.5148,
+      "step": 5924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6233008463708646e-05,
+      "loss": 0.578,
+      "step": 5925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623236727365992e-05,
+      "loss": 0.6326,
+      "step": 5926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623172608361118e-05,
+      "loss": 0.5758,
+      "step": 5927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6231084893562454e-05,
+      "loss": 0.6133,
+      "step": 5928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6230443703513726e-05,
+      "loss": 0.6324,
+      "step": 5929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622980251346499e-05,
+      "loss": 0.5162,
+      "step": 5930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622916132341626e-05,
+      "loss": 0.6086,
+      "step": 5931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6228520133367534e-05,
+      "loss": 0.7215,
+      "step": 5932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6227878943318806e-05,
+      "loss": 0.6307,
+      "step": 5933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622723775327007e-05,
+      "loss": 0.6129,
+      "step": 5934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622659656322134e-05,
+      "loss": 0.6414,
+      "step": 5935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622595537317261e-05,
+      "loss": 0.5673,
+      "step": 5936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622531418312388e-05,
+      "loss": 0.5046,
+      "step": 5937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622467299307515e-05,
+      "loss": 0.5373,
+      "step": 5938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6224031803026416e-05,
+      "loss": 0.5754,
+      "step": 5939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622339061297769e-05,
+      "loss": 0.5567,
+      "step": 5940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622274942292896e-05,
+      "loss": 0.6566,
+      "step": 5941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622210823288023e-05,
+      "loss": 0.5699,
+      "step": 5942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6221467042831496e-05,
+      "loss": 0.5069,
+      "step": 5943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622082585278277e-05,
+      "loss": 0.6328,
+      "step": 5944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622018466273403e-05,
+      "loss": 0.5732,
+      "step": 5945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6219543472685304e-05,
+      "loss": 0.5305,
+      "step": 5946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6218902282636576e-05,
+      "loss": 0.607,
+      "step": 5947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621826109258785e-05,
+      "loss": 0.6067,
+      "step": 5948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621761990253912e-05,
+      "loss": 0.6104,
+      "step": 5949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6216978712490384e-05,
+      "loss": 0.6207,
+      "step": 5950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6216337522441656e-05,
+      "loss": 0.6952,
+      "step": 5951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621569633239292e-05,
+      "loss": 0.6317,
+      "step": 5952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621505514234419e-05,
+      "loss": 0.6649,
+      "step": 5953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621441395229546e-05,
+      "loss": 0.6678,
+      "step": 5954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621377276224673e-05,
+      "loss": 0.5548,
+      "step": 5955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6213131572198e-05,
+      "loss": 0.5464,
+      "step": 5956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621249038214927e-05,
+      "loss": 0.5816,
+      "step": 5957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6211849192100545e-05,
+      "loss": 0.54,
+      "step": 5958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621120800205181e-05,
+      "loss": 0.6094,
+      "step": 5959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621056681200308e-05,
+      "loss": 0.633,
+      "step": 5960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6209925621954346e-05,
+      "loss": 0.6383,
+      "step": 5961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620928443190562e-05,
+      "loss": 0.6573,
+      "step": 5962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620864324185688e-05,
+      "loss": 0.597,
+      "step": 5963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620800205180816e-05,
+      "loss": 0.6484,
+      "step": 5964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620736086175943e-05,
+      "loss": 0.6276,
+      "step": 5965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62067196717107e-05,
+      "loss": 0.6105,
+      "step": 5966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620607848166197e-05,
+      "loss": 0.5778,
+      "step": 5967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6205437291613235e-05,
+      "loss": 0.6225,
+      "step": 5968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6204796101564506e-05,
+      "loss": 0.5402,
+      "step": 5969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620415491151577e-05,
+      "loss": 0.6649,
+      "step": 5970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620351372146704e-05,
+      "loss": 0.5124,
+      "step": 5971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6202872531418315e-05,
+      "loss": 0.5504,
+      "step": 5972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6202231341369586e-05,
+      "loss": 0.6115,
+      "step": 5973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620159015132086e-05,
+      "loss": 0.5358,
+      "step": 5974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620094896127212e-05,
+      "loss": 0.579,
+      "step": 5975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6200307771223395e-05,
+      "loss": 0.6116,
+      "step": 5976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619966658117466e-05,
+      "loss": 0.5972,
+      "step": 5977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619902539112593e-05,
+      "loss": 0.5908,
+      "step": 5978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6198384201077196e-05,
+      "loss": 0.6189,
+      "step": 5979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6197743011028475e-05,
+      "loss": 0.6632,
+      "step": 5980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619710182097974e-05,
+      "loss": 0.6326,
+      "step": 5981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619646063093101e-05,
+      "loss": 0.5087,
+      "step": 5982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619581944088228e-05,
+      "loss": 0.519,
+      "step": 5983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619517825083355e-05,
+      "loss": 0.4974,
+      "step": 5984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619453706078482e-05,
+      "loss": 0.5338,
+      "step": 5985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6193895870736085e-05,
+      "loss": 0.5305,
+      "step": 5986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6193254680687357e-05,
+      "loss": 0.5598,
+      "step": 5987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619261349063863e-05,
+      "loss": 0.5538,
+      "step": 5988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61919723005899e-05,
+      "loss": 0.5909,
+      "step": 5989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6191331110541165e-05,
+      "loss": 0.7141,
+      "step": 5990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619068992049244e-05,
+      "loss": 0.5987,
+      "step": 5991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619004873044371e-05,
+      "loss": 0.5889,
+      "step": 5992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618940754039497e-05,
+      "loss": 0.6761,
+      "step": 5993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6188766350346245e-05,
+      "loss": 0.5909,
+      "step": 5994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618812516029751e-05,
+      "loss": 0.6906,
+      "step": 5995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618748397024878e-05,
+      "loss": 0.5841,
+      "step": 5996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618684278020005e-05,
+      "loss": 0.5556,
+      "step": 5997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6186201590151325e-05,
+      "loss": 0.6706,
+      "step": 5998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618556040010259e-05,
+      "loss": 0.5151,
+      "step": 5999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618491921005386e-05,
+      "loss": 0.6174,
+      "step": 6000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6184278020005133e-05,
+      "loss": 0.5677,
+      "step": 6001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61836368299564e-05,
+      "loss": 0.6347,
+      "step": 6002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618299563990767e-05,
+      "loss": 0.5541,
+      "step": 6003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618235444985894e-05,
+      "loss": 0.5345,
+      "step": 6004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6181713259810214e-05,
+      "loss": 0.5893,
+      "step": 6005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618107206976148e-05,
+      "loss": 0.6355,
+      "step": 6006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618043087971275e-05,
+      "loss": 0.5532,
+      "step": 6007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6179789689664015e-05,
+      "loss": 0.6266,
+      "step": 6008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617914849961529e-05,
+      "loss": 0.6379,
+      "step": 6009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617850730956656e-05,
+      "loss": 0.621,
+      "step": 6010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6177866119517824e-05,
+      "loss": 0.5495,
+      "step": 6011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6177224929469095e-05,
+      "loss": 0.6342,
+      "step": 6012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617658373942037e-05,
+      "loss": 0.6084,
+      "step": 6013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617594254937164e-05,
+      "loss": 0.5716,
+      "step": 6014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6175301359322904e-05,
+      "loss": 0.6521,
+      "step": 6015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6174660169274175e-05,
+      "loss": 0.6254,
+      "step": 6016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617401897922545e-05,
+      "loss": 0.5272,
+      "step": 6017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617337778917671e-05,
+      "loss": 0.5703,
+      "step": 6018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6172736599127984e-05,
+      "loss": 0.5562,
+      "step": 6019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6172095409079255e-05,
+      "loss": 0.5542,
+      "step": 6020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617145421903053e-05,
+      "loss": 0.5391,
+      "step": 6021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.617081302898179e-05,
+      "loss": 0.5786,
+      "step": 6022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6170171838933064e-05,
+      "loss": 0.5329,
+      "step": 6023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616953064888433e-05,
+      "loss": 0.5426,
+      "step": 6024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61688894588356e-05,
+      "loss": 0.5155,
+      "step": 6025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616824826878687e-05,
+      "loss": 0.5329,
+      "step": 6026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616760707873814e-05,
+      "loss": 0.5004,
+      "step": 6027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616696588868941e-05,
+      "loss": 0.5695,
+      "step": 6028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616632469864068e-05,
+      "loss": 0.6446,
+      "step": 6029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616568350859195e-05,
+      "loss": 0.5484,
+      "step": 6030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616504231854322e-05,
+      "loss": 0.5256,
+      "step": 6031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616440112849449e-05,
+      "loss": 0.557,
+      "step": 6032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6163759938445754e-05,
+      "loss": 0.6909,
+      "step": 6033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6163118748397026e-05,
+      "loss": 0.6541,
+      "step": 6034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61624775583483e-05,
+      "loss": 0.6766,
+      "step": 6035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616183636829957e-05,
+      "loss": 0.5117,
+      "step": 6036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.616119517825084e-05,
+      "loss": 0.5983,
+      "step": 6037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6160553988202106e-05,
+      "loss": 0.5281,
+      "step": 6038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615991279815338e-05,
+      "loss": 0.5756,
+      "step": 6039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615927160810464e-05,
+      "loss": 0.5518,
+      "step": 6040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6158630418055914e-05,
+      "loss": 0.5817,
+      "step": 6041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615798922800718e-05,
+      "loss": 0.591,
+      "step": 6042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615734803795845e-05,
+      "loss": 0.7059,
+      "step": 6043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615670684790972e-05,
+      "loss": 0.6569,
+      "step": 6044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6156065657860994e-05,
+      "loss": 0.64,
+      "step": 6045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6155424467812266e-05,
+      "loss": 0.5862,
+      "step": 6046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615478327776353e-05,
+      "loss": 0.5562,
+      "step": 6047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61541420877148e-05,
+      "loss": 0.5319,
+      "step": 6048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615350089766607e-05,
+      "loss": 0.7351,
+      "step": 6049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615285970761734e-05,
+      "loss": 0.5951,
+      "step": 6050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6152218517568604e-05,
+      "loss": 0.6389,
+      "step": 6051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615157732751988e-05,
+      "loss": 0.5636,
+      "step": 6052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6150936137471154e-05,
+      "loss": 0.6227,
+      "step": 6053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615029494742242e-05,
+      "loss": 0.5062,
+      "step": 6054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614965375737369e-05,
+      "loss": 0.6332,
+      "step": 6055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6149012567324956e-05,
+      "loss": 0.6113,
+      "step": 6056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614837137727623e-05,
+      "loss": 0.6059,
+      "step": 6057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614773018722749e-05,
+      "loss": 0.5658,
+      "step": 6058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6147088997178764e-05,
+      "loss": 0.5739,
+      "step": 6059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6146447807130036e-05,
+      "loss": 0.5191,
+      "step": 6060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614580661708131e-05,
+      "loss": 0.6188,
+      "step": 6061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614516542703258e-05,
+      "loss": 0.5981,
+      "step": 6062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6144524236983844e-05,
+      "loss": 0.6429,
+      "step": 6063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6143883046935116e-05,
+      "loss": 0.5869,
+      "step": 6064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614324185688638e-05,
+      "loss": 0.5157,
+      "step": 6065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614260066683765e-05,
+      "loss": 0.6358,
+      "step": 6066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614195947678892e-05,
+      "loss": 0.6167,
+      "step": 6067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614131828674019e-05,
+      "loss": 0.6407,
+      "step": 6068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614067709669146e-05,
+      "loss": 0.585,
+      "step": 6069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.614003590664273e-05,
+      "loss": 0.5906,
+      "step": 6070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6139394716594005e-05,
+      "loss": 0.4804,
+      "step": 6071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.613875352654527e-05,
+      "loss": 0.522,
+      "step": 6072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.613811233649654e-05,
+      "loss": 0.6403,
+      "step": 6073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6137471146447806e-05,
+      "loss": 0.5789,
+      "step": 6074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.613682995639908e-05,
+      "loss": 0.477,
+      "step": 6075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.613618876635035e-05,
+      "loss": 0.6475,
+      "step": 6076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.613554757630162e-05,
+      "loss": 0.5668,
+      "step": 6077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6134906386252886e-05,
+      "loss": 0.57,
+      "step": 6078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.613426519620416e-05,
+      "loss": 0.6119,
+      "step": 6079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.613362400615543e-05,
+      "loss": 0.4616,
+      "step": 6080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6132982816106695e-05,
+      "loss": 0.6168,
+      "step": 6081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6132341626057966e-05,
+      "loss": 0.5913,
+      "step": 6082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.613170043600923e-05,
+      "loss": 0.5433,
+      "step": 6083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61310592459605e-05,
+      "loss": 0.6161,
+      "step": 6084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6130418055911775e-05,
+      "loss": 0.5519,
+      "step": 6085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6129776865863046e-05,
+      "loss": 0.549,
+      "step": 6086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612913567581431e-05,
+      "loss": 0.6524,
+      "step": 6087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612849448576558e-05,
+      "loss": 0.5937,
+      "step": 6088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6127853295716855e-05,
+      "loss": 0.4922,
+      "step": 6089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612721210566812e-05,
+      "loss": 0.6147,
+      "step": 6090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612657091561939e-05,
+      "loss": 0.6816,
+      "step": 6091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612592972557066e-05,
+      "loss": 0.5596,
+      "step": 6092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6125288535521935e-05,
+      "loss": 0.5087,
+      "step": 6093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61246473454732e-05,
+      "loss": 0.6579,
+      "step": 6094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612400615542447e-05,
+      "loss": 0.5068,
+      "step": 6095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6123364965375736e-05,
+      "loss": 0.547,
+      "step": 6096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612272377532701e-05,
+      "loss": 0.5608,
+      "step": 6097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612208258527828e-05,
+      "loss": 0.5428,
+      "step": 6098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6121441395229545e-05,
+      "loss": 0.6018,
+      "step": 6099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6120800205180817e-05,
+      "loss": 0.5365,
+      "step": 6100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612015901513209e-05,
+      "loss": 0.5825,
+      "step": 6101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.611951782508336e-05,
+      "loss": 0.5596,
+      "step": 6102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6118876635034625e-05,
+      "loss": 0.6014,
+      "step": 6103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61182354449859e-05,
+      "loss": 0.6752,
+      "step": 6104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.611759425493716e-05,
+      "loss": 0.6246,
+      "step": 6105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.611695306488843e-05,
+      "loss": 0.58,
+      "step": 6106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6116311874839705e-05,
+      "loss": 0.6046,
+      "step": 6107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.611567068479098e-05,
+      "loss": 0.5292,
+      "step": 6108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.611502949474225e-05,
+      "loss": 0.5402,
+      "step": 6109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6114388304693513e-05,
+      "loss": 0.6048,
+      "step": 6110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6113747114644785e-05,
+      "loss": 0.6445,
+      "step": 6111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.611310592459605e-05,
+      "loss": 0.4978,
+      "step": 6112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.611246473454732e-05,
+      "loss": 0.6019,
+      "step": 6113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.611182354449859e-05,
+      "loss": 0.6249,
+      "step": 6114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.611118235444986e-05,
+      "loss": 0.5328,
+      "step": 6115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.611054116440113e-05,
+      "loss": 0.5431,
+      "step": 6116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61098999743524e-05,
+      "loss": 0.6271,
+      "step": 6117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6109258784303674e-05,
+      "loss": 0.6261,
+      "step": 6118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610861759425494e-05,
+      "loss": 0.6652,
+      "step": 6119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610797640420621e-05,
+      "loss": 0.6685,
+      "step": 6120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6107335214157475e-05,
+      "loss": 0.5423,
+      "step": 6121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610669402410875e-05,
+      "loss": 0.6413,
+      "step": 6122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610605283406001e-05,
+      "loss": 0.5567,
+      "step": 6123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610541164401129e-05,
+      "loss": 0.6245,
+      "step": 6124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610477045396256e-05,
+      "loss": 0.4717,
+      "step": 6125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610412926391383e-05,
+      "loss": 0.5756,
+      "step": 6126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61034880738651e-05,
+      "loss": 0.5149,
+      "step": 6127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6102846883816364e-05,
+      "loss": 0.5734,
+      "step": 6128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6102205693767635e-05,
+      "loss": 0.5934,
+      "step": 6129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61015645037189e-05,
+      "loss": 0.6365,
+      "step": 6130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610092331367017e-05,
+      "loss": 0.6593,
+      "step": 6131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6100282123621444e-05,
+      "loss": 0.582,
+      "step": 6132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6099640933572715e-05,
+      "loss": 0.6276,
+      "step": 6133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609899974352399e-05,
+      "loss": 0.603,
+      "step": 6134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609835855347525e-05,
+      "loss": 0.5733,
+      "step": 6135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6097717363426524e-05,
+      "loss": 0.566,
+      "step": 6136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609707617337779e-05,
+      "loss": 0.5222,
+      "step": 6137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609643498332906e-05,
+      "loss": 0.5549,
+      "step": 6138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6095793793280325e-05,
+      "loss": 0.6373,
+      "step": 6139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6095152603231604e-05,
+      "loss": 0.5414,
+      "step": 6140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609451141318287e-05,
+      "loss": 0.6174,
+      "step": 6141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609387022313414e-05,
+      "loss": 0.578,
+      "step": 6142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609322903308541e-05,
+      "loss": 0.6216,
+      "step": 6143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609258784303668e-05,
+      "loss": 0.5332,
+      "step": 6144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609194665298795e-05,
+      "loss": 0.581,
+      "step": 6145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6091305462939214e-05,
+      "loss": 0.591,
+      "step": 6146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6090664272890486e-05,
+      "loss": 0.5505,
+      "step": 6147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609002308284176e-05,
+      "loss": 0.5069,
+      "step": 6148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608938189279303e-05,
+      "loss": 0.6147,
+      "step": 6149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6088740702744294e-05,
+      "loss": 0.606,
+      "step": 6150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6088099512695566e-05,
+      "loss": 0.4822,
+      "step": 6151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608745832264684e-05,
+      "loss": 0.575,
+      "step": 6152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60868171325981e-05,
+      "loss": 0.5584,
+      "step": 6153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6086175942549374e-05,
+      "loss": 0.6322,
+      "step": 6154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608553475250064e-05,
+      "loss": 0.4814,
+      "step": 6155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608489356245191e-05,
+      "loss": 0.6738,
+      "step": 6156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608425237240318e-05,
+      "loss": 0.6168,
+      "step": 6157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6083611182354454e-05,
+      "loss": 0.6255,
+      "step": 6158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608296999230572e-05,
+      "loss": 0.5121,
+      "step": 6159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608232880225699e-05,
+      "loss": 0.5655,
+      "step": 6160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608168761220826e-05,
+      "loss": 0.6404,
+      "step": 6161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608104642215953e-05,
+      "loss": 0.5396,
+      "step": 6162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60804052321108e-05,
+      "loss": 0.626,
+      "step": 6163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607976404206207e-05,
+      "loss": 0.6246,
+      "step": 6164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607912285201334e-05,
+      "loss": 0.6368,
+      "step": 6165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607848166196461e-05,
+      "loss": 0.5394,
+      "step": 6166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607784047191588e-05,
+      "loss": 0.5953,
+      "step": 6167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6077199281867144e-05,
+      "loss": 0.6012,
+      "step": 6168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6076558091818416e-05,
+      "loss": 0.5205,
+      "step": 6169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607591690176969e-05,
+      "loss": 0.4994,
+      "step": 6170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607527571172095e-05,
+      "loss": 0.6054,
+      "step": 6171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6074634521672224e-05,
+      "loss": 0.6038,
+      "step": 6172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6073993331623496e-05,
+      "loss": 0.7019,
+      "step": 6173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607335214157477e-05,
+      "loss": 0.6601,
+      "step": 6174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607271095152603e-05,
+      "loss": 0.5021,
+      "step": 6175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6072069761477304e-05,
+      "loss": 0.5935,
+      "step": 6176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607142857142857e-05,
+      "loss": 0.4836,
+      "step": 6177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607078738137984e-05,
+      "loss": 0.6146,
+      "step": 6178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607014619133111e-05,
+      "loss": 0.6164,
+      "step": 6179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6069505001282384e-05,
+      "loss": 0.5735,
+      "step": 6180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6068863811233656e-05,
+      "loss": 0.5589,
+      "step": 6181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606822262118492e-05,
+      "loss": 0.7249,
+      "step": 6182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606758143113619e-05,
+      "loss": 0.6363,
+      "step": 6183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606694024108746e-05,
+      "loss": 0.5387,
+      "step": 6184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606629905103873e-05,
+      "loss": 0.5789,
+      "step": 6185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6065657860989994e-05,
+      "loss": 0.5432,
+      "step": 6186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6065016670941266e-05,
+      "loss": 0.5301,
+      "step": 6187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606437548089254e-05,
+      "loss": 0.5345,
+      "step": 6188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606373429084381e-05,
+      "loss": 0.6192,
+      "step": 6189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606309310079508e-05,
+      "loss": 0.5586,
+      "step": 6190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6062451910746346e-05,
+      "loss": 0.5051,
+      "step": 6191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606181072069762e-05,
+      "loss": 0.5642,
+      "step": 6192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606116953064888e-05,
+      "loss": 0.5772,
+      "step": 6193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6060528340600155e-05,
+      "loss": 0.5757,
+      "step": 6194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605988715055142e-05,
+      "loss": 0.5657,
+      "step": 6195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60592459605027e-05,
+      "loss": 0.7057,
+      "step": 6196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605860477045397e-05,
+      "loss": 0.5289,
+      "step": 6197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6057963580405235e-05,
+      "loss": 0.4925,
+      "step": 6198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6057322390356506e-05,
+      "loss": 0.4999,
+      "step": 6199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605668120030777e-05,
+      "loss": 0.6793,
+      "step": 6200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605604001025904e-05,
+      "loss": 0.5546,
+      "step": 6201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605539882021031e-05,
+      "loss": 0.5522,
+      "step": 6202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605475763016158e-05,
+      "loss": 0.5371,
+      "step": 6203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605411644011285e-05,
+      "loss": 0.5577,
+      "step": 6204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605347525006412e-05,
+      "loss": 0.591,
+      "step": 6205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6052834060015395e-05,
+      "loss": 0.5749,
+      "step": 6206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605219286996666e-05,
+      "loss": 0.6605,
+      "step": 6207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605155167991793e-05,
+      "loss": 0.6135,
+      "step": 6208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6050910489869196e-05,
+      "loss": 0.5394,
+      "step": 6209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605026929982047e-05,
+      "loss": 0.5066,
+      "step": 6210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604962810977173e-05,
+      "loss": 0.6791,
+      "step": 6211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604898691972301e-05,
+      "loss": 0.6146,
+      "step": 6212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604834572967428e-05,
+      "loss": 0.5194,
+      "step": 6213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604770453962555e-05,
+      "loss": 0.5181,
+      "step": 6214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604706334957682e-05,
+      "loss": 0.5287,
+      "step": 6215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6046422159528085e-05,
+      "loss": 0.5754,
+      "step": 6216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604578096947936e-05,
+      "loss": 0.6055,
+      "step": 6217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604513977943062e-05,
+      "loss": 0.5094,
+      "step": 6218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604449858938189e-05,
+      "loss": 0.6622,
+      "step": 6219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6043857399333165e-05,
+      "loss": 0.5579,
+      "step": 6220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604321620928444e-05,
+      "loss": 0.6123,
+      "step": 6221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604257501923571e-05,
+      "loss": 0.534,
+      "step": 6222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6041933829186973e-05,
+      "loss": 0.6074,
+      "step": 6223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6041292639138245e-05,
+      "loss": 0.67,
+      "step": 6224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604065144908951e-05,
+      "loss": 0.5935,
+      "step": 6225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604001025904078e-05,
+      "loss": 0.5572,
+      "step": 6226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603936906899205e-05,
+      "loss": 0.653,
+      "step": 6227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603872787894332e-05,
+      "loss": 0.5891,
+      "step": 6228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603808668889459e-05,
+      "loss": 0.5647,
+      "step": 6229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603744549884586e-05,
+      "loss": 0.5638,
+      "step": 6230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6036804308797134e-05,
+      "loss": 0.5488,
+      "step": 6231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60361631187484e-05,
+      "loss": 0.6288,
+      "step": 6232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603552192869967e-05,
+      "loss": 0.5457,
+      "step": 6233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6034880738650935e-05,
+      "loss": 0.5186,
+      "step": 6234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603423954860221e-05,
+      "loss": 0.5557,
+      "step": 6235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603359835855348e-05,
+      "loss": 0.6138,
+      "step": 6236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603295716850475e-05,
+      "loss": 0.638,
+      "step": 6237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6032315978456015e-05,
+      "loss": 0.6018,
+      "step": 6238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603167478840729e-05,
+      "loss": 0.704,
+      "step": 6239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603103359835856e-05,
+      "loss": 0.6363,
+      "step": 6240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6030392408309824e-05,
+      "loss": 0.6208,
+      "step": 6241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6029751218261095e-05,
+      "loss": 0.54,
+      "step": 6242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602911002821236e-05,
+      "loss": 0.792,
+      "step": 6243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602846883816363e-05,
+      "loss": 0.5381,
+      "step": 6244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6027827648114904e-05,
+      "loss": 0.5377,
+      "step": 6245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6027186458066175e-05,
+      "loss": 0.6523,
+      "step": 6246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602654526801744e-05,
+      "loss": 0.6986,
+      "step": 6247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602590407796871e-05,
+      "loss": 0.5532,
+      "step": 6248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6025262887919984e-05,
+      "loss": 0.5568,
+      "step": 6249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602462169787125e-05,
+      "loss": 0.5819,
+      "step": 6250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602398050782252e-05,
+      "loss": 0.6041,
+      "step": 6251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602333931777379e-05,
+      "loss": 0.6746,
+      "step": 6252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6022698127725064e-05,
+      "loss": 0.6277,
+      "step": 6253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602205693767633e-05,
+      "loss": 0.5177,
+      "step": 6254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60214157476276e-05,
+      "loss": 0.5763,
+      "step": 6255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6020774557578866e-05,
+      "loss": 0.5992,
+      "step": 6256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602013336753014e-05,
+      "loss": 0.6385,
+      "step": 6257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601949217748141e-05,
+      "loss": 0.551,
+      "step": 6258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6018850987432674e-05,
+      "loss": 0.6754,
+      "step": 6259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6018209797383946e-05,
+      "loss": 0.6148,
+      "step": 6260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601756860733522e-05,
+      "loss": 0.5925,
+      "step": 6261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601692741728649e-05,
+      "loss": 0.5378,
+      "step": 6262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6016286227237754e-05,
+      "loss": 0.5641,
+      "step": 6263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6015645037189026e-05,
+      "loss": 0.4311,
+      "step": 6264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601500384714029e-05,
+      "loss": 0.5801,
+      "step": 6265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601436265709156e-05,
+      "loss": 0.6468,
+      "step": 6266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6013721467042834e-05,
+      "loss": 0.585,
+      "step": 6267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6013080276994106e-05,
+      "loss": 0.6846,
+      "step": 6268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601243908694538e-05,
+      "loss": 0.6196,
+      "step": 6269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601179789689664e-05,
+      "loss": 0.5874,
+      "step": 6270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6011156706847914e-05,
+      "loss": 0.5378,
+      "step": 6271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601051551679918e-05,
+      "loss": 0.5975,
+      "step": 6272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600987432675045e-05,
+      "loss": 0.5796,
+      "step": 6273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6009233136701716e-05,
+      "loss": 0.5805,
+      "step": 6274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600859194665299e-05,
+      "loss": 0.6439,
+      "step": 6275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600795075660426e-05,
+      "loss": 0.619,
+      "step": 6276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600730956655553e-05,
+      "loss": 0.4708,
+      "step": 6277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60066683765068e-05,
+      "loss": 0.5633,
+      "step": 6278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600602718645807e-05,
+      "loss": 0.4858,
+      "step": 6279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600538599640934e-05,
+      "loss": 0.5028,
+      "step": 6280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6004744806360604e-05,
+      "loss": 0.559,
+      "step": 6281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6004103616311876e-05,
+      "loss": 0.5585,
+      "step": 6282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600346242626314e-05,
+      "loss": 0.6232,
+      "step": 6283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600282123621442e-05,
+      "loss": 0.6454,
+      "step": 6284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600218004616569e-05,
+      "loss": 0.5029,
+      "step": 6285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6001538856116956e-05,
+      "loss": 0.5314,
+      "step": 6286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600089766606823e-05,
+      "loss": 0.6847,
+      "step": 6287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.600025647601949e-05,
+      "loss": 0.5464,
+      "step": 6288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5999615285970764e-05,
+      "loss": 0.5775,
+      "step": 6289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599897409592203e-05,
+      "loss": 0.6973,
+      "step": 6290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59983329058733e-05,
+      "loss": 0.4945,
+      "step": 6291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599769171582457e-05,
+      "loss": 0.6407,
+      "step": 6292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5997050525775845e-05,
+      "loss": 0.6703,
+      "step": 6293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5996409335727116e-05,
+      "loss": 0.5926,
+      "step": 6294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599576814567838e-05,
+      "loss": 0.5232,
+      "step": 6295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599512695562965e-05,
+      "loss": 0.5571,
+      "step": 6296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599448576558092e-05,
+      "loss": 0.4914,
+      "step": 6297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599384457553219e-05,
+      "loss": 0.595,
+      "step": 6298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5993203385483454e-05,
+      "loss": 0.5817,
+      "step": 6299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599256219543473e-05,
+      "loss": 0.5784,
+      "step": 6300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5991921005386e-05,
+      "loss": 0.5444,
+      "step": 6301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599127981533727e-05,
+      "loss": 0.629,
+      "step": 6302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.599063862528854e-05,
+      "loss": 0.6729,
+      "step": 6303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5989997435239806e-05,
+      "loss": 0.6282,
+      "step": 6304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598935624519108e-05,
+      "loss": 0.5706,
+      "step": 6305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598871505514234e-05,
+      "loss": 0.5921,
+      "step": 6306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5988073865093615e-05,
+      "loss": 0.5986,
+      "step": 6307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5987432675044886e-05,
+      "loss": 0.5852,
+      "step": 6308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598679148499616e-05,
+      "loss": 0.5939,
+      "step": 6309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598615029494742e-05,
+      "loss": 0.6385,
+      "step": 6310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5985509104898695e-05,
+      "loss": 0.6158,
+      "step": 6311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5984867914849966e-05,
+      "loss": 0.7266,
+      "step": 6312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598422672480123e-05,
+      "loss": 0.6473,
+      "step": 6313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59835855347525e-05,
+      "loss": 0.5546,
+      "step": 6314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598294434470377e-05,
+      "loss": 0.6311,
+      "step": 6315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598230315465504e-05,
+      "loss": 0.5478,
+      "step": 6316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598166196460631e-05,
+      "loss": 0.5656,
+      "step": 6317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598102077455758e-05,
+      "loss": 0.5409,
+      "step": 6318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.598037958450885e-05,
+      "loss": 0.5995,
+      "step": 6319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597973839446012e-05,
+      "loss": 0.5936,
+      "step": 6320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597909720441139e-05,
+      "loss": 0.5803,
+      "step": 6321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5978456014362657e-05,
+      "loss": 0.6117,
+      "step": 6322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597781482431393e-05,
+      "loss": 0.6598,
+      "step": 6323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59771736342652e-05,
+      "loss": 0.5779,
+      "step": 6324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597653244421647e-05,
+      "loss": 0.576,
+      "step": 6325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597589125416774e-05,
+      "loss": 0.5318,
+      "step": 6326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597525006411901e-05,
+      "loss": 0.5791,
+      "step": 6327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597460887407027e-05,
+      "loss": 0.5312,
+      "step": 6328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5973967684021545e-05,
+      "loss": 0.584,
+      "step": 6329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597332649397282e-05,
+      "loss": 0.5315,
+      "step": 6330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597268530392408e-05,
+      "loss": 0.6176,
+      "step": 6331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597204411387535e-05,
+      "loss": 0.5575,
+      "step": 6332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5971402923826625e-05,
+      "loss": 0.554,
+      "step": 6333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59707617337779e-05,
+      "loss": 0.6603,
+      "step": 6334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597012054372916e-05,
+      "loss": 0.5841,
+      "step": 6335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5969479353680433e-05,
+      "loss": 0.5684,
+      "step": 6336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59688381636317e-05,
+      "loss": 0.5651,
+      "step": 6337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596819697358297e-05,
+      "loss": 0.542,
+      "step": 6338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596755578353424e-05,
+      "loss": 0.6341,
+      "step": 6339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5966914593485514e-05,
+      "loss": 0.5364,
+      "step": 6340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5966273403436785e-05,
+      "loss": 0.5162,
+      "step": 6341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596563221338805e-05,
+      "loss": 0.5122,
+      "step": 6342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596499102333932e-05,
+      "loss": 0.6079,
+      "step": 6343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596434983329059e-05,
+      "loss": 0.6349,
+      "step": 6344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596370864324186e-05,
+      "loss": 0.5116,
+      "step": 6345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5963067453193124e-05,
+      "loss": 0.6316,
+      "step": 6346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5962426263144395e-05,
+      "loss": 0.6367,
+      "step": 6347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596178507309567e-05,
+      "loss": 0.6036,
+      "step": 6348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596114388304694e-05,
+      "loss": 0.5808,
+      "step": 6349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596050269299821e-05,
+      "loss": 0.6588,
+      "step": 6350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5959861502949475e-05,
+      "loss": 0.6268,
+      "step": 6351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595922031290075e-05,
+      "loss": 0.5382,
+      "step": 6352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595857912285201e-05,
+      "loss": 0.6129,
+      "step": 6353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5957937932803284e-05,
+      "loss": 0.5343,
+      "step": 6354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595729674275455e-05,
+      "loss": 0.6501,
+      "step": 6355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595665555270583e-05,
+      "loss": 0.5372,
+      "step": 6356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59560143626571e-05,
+      "loss": 0.5134,
+      "step": 6357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5955373172608364e-05,
+      "loss": 0.5991,
+      "step": 6358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5954731982559636e-05,
+      "loss": 0.5824,
+      "step": 6359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59540907925109e-05,
+      "loss": 0.6238,
+      "step": 6360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595344960246217e-05,
+      "loss": 0.5708,
+      "step": 6361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595280841241344e-05,
+      "loss": 0.5599,
+      "step": 6362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595216722236471e-05,
+      "loss": 0.525,
+      "step": 6363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595152603231598e-05,
+      "loss": 0.5772,
+      "step": 6364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.595088484226725e-05,
+      "loss": 0.5347,
+      "step": 6365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5950243652218524e-05,
+      "loss": 0.7044,
+      "step": 6366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.594960246216979e-05,
+      "loss": 0.4502,
+      "step": 6367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.594896127212106e-05,
+      "loss": 0.6131,
+      "step": 6368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5948320082072326e-05,
+      "loss": 0.5537,
+      "step": 6369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59476788920236e-05,
+      "loss": 0.6036,
+      "step": 6370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.594703770197486e-05,
+      "loss": 0.6596,
+      "step": 6371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.594639651192614e-05,
+      "loss": 0.5488,
+      "step": 6372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5945755321877406e-05,
+      "loss": 0.6098,
+      "step": 6373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594511413182868e-05,
+      "loss": 0.6044,
+      "step": 6374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594447294177995e-05,
+      "loss": 0.6423,
+      "step": 6375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5943831751731214e-05,
+      "loss": 0.5949,
+      "step": 6376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5943190561682486e-05,
+      "loss": 0.5951,
+      "step": 6377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594254937163375e-05,
+      "loss": 0.6238,
+      "step": 6378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594190818158502e-05,
+      "loss": 0.5847,
+      "step": 6379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5941266991536294e-05,
+      "loss": 0.6243,
+      "step": 6380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5940625801487566e-05,
+      "loss": 0.6115,
+      "step": 6381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593998461143883e-05,
+      "loss": 0.6234,
+      "step": 6382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.59393434213901e-05,
+      "loss": 0.6111,
+      "step": 6383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5938702231341374e-05,
+      "loss": 0.6136,
+      "step": 6384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593806104129264e-05,
+      "loss": 0.5509,
+      "step": 6385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593741985124391e-05,
+      "loss": 0.5956,
+      "step": 6386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5936778661195176e-05,
+      "loss": 0.6449,
+      "step": 6387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5936137471146454e-05,
+      "loss": 0.6022,
+      "step": 6388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593549628109772e-05,
+      "loss": 0.5207,
+      "step": 6389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593485509104899e-05,
+      "loss": 0.5693,
+      "step": 6390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593421390100026e-05,
+      "loss": 0.5047,
+      "step": 6391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593357271095153e-05,
+      "loss": 0.625,
+      "step": 6392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.59329315209028e-05,
+      "loss": 0.5473,
+      "step": 6393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5932290330854064e-05,
+      "loss": 0.608,
+      "step": 6394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5931649140805336e-05,
+      "loss": 0.637,
+      "step": 6395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593100795075661e-05,
+      "loss": 0.7163,
+      "step": 6396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593036676070788e-05,
+      "loss": 0.5941,
+      "step": 6397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5929725570659144e-05,
+      "loss": 0.5261,
+      "step": 6398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5929084380610416e-05,
+      "loss": 0.5011,
+      "step": 6399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592844319056169e-05,
+      "loss": 0.577,
+      "step": 6400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592780200051295e-05,
+      "loss": 0.6783,
+      "step": 6401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5927160810464224e-05,
+      "loss": 0.5689,
+      "step": 6402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592651962041549e-05,
+      "loss": 0.6306,
+      "step": 6403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592587843036676e-05,
+      "loss": 0.5784,
+      "step": 6404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592523724031803e-05,
+      "loss": 0.6583,
+      "step": 6405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5924596050269305e-05,
+      "loss": 0.568,
+      "step": 6406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592395486022057e-05,
+      "loss": 0.5712,
+      "step": 6407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592331367017184e-05,
+      "loss": 0.5974,
+      "step": 6408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592267248012311e-05,
+      "loss": 0.5873,
+      "step": 6409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592203129007438e-05,
+      "loss": 0.629,
+      "step": 6410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592139010002565e-05,
+      "loss": 0.5996,
+      "step": 6411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592074890997692e-05,
+      "loss": 0.6852,
+      "step": 6412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592010771992819e-05,
+      "loss": 0.6338,
+      "step": 6413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591946652987946e-05,
+      "loss": 0.5276,
+      "step": 6414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591882533983073e-05,
+      "loss": 0.5289,
+      "step": 6415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5918184149781995e-05,
+      "loss": 0.548,
+      "step": 6416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5917542959733266e-05,
+      "loss": 0.5436,
+      "step": 6417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591690176968454e-05,
+      "loss": 0.5285,
+      "step": 6418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.59162605796358e-05,
+      "loss": 0.5589,
+      "step": 6419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5915619389587075e-05,
+      "loss": 0.5051,
+      "step": 6420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5914978199538346e-05,
+      "loss": 0.5653,
+      "step": 6421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591433700948962e-05,
+      "loss": 0.5663,
+      "step": 6422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591369581944088e-05,
+      "loss": 0.5911,
+      "step": 6423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5913054629392155e-05,
+      "loss": 0.5395,
+      "step": 6424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591241343934342e-05,
+      "loss": 0.5375,
+      "step": 6425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591177224929469e-05,
+      "loss": 0.5487,
+      "step": 6426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591113105924596e-05,
+      "loss": 0.5391,
+      "step": 6427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5910489869197235e-05,
+      "loss": 0.518,
+      "step": 6428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5909848679148507e-05,
+      "loss": 0.5974,
+      "step": 6429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590920748909977e-05,
+      "loss": 0.6032,
+      "step": 6430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590856629905104e-05,
+      "loss": 0.5045,
+      "step": 6431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590792510900231e-05,
+      "loss": 0.6319,
+      "step": 6432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590728391895358e-05,
+      "loss": 0.609,
+      "step": 6433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5906642728904845e-05,
+      "loss": 0.5623,
+      "step": 6434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5906001538856117e-05,
+      "loss": 0.6023,
+      "step": 6435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590536034880739e-05,
+      "loss": 0.676,
+      "step": 6436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590471915875866e-05,
+      "loss": 0.5371,
+      "step": 6437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590407796870993e-05,
+      "loss": 0.6278,
+      "step": 6438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.59034367786612e-05,
+      "loss": 0.532,
+      "step": 6439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590279558861247e-05,
+      "loss": 0.6171,
+      "step": 6440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590215439856373e-05,
+      "loss": 0.5539,
+      "step": 6441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5901513208515005e-05,
+      "loss": 0.5397,
+      "step": 6442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590087201846627e-05,
+      "loss": 0.569,
+      "step": 6443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590023082841755e-05,
+      "loss": 0.5313,
+      "step": 6444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589958963836882e-05,
+      "loss": 0.6468,
+      "step": 6445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5898948448320085e-05,
+      "loss": 0.5581,
+      "step": 6446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589830725827136e-05,
+      "loss": 0.5349,
+      "step": 6447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589766606822262e-05,
+      "loss": 0.6538,
+      "step": 6448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5897024878173893e-05,
+      "loss": 0.5774,
+      "step": 6449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589638368812516e-05,
+      "loss": 0.5196,
+      "step": 6450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589574249807643e-05,
+      "loss": 0.5,
+      "step": 6451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58951013080277e-05,
+      "loss": 0.626,
+      "step": 6452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5894460117978974e-05,
+      "loss": 0.5552,
+      "step": 6453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5893818927930245e-05,
+      "loss": 0.6161,
+      "step": 6454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589317773788151e-05,
+      "loss": 0.6253,
+      "step": 6455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589253654783278e-05,
+      "loss": 0.6359,
+      "step": 6456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589189535778405e-05,
+      "loss": 0.5712,
+      "step": 6457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589125416773532e-05,
+      "loss": 0.6155,
+      "step": 6458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5890612977686584e-05,
+      "loss": 0.5986,
+      "step": 6459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588997178763786e-05,
+      "loss": 0.6397,
+      "step": 6460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588933059758913e-05,
+      "loss": 0.5965,
+      "step": 6461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58886894075404e-05,
+      "loss": 0.6234,
+      "step": 6462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588804821749167e-05,
+      "loss": 0.606,
+      "step": 6463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5887407027442935e-05,
+      "loss": 0.6014,
+      "step": 6464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588676583739421e-05,
+      "loss": 0.5272,
+      "step": 6465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588612464734547e-05,
+      "loss": 0.5911,
+      "step": 6466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5885483457296744e-05,
+      "loss": 0.5509,
+      "step": 6467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5884842267248015e-05,
+      "loss": 0.6349,
+      "step": 6468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588420107719929e-05,
+      "loss": 0.5839,
+      "step": 6469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588355988715055e-05,
+      "loss": 0.5176,
+      "step": 6470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5882918697101824e-05,
+      "loss": 0.5659,
+      "step": 6471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5882277507053096e-05,
+      "loss": 0.6735,
+      "step": 6472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588163631700436e-05,
+      "loss": 0.4758,
+      "step": 6473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588099512695563e-05,
+      "loss": 0.5928,
+      "step": 6474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58803539369069e-05,
+      "loss": 0.5516,
+      "step": 6475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587971274685817e-05,
+      "loss": 0.5851,
+      "step": 6476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587907155680944e-05,
+      "loss": 0.5381,
+      "step": 6477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587843036676071e-05,
+      "loss": 0.5725,
+      "step": 6478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587778917671198e-05,
+      "loss": 0.5635,
+      "step": 6479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587714798666325e-05,
+      "loss": 0.5695,
+      "step": 6480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587650679661452e-05,
+      "loss": 0.5876,
+      "step": 6481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5875865606565786e-05,
+      "loss": 0.5823,
+      "step": 6482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587522441651706e-05,
+      "loss": 0.5202,
+      "step": 6483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587458322646833e-05,
+      "loss": 0.604,
+      "step": 6484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58739420364196e-05,
+      "loss": 0.5383,
+      "step": 6485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5873300846370866e-05,
+      "loss": 0.6686,
+      "step": 6486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587265965632214e-05,
+      "loss": 0.5384,
+      "step": 6487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58720184662734e-05,
+      "loss": 0.4824,
+      "step": 6488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5871377276224674e-05,
+      "loss": 0.6148,
+      "step": 6489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5870736086175946e-05,
+      "loss": 0.5645,
+      "step": 6490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587009489612721e-05,
+      "loss": 0.5493,
+      "step": 6491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586945370607848e-05,
+      "loss": 0.6073,
+      "step": 6492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5868812516029754e-05,
+      "loss": 0.516,
+      "step": 6493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5868171325981026e-05,
+      "loss": 0.5885,
+      "step": 6494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586753013593229e-05,
+      "loss": 0.5885,
+      "step": 6495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586688894588356e-05,
+      "loss": 0.5209,
+      "step": 6496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586624775583483e-05,
+      "loss": 0.6111,
+      "step": 6497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58656065657861e-05,
+      "loss": 0.6133,
+      "step": 6498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586496537573737e-05,
+      "loss": 0.5724,
+      "step": 6499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586432418568864e-05,
+      "loss": 0.6271,
+      "step": 6500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5863682995639914e-05,
+      "loss": 0.6271,
+      "step": 6501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586304180559118e-05,
+      "loss": 0.6045,
+      "step": 6502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586240061554245e-05,
+      "loss": 0.604,
+      "step": 6503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5861759425493716e-05,
+      "loss": 0.498,
+      "step": 6504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586111823544499e-05,
+      "loss": 0.5711,
+      "step": 6505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586047704539625e-05,
+      "loss": 0.5217,
+      "step": 6506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5859835855347524e-05,
+      "loss": 0.5287,
+      "step": 6507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5859194665298796e-05,
+      "loss": 0.5921,
+      "step": 6508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.585855347525007e-05,
+      "loss": 0.5874,
+      "step": 6509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.585791228520134e-05,
+      "loss": 0.5523,
+      "step": 6510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5857271095152604e-05,
+      "loss": 0.5917,
+      "step": 6511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5856629905103876e-05,
+      "loss": 0.5999,
+      "step": 6512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.585598871505514e-05,
+      "loss": 0.5679,
+      "step": 6513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.585534752500641e-05,
+      "loss": 0.6271,
+      "step": 6514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.585470633495768e-05,
+      "loss": 0.636,
+      "step": 6515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5854065144908956e-05,
+      "loss": 0.5215,
+      "step": 6516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.585342395486023e-05,
+      "loss": 0.5519,
+      "step": 6517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.585278276481149e-05,
+      "loss": 0.5928,
+      "step": 6518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5852141574762765e-05,
+      "loss": 0.5576,
+      "step": 6519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.585150038471403e-05,
+      "loss": 0.5804,
+      "step": 6520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58508591946653e-05,
+      "loss": 0.5912,
+      "step": 6521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5850218004616566e-05,
+      "loss": 0.5188,
+      "step": 6522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584957681456784e-05,
+      "loss": 0.6246,
+      "step": 6523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584893562451911e-05,
+      "loss": 0.5795,
+      "step": 6524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584829443447038e-05,
+      "loss": 0.5415,
+      "step": 6525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584765324442165e-05,
+      "loss": 0.5335,
+      "step": 6526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584701205437292e-05,
+      "loss": 0.605,
+      "step": 6527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584637086432419e-05,
+      "loss": 0.5116,
+      "step": 6528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5845729674275455e-05,
+      "loss": 0.5525,
+      "step": 6529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5845088484226726e-05,
+      "loss": 0.5354,
+      "step": 6530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584444729417799e-05,
+      "loss": 0.6093,
+      "step": 6531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584380610412927e-05,
+      "loss": 0.5288,
+      "step": 6532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5843164914080535e-05,
+      "loss": 0.6368,
+      "step": 6533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5842523724031806e-05,
+      "loss": 0.6075,
+      "step": 6534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584188253398308e-05,
+      "loss": 0.5257,
+      "step": 6535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.584124134393434e-05,
+      "loss": 0.6076,
+      "step": 6536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5840600153885615e-05,
+      "loss": 0.5743,
+      "step": 6537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583995896383688e-05,
+      "loss": 0.5817,
+      "step": 6538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583931777378815e-05,
+      "loss": 0.5981,
+      "step": 6539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583867658373942e-05,
+      "loss": 0.5181,
+      "step": 6540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5838035393690695e-05,
+      "loss": 0.5737,
+      "step": 6541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583739420364196e-05,
+      "loss": 0.6536,
+      "step": 6542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583675301359323e-05,
+      "loss": 0.4801,
+      "step": 6543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58361118235445e-05,
+      "loss": 0.5046,
+      "step": 6544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583547063349577e-05,
+      "loss": 0.5334,
+      "step": 6545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583482944344704e-05,
+      "loss": 0.5108,
+      "step": 6546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5834188253398305e-05,
+      "loss": 0.5691,
+      "step": 6547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583354706334958e-05,
+      "loss": 0.5977,
+      "step": 6548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583290587330085e-05,
+      "loss": 0.6014,
+      "step": 6549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583226468325212e-05,
+      "loss": 0.5727,
+      "step": 6550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5831623493203385e-05,
+      "loss": 0.5282,
+      "step": 6551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583098230315466e-05,
+      "loss": 0.5121,
+      "step": 6552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583034111310593e-05,
+      "loss": 0.5503,
+      "step": 6553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582969992305719e-05,
+      "loss": 0.5239,
+      "step": 6554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5829058733008465e-05,
+      "loss": 0.6199,
+      "step": 6555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582841754295974e-05,
+      "loss": 0.5306,
+      "step": 6556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582777635291101e-05,
+      "loss": 0.5985,
+      "step": 6557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5827135162862273e-05,
+      "loss": 0.6763,
+      "step": 6558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5826493972813545e-05,
+      "loss": 0.666,
+      "step": 6559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582585278276481e-05,
+      "loss": 0.7133,
+      "step": 6560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582521159271608e-05,
+      "loss": 0.599,
+      "step": 6561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5824570402667354e-05,
+      "loss": 0.6222,
+      "step": 6562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582392921261862e-05,
+      "loss": 0.6234,
+      "step": 6563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582328802256989e-05,
+      "loss": 0.5927,
+      "step": 6564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582264683252116e-05,
+      "loss": 0.6262,
+      "step": 6565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5822005642472434e-05,
+      "loss": 0.648,
+      "step": 6566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58213644524237e-05,
+      "loss": 0.543,
+      "step": 6567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582072326237497e-05,
+      "loss": 0.6333,
+      "step": 6568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5820082072326235e-05,
+      "loss": 0.5758,
+      "step": 6569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581944088227751e-05,
+      "loss": 0.587,
+      "step": 6570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581879969222878e-05,
+      "loss": 0.5627,
+      "step": 6571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581815850218005e-05,
+      "loss": 0.5882,
+      "step": 6572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581751731213132e-05,
+      "loss": 0.6138,
+      "step": 6573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581687612208259e-05,
+      "loss": 0.6367,
+      "step": 6574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581623493203386e-05,
+      "loss": 0.5952,
+      "step": 6575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5815593741985124e-05,
+      "loss": 0.5009,
+      "step": 6576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5814952551936395e-05,
+      "loss": 0.6016,
+      "step": 6577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581431136188766e-05,
+      "loss": 0.6068,
+      "step": 6578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581367017183893e-05,
+      "loss": 0.5109,
+      "step": 6579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5813028981790204e-05,
+      "loss": 0.5945,
+      "step": 6580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5812387791741475e-05,
+      "loss": 0.5288,
+      "step": 6581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581174660169275e-05,
+      "loss": 0.4935,
+      "step": 6582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.581110541164401e-05,
+      "loss": 0.5928,
+      "step": 6583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5810464221595284e-05,
+      "loss": 0.6098,
+      "step": 6584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580982303154655e-05,
+      "loss": 0.5122,
+      "step": 6585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580918184149782e-05,
+      "loss": 0.5579,
+      "step": 6586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580854065144909e-05,
+      "loss": 0.6238,
+      "step": 6587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5807899461400364e-05,
+      "loss": 0.6052,
+      "step": 6588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5807258271351636e-05,
+      "loss": 0.615,
+      "step": 6589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58066170813029e-05,
+      "loss": 0.6555,
+      "step": 6590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580597589125417e-05,
+      "loss": 0.6282,
+      "step": 6591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580533470120544e-05,
+      "loss": 0.7015,
+      "step": 6592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580469351115671e-05,
+      "loss": 0.6178,
+      "step": 6593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5804052321107974e-05,
+      "loss": 0.5236,
+      "step": 6594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5803411131059246e-05,
+      "loss": 0.5617,
+      "step": 6595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580276994101052e-05,
+      "loss": 0.5464,
+      "step": 6596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580212875096179e-05,
+      "loss": 0.5842,
+      "step": 6597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580148756091306e-05,
+      "loss": 0.5508,
+      "step": 6598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5800846370864326e-05,
+      "loss": 0.5848,
+      "step": 6599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58002051808156e-05,
+      "loss": 0.6581,
+      "step": 6600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579956399076686e-05,
+      "loss": 0.4694,
+      "step": 6601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5798922800718134e-05,
+      "loss": 0.534,
+      "step": 6602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.57982816106694e-05,
+      "loss": 0.5952,
+      "step": 6603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579764042062068e-05,
+      "loss": 0.6302,
+      "step": 6604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579699923057195e-05,
+      "loss": 0.4754,
+      "step": 6605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5796358040523214e-05,
+      "loss": 0.5433,
+      "step": 6606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5795716850474486e-05,
+      "loss": 0.5184,
+      "step": 6607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579507566042575e-05,
+      "loss": 0.5124,
+      "step": 6608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579443447037702e-05,
+      "loss": 0.5921,
+      "step": 6609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579379328032829e-05,
+      "loss": 0.7201,
+      "step": 6610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579315209027956e-05,
+      "loss": 0.5862,
+      "step": 6611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579251090023083e-05,
+      "loss": 0.613,
+      "step": 6612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.57918697101821e-05,
+      "loss": 0.4656,
+      "step": 6613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5791228520133374e-05,
+      "loss": 0.5923,
+      "step": 6614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.579058733008464e-05,
+      "loss": 0.6342,
+      "step": 6615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.578994614003591e-05,
+      "loss": 0.4485,
+      "step": 6616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5789304949987176e-05,
+      "loss": 0.5142,
+      "step": 6617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.578866375993845e-05,
+      "loss": 0.5755,
+      "step": 6618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.578802256988971e-05,
+      "loss": 0.6033,
+      "step": 6619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.578738137984099e-05,
+      "loss": 0.6084,
+      "step": 6620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5786740189792256e-05,
+      "loss": 0.5439,
+      "step": 6621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.578609899974353e-05,
+      "loss": 0.6245,
+      "step": 6622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.57854578096948e-05,
+      "loss": 0.6131,
+      "step": 6623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5784816619646064e-05,
+      "loss": 0.5613,
+      "step": 6624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5784175429597336e-05,
+      "loss": 0.5416,
+      "step": 6625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.57835342395486e-05,
+      "loss": 0.6287,
+      "step": 6626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.578289304949987e-05,
+      "loss": 0.6506,
+      "step": 6627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5782251859451145e-05,
+      "loss": 0.5348,
+      "step": 6628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5781610669402416e-05,
+      "loss": 0.6056,
+      "step": 6629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.578096947935368e-05,
+      "loss": 0.6527,
+      "step": 6630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.578032828930495e-05,
+      "loss": 0.6311,
+      "step": 6631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5779687099256225e-05,
+      "loss": 0.603,
+      "step": 6632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577904590920749e-05,
+      "loss": 0.5476,
+      "step": 6633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577840471915876e-05,
+      "loss": 0.5824,
+      "step": 6634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5777763529110026e-05,
+      "loss": 0.5917,
+      "step": 6635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5777122339061305e-05,
+      "loss": 0.5043,
+      "step": 6636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577648114901257e-05,
+      "loss": 0.6001,
+      "step": 6637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577583995896384e-05,
+      "loss": 0.6075,
+      "step": 6638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5775198768915106e-05,
+      "loss": 0.538,
+      "step": 6639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577455757886638e-05,
+      "loss": 0.6252,
+      "step": 6640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577391638881765e-05,
+      "loss": 0.6145,
+      "step": 6641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5773275198768915e-05,
+      "loss": 0.5066,
+      "step": 6642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5772634008720186e-05,
+      "loss": 0.5869,
+      "step": 6643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577199281867146e-05,
+      "loss": 0.4995,
+      "step": 6644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577135162862273e-05,
+      "loss": 0.5054,
+      "step": 6645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5770710438573995e-05,
+      "loss": 0.5393,
+      "step": 6646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5770069248525266e-05,
+      "loss": 0.6076,
+      "step": 6647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576942805847653e-05,
+      "loss": 0.5516,
+      "step": 6648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57687868684278e-05,
+      "loss": 0.4264,
+      "step": 6649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5768145678379075e-05,
+      "loss": 0.5523,
+      "step": 6650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576750448833034e-05,
+      "loss": 0.5339,
+      "step": 6651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576686329828161e-05,
+      "loss": 0.5786,
+      "step": 6652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576622210823288e-05,
+      "loss": 0.6487,
+      "step": 6653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5765580918184155e-05,
+      "loss": 0.4725,
+      "step": 6654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576493972813542e-05,
+      "loss": 0.5844,
+      "step": 6655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576429853808669e-05,
+      "loss": 0.6294,
+      "step": 6656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5763657348037957e-05,
+      "loss": 0.573,
+      "step": 6657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576301615798923e-05,
+      "loss": 0.628,
+      "step": 6658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57623749679405e-05,
+      "loss": 0.6416,
+      "step": 6659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576173377789177e-05,
+      "loss": 0.5653,
+      "step": 6660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5761092587843043e-05,
+      "loss": 0.5054,
+      "step": 6661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576045139779431e-05,
+      "loss": 0.6188,
+      "step": 6662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575981020774558e-05,
+      "loss": 0.5523,
+      "step": 6663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5759169017696845e-05,
+      "loss": 0.5688,
+      "step": 6664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575852782764812e-05,
+      "loss": 0.6205,
+      "step": 6665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575788663759938e-05,
+      "loss": 0.5579,
+      "step": 6666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575724544755065e-05,
+      "loss": 0.5963,
+      "step": 6667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5756604257501925e-05,
+      "loss": 0.5528,
+      "step": 6668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57559630674532e-05,
+      "loss": 0.552,
+      "step": 6669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575532187740447e-05,
+      "loss": 0.5543,
+      "step": 6670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5754680687355733e-05,
+      "loss": 0.6105,
+      "step": 6671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5754039497307005e-05,
+      "loss": 0.5138,
+      "step": 6672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575339830725827e-05,
+      "loss": 0.5497,
+      "step": 6673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575275711720954e-05,
+      "loss": 0.5337,
+      "step": 6674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5752115927160814e-05,
+      "loss": 0.5367,
+      "step": 6675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5751474737112085e-05,
+      "loss": 0.6414,
+      "step": 6676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575083354706336e-05,
+      "loss": 0.5404,
+      "step": 6677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575019235701462e-05,
+      "loss": 0.5879,
+      "step": 6678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5749551166965894e-05,
+      "loss": 0.7301,
+      "step": 6679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574890997691716e-05,
+      "loss": 0.5658,
+      "step": 6680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574826878686843e-05,
+      "loss": 0.5239,
+      "step": 6681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5747627596819695e-05,
+      "loss": 0.551,
+      "step": 6682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574698640677097e-05,
+      "loss": 0.5821,
+      "step": 6683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574634521672224e-05,
+      "loss": 0.5777,
+      "step": 6684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574570402667351e-05,
+      "loss": 0.4886,
+      "step": 6685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574506283662478e-05,
+      "loss": 0.6235,
+      "step": 6686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574442164657605e-05,
+      "loss": 0.6043,
+      "step": 6687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574378045652732e-05,
+      "loss": 0.5181,
+      "step": 6688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5743139266478584e-05,
+      "loss": 0.496,
+      "step": 6689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5742498076429855e-05,
+      "loss": 0.6316,
+      "step": 6690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574185688638112e-05,
+      "loss": 0.599,
+      "step": 6691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57412156963324e-05,
+      "loss": 0.5443,
+      "step": 6692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5740574506283664e-05,
+      "loss": 0.5643,
+      "step": 6693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5739933316234936e-05,
+      "loss": 0.6395,
+      "step": 6694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573929212618621e-05,
+      "loss": 0.5862,
+      "step": 6695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573865093613747e-05,
+      "loss": 0.5705,
+      "step": 6696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5738009746088744e-05,
+      "loss": 0.5098,
+      "step": 6697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573736855604001e-05,
+      "loss": 0.5931,
+      "step": 6698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573672736599128e-05,
+      "loss": 0.5811,
+      "step": 6699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573608617594255e-05,
+      "loss": 0.5904,
+      "step": 6700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5735444985893824e-05,
+      "loss": 0.5391,
+      "step": 6701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573480379584509e-05,
+      "loss": 0.6247,
+      "step": 6702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573416260579636e-05,
+      "loss": 0.6344,
+      "step": 6703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573352141574763e-05,
+      "loss": 0.6459,
+      "step": 6704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57328802256989e-05,
+      "loss": 0.6105,
+      "step": 6705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573223903565017e-05,
+      "loss": 0.5496,
+      "step": 6706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5731597845601434e-05,
+      "loss": 0.5903,
+      "step": 6707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573095665555271e-05,
+      "loss": 0.5842,
+      "step": 6708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573031546550398e-05,
+      "loss": 0.5971,
+      "step": 6709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572967427545525e-05,
+      "loss": 0.533,
+      "step": 6710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5729033085406514e-05,
+      "loss": 0.5423,
+      "step": 6711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5728391895357786e-05,
+      "loss": 0.5622,
+      "step": 6712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572775070530906e-05,
+      "loss": 0.5394,
+      "step": 6713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572710951526032e-05,
+      "loss": 0.5383,
+      "step": 6714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5726468325211594e-05,
+      "loss": 0.6094,
+      "step": 6715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5725827135162866e-05,
+      "loss": 0.6462,
+      "step": 6716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572518594511414e-05,
+      "loss": 0.5362,
+      "step": 6717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57245447550654e-05,
+      "loss": 0.5498,
+      "step": 6718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5723903565016674e-05,
+      "loss": 0.5046,
+      "step": 6719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572326237496794e-05,
+      "loss": 0.5551,
+      "step": 6720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572262118491921e-05,
+      "loss": 0.6073,
+      "step": 6721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572197999487048e-05,
+      "loss": 0.618,
+      "step": 6722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572133880482175e-05,
+      "loss": 0.5544,
+      "step": 6723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572069761477302e-05,
+      "loss": 0.6053,
+      "step": 6724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572005642472429e-05,
+      "loss": 0.642,
+      "step": 6725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571941523467556e-05,
+      "loss": 0.5491,
+      "step": 6726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571877404462683e-05,
+      "loss": 0.6339,
+      "step": 6727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57181328545781e-05,
+      "loss": 0.61,
+      "step": 6728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5717491664529364e-05,
+      "loss": 0.5942,
+      "step": 6729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5716850474480636e-05,
+      "loss": 0.5907,
+      "step": 6730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571620928443191e-05,
+      "loss": 0.5075,
+      "step": 6731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571556809438318e-05,
+      "loss": 0.6427,
+      "step": 6732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571492690433445e-05,
+      "loss": 0.6082,
+      "step": 6733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5714285714285716e-05,
+      "loss": 0.514,
+      "step": 6734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571364452423699e-05,
+      "loss": 0.6093,
+      "step": 6735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571300333418825e-05,
+      "loss": 0.5464,
+      "step": 6736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5712362144139524e-05,
+      "loss": 0.5774,
+      "step": 6737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571172095409079e-05,
+      "loss": 0.5729,
+      "step": 6738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571107976404206e-05,
+      "loss": 0.5805,
+      "step": 6739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571043857399333e-05,
+      "loss": 0.5905,
+      "step": 6740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5709797383944605e-05,
+      "loss": 0.5141,
+      "step": 6741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5709156193895876e-05,
+      "loss": 0.5279,
+      "step": 6742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570851500384714e-05,
+      "loss": 0.5052,
+      "step": 6743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570787381379841e-05,
+      "loss": 0.6382,
+      "step": 6744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570723262374968e-05,
+      "loss": 0.5823,
+      "step": 6745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570659143370095e-05,
+      "loss": 0.5377,
+      "step": 6746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570595024365222e-05,
+      "loss": 0.5965,
+      "step": 6747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570530905360349e-05,
+      "loss": 0.6889,
+      "step": 6748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5704667863554765e-05,
+      "loss": 0.566,
+      "step": 6749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570402667350603e-05,
+      "loss": 0.5339,
+      "step": 6750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57033854834573e-05,
+      "loss": 0.5808,
+      "step": 6751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5702744293408566e-05,
+      "loss": 0.5289,
+      "step": 6752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570210310335984e-05,
+      "loss": 0.6074,
+      "step": 6753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57014619133111e-05,
+      "loss": 0.4924,
+      "step": 6754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5700820723262375e-05,
+      "loss": 0.5906,
+      "step": 6755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5700179533213646e-05,
+      "loss": 0.5996,
+      "step": 6756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569953834316492e-05,
+      "loss": 0.5984,
+      "step": 6757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569889715311619e-05,
+      "loss": 0.6152,
+      "step": 6758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5698255963067455e-05,
+      "loss": 0.6147,
+      "step": 6759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5697614773018727e-05,
+      "loss": 0.5853,
+      "step": 6760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569697358296999e-05,
+      "loss": 0.5384,
+      "step": 6761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569633239292126e-05,
+      "loss": 0.6162,
+      "step": 6762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569569120287253e-05,
+      "loss": 0.6251,
+      "step": 6763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5695050012823807e-05,
+      "loss": 0.6484,
+      "step": 6764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569440882277507e-05,
+      "loss": 0.6177,
+      "step": 6765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569376763272634e-05,
+      "loss": 0.6295,
+      "step": 6766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5693126442677615e-05,
+      "loss": 0.5266,
+      "step": 6767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569248525262888e-05,
+      "loss": 0.5782,
+      "step": 6768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569184406258015e-05,
+      "loss": 0.5451,
+      "step": 6769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5691202872531417e-05,
+      "loss": 0.564,
+      "step": 6770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.569056168248269e-05,
+      "loss": 0.6213,
+      "step": 6771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568992049243396e-05,
+      "loss": 0.5826,
+      "step": 6772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568927930238523e-05,
+      "loss": 0.5633,
+      "step": 6773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5688638112336503e-05,
+      "loss": 0.6256,
+      "step": 6774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568799692228777e-05,
+      "loss": 0.6109,
+      "step": 6775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568735573223904e-05,
+      "loss": 0.6121,
+      "step": 6776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5686714542190305e-05,
+      "loss": 0.5725,
+      "step": 6777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568607335214158e-05,
+      "loss": 0.5941,
+      "step": 6778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568543216209284e-05,
+      "loss": 0.5831,
+      "step": 6779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568479097204412e-05,
+      "loss": 0.6597,
+      "step": 6780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5684149781995385e-05,
+      "loss": 0.5839,
+      "step": 6781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568350859194666e-05,
+      "loss": 0.6222,
+      "step": 6782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568286740189793e-05,
+      "loss": 0.6304,
+      "step": 6783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5682226211849194e-05,
+      "loss": 0.6545,
+      "step": 6784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5681585021800465e-05,
+      "loss": 0.5053,
+      "step": 6785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568094383175173e-05,
+      "loss": 0.5334,
+      "step": 6786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5680302641703e-05,
+      "loss": 0.5469,
+      "step": 6787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5679661451654274e-05,
+      "loss": 0.4691,
+      "step": 6788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5679020261605545e-05,
+      "loss": 0.6102,
+      "step": 6789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.567837907155681e-05,
+      "loss": 0.6669,
+      "step": 6790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.567773788150808e-05,
+      "loss": 0.6349,
+      "step": 6791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5677096691459354e-05,
+      "loss": 0.6148,
+      "step": 6792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.567645550141062e-05,
+      "loss": 0.4918,
+      "step": 6793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.567581431136189e-05,
+      "loss": 0.572,
+      "step": 6794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5675173121313155e-05,
+      "loss": 0.5411,
+      "step": 6795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5674531931264434e-05,
+      "loss": 0.618,
+      "step": 6796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.56738907412157e-05,
+      "loss": 0.5611,
+      "step": 6797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.567324955116697e-05,
+      "loss": 0.5854,
+      "step": 6798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5672608361118235e-05,
+      "loss": 0.5589,
+      "step": 6799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.567196717106951e-05,
+      "loss": 0.5643,
+      "step": 6800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.567132598102078e-05,
+      "loss": 0.6132,
+      "step": 6801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5670684790972044e-05,
+      "loss": 0.5558,
+      "step": 6802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5670043600923315e-05,
+      "loss": 0.5847,
+      "step": 6803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566940241087459e-05,
+      "loss": 0.6526,
+      "step": 6804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566876122082586e-05,
+      "loss": 0.5358,
+      "step": 6805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5668120030777124e-05,
+      "loss": 0.5741,
+      "step": 6806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5667478840728396e-05,
+      "loss": 0.6329,
+      "step": 6807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566683765067966e-05,
+      "loss": 0.5465,
+      "step": 6808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566619646063093e-05,
+      "loss": 0.5671,
+      "step": 6809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5665555270582204e-05,
+      "loss": 0.5319,
+      "step": 6810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566491408053347e-05,
+      "loss": 0.5339,
+      "step": 6811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566427289048474e-05,
+      "loss": 0.5986,
+      "step": 6812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566363170043601e-05,
+      "loss": 0.714,
+      "step": 6813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5662990510387284e-05,
+      "loss": 0.5541,
+      "step": 6814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566234932033855e-05,
+      "loss": 0.5633,
+      "step": 6815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566170813028982e-05,
+      "loss": 0.5141,
+      "step": 6816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5661066940241086e-05,
+      "loss": 0.646,
+      "step": 6817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.566042575019236e-05,
+      "loss": 0.6037,
+      "step": 6818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.565978456014363e-05,
+      "loss": 0.5549,
+      "step": 6819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.56591433700949e-05,
+      "loss": 0.6267,
+      "step": 6820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.565850218004617e-05,
+      "loss": 0.5958,
+      "step": 6821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.565786098999744e-05,
+      "loss": 0.5062,
+      "step": 6822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.565721979994871e-05,
+      "loss": 0.6329,
+      "step": 6823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5656578609899974e-05,
+      "loss": 0.5686,
+      "step": 6824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5655937419851246e-05,
+      "loss": 0.5609,
+      "step": 6825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.565529622980251e-05,
+      "loss": 0.5842,
+      "step": 6826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.565465503975378e-05,
+      "loss": 0.5275,
+      "step": 6827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5654013849705054e-05,
+      "loss": 0.566,
+      "step": 6828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5653372659656326e-05,
+      "loss": 0.5593,
+      "step": 6829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.56527314696076e-05,
+      "loss": 0.5801,
+      "step": 6830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.565209027955886e-05,
+      "loss": 0.6023,
+      "step": 6831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5651449089510134e-05,
+      "loss": 0.5501,
+      "step": 6832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.56508078994614e-05,
+      "loss": 0.5966,
+      "step": 6833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.565016670941267e-05,
+      "loss": 0.5815,
+      "step": 6834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564952551936394e-05,
+      "loss": 0.6026,
+      "step": 6835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5648884329315214e-05,
+      "loss": 0.662,
+      "step": 6836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5648243139266486e-05,
+      "loss": 0.6095,
+      "step": 6837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564760194921775e-05,
+      "loss": 0.653,
+      "step": 6838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564696075916902e-05,
+      "loss": 0.5636,
+      "step": 6839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564631956912029e-05,
+      "loss": 0.5577,
+      "step": 6840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564567837907156e-05,
+      "loss": 0.7025,
+      "step": 6841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5645037189022824e-05,
+      "loss": 0.5811,
+      "step": 6842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5644395998974096e-05,
+      "loss": 0.5874,
+      "step": 6843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564375480892537e-05,
+      "loss": 0.5873,
+      "step": 6844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564311361887664e-05,
+      "loss": 0.6004,
+      "step": 6845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564247242882791e-05,
+      "loss": 0.5716,
+      "step": 6846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5641831238779176e-05,
+      "loss": 0.4999,
+      "step": 6847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564119004873045e-05,
+      "loss": 0.5173,
+      "step": 6848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564054885868171e-05,
+      "loss": 0.4893,
+      "step": 6849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5639907668632984e-05,
+      "loss": 0.5675,
+      "step": 6850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563926647858425e-05,
+      "loss": 0.6027,
+      "step": 6851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563862528853553e-05,
+      "loss": 0.5779,
+      "step": 6852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563798409848679e-05,
+      "loss": 0.6083,
+      "step": 6853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5637342908438065e-05,
+      "loss": 0.6425,
+      "step": 6854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5636701718389336e-05,
+      "loss": 0.4939,
+      "step": 6855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.56360605283406e-05,
+      "loss": 0.644,
+      "step": 6856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563541933829187e-05,
+      "loss": 0.5346,
+      "step": 6857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563477814824314e-05,
+      "loss": 0.5509,
+      "step": 6858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563413695819441e-05,
+      "loss": 0.6687,
+      "step": 6859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563349576814568e-05,
+      "loss": 0.4958,
+      "step": 6860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563285457809695e-05,
+      "loss": 0.6196,
+      "step": 6861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563221338804822e-05,
+      "loss": 0.586,
+      "step": 6862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563157219799949e-05,
+      "loss": 0.6328,
+      "step": 6863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.563093100795076e-05,
+      "loss": 0.5574,
+      "step": 6864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5630289817902026e-05,
+      "loss": 0.5621,
+      "step": 6865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.56296486278533e-05,
+      "loss": 0.5452,
+      "step": 6866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562900743780456e-05,
+      "loss": 0.5599,
+      "step": 6867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562836624775584e-05,
+      "loss": 0.5509,
+      "step": 6868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5627725057707106e-05,
+      "loss": 0.5439,
+      "step": 6869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562708386765838e-05,
+      "loss": 0.6168,
+      "step": 6870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562644267760964e-05,
+      "loss": 0.6177,
+      "step": 6871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5625801487560915e-05,
+      "loss": 0.4974,
+      "step": 6872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5625160297512187e-05,
+      "loss": 0.6118,
+      "step": 6873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562451910746345e-05,
+      "loss": 0.7857,
+      "step": 6874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562387791741472e-05,
+      "loss": 0.5902,
+      "step": 6875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5623236727365995e-05,
+      "loss": 0.6203,
+      "step": 6876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562259553731727e-05,
+      "loss": 0.5493,
+      "step": 6877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562195434726853e-05,
+      "loss": 0.6108,
+      "step": 6878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.56213131572198e-05,
+      "loss": 0.5462,
+      "step": 6879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562067196717107e-05,
+      "loss": 0.5877,
+      "step": 6880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562003077712234e-05,
+      "loss": 0.6352,
+      "step": 6881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561938958707361e-05,
+      "loss": 0.689,
+      "step": 6882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5618748397024877e-05,
+      "loss": 0.5671,
+      "step": 6883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561810720697615e-05,
+      "loss": 0.5493,
+      "step": 6884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561746601692742e-05,
+      "loss": 0.5921,
+      "step": 6885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561682482687869e-05,
+      "loss": 0.4715,
+      "step": 6886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561618363682996e-05,
+      "loss": 0.6091,
+      "step": 6887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561554244678123e-05,
+      "loss": 0.5923,
+      "step": 6888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561490125673249e-05,
+      "loss": 0.6306,
+      "step": 6889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5614260066683765e-05,
+      "loss": 0.5856,
+      "step": 6890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561361887663504e-05,
+      "loss": 0.5525,
+      "step": 6891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561297768658631e-05,
+      "loss": 0.6269,
+      "step": 6892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.561233649653758e-05,
+      "loss": 0.6072,
+      "step": 6893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5611695306488845e-05,
+      "loss": 0.5077,
+      "step": 6894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.561105411644012e-05,
+      "loss": 0.5631,
+      "step": 6895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.561041292639138e-05,
+      "loss": 0.5704,
+      "step": 6896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5609771736342654e-05,
+      "loss": 0.5832,
+      "step": 6897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560913054629392e-05,
+      "loss": 0.5525,
+      "step": 6898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560848935624519e-05,
+      "loss": 0.5667,
+      "step": 6899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560784816619646e-05,
+      "loss": 0.5324,
+      "step": 6900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5607206976147734e-05,
+      "loss": 0.5661,
+      "step": 6901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5606565786099005e-05,
+      "loss": 0.6146,
+      "step": 6902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560592459605027e-05,
+      "loss": 0.5538,
+      "step": 6903
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560528340600154e-05,
+      "loss": 0.599,
+      "step": 6904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560464221595281e-05,
+      "loss": 0.5447,
+      "step": 6905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560400102590408e-05,
+      "loss": 0.6151,
+      "step": 6906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560335983585535e-05,
+      "loss": 0.6214,
+      "step": 6907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560271864580662e-05,
+      "loss": 0.4803,
+      "step": 6908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5602077455757894e-05,
+      "loss": 0.5611,
+      "step": 6909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560143626570916e-05,
+      "loss": 0.54,
+      "step": 6910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560079507566043e-05,
+      "loss": 0.614,
+      "step": 6911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5600153885611695e-05,
+      "loss": 0.5373,
+      "step": 6912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559951269556297e-05,
+      "loss": 0.6067,
+      "step": 6913
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559887150551423e-05,
+      "loss": 0.549,
+      "step": 6914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5598230315465504e-05,
+      "loss": 0.5307,
+      "step": 6915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5597589125416775e-05,
+      "loss": 0.6186,
+      "step": 6916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559694793536805e-05,
+      "loss": 0.5549,
+      "step": 6917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559630674531932e-05,
+      "loss": 0.6294,
+      "step": 6918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5595665555270584e-05,
+      "loss": 0.4727,
+      "step": 6919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5595024365221856e-05,
+      "loss": 0.4728,
+      "step": 6920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559438317517312e-05,
+      "loss": 0.6172,
+      "step": 6921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559374198512439e-05,
+      "loss": 0.5592,
+      "step": 6922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5593100795075664e-05,
+      "loss": 0.6671,
+      "step": 6923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5592459605026936e-05,
+      "loss": 0.592,
+      "step": 6924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55918184149782e-05,
+      "loss": 0.4372,
+      "step": 6925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559117722492947e-05,
+      "loss": 0.4929,
+      "step": 6926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5590536034880744e-05,
+      "loss": 0.631,
+      "step": 6927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558989484483201e-05,
+      "loss": 0.6556,
+      "step": 6928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558925365478328e-05,
+      "loss": 0.6305,
+      "step": 6929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5588612464734546e-05,
+      "loss": 0.6041,
+      "step": 6930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558797127468582e-05,
+      "loss": 0.6017,
+      "step": 6931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558733008463709e-05,
+      "loss": 0.59,
+      "step": 6932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558668889458836e-05,
+      "loss": 0.5323,
+      "step": 6933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5586047704539626e-05,
+      "loss": 0.6266,
+      "step": 6934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55854065144909e-05,
+      "loss": 0.5003,
+      "step": 6935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558476532444217e-05,
+      "loss": 0.5572,
+      "step": 6936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5584124134393434e-05,
+      "loss": 0.5801,
+      "step": 6937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5583482944344706e-05,
+      "loss": 0.5261,
+      "step": 6938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558284175429597e-05,
+      "loss": 0.5827,
+      "step": 6939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558220056424725e-05,
+      "loss": 0.6387,
+      "step": 6940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5581559374198514e-05,
+      "loss": 0.5711,
+      "step": 6941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5580918184149786e-05,
+      "loss": 0.6041,
+      "step": 6942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558027699410105e-05,
+      "loss": 0.5832,
+      "step": 6943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557963580405232e-05,
+      "loss": 0.5197,
+      "step": 6944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5578994614003594e-05,
+      "loss": 0.6342,
+      "step": 6945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557835342395486e-05,
+      "loss": 0.5437,
+      "step": 6946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557771223390613e-05,
+      "loss": 0.5702,
+      "step": 6947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55770710438574e-05,
+      "loss": 0.6241,
+      "step": 6948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5576429853808674e-05,
+      "loss": 0.4985,
+      "step": 6949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557578866375994e-05,
+      "loss": 0.5562,
+      "step": 6950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557514747371121e-05,
+      "loss": 0.556,
+      "step": 6951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5574506283662476e-05,
+      "loss": 0.698,
+      "step": 6952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557386509361375e-05,
+      "loss": 0.6031,
+      "step": 6953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557322390356502e-05,
+      "loss": 0.6465,
+      "step": 6954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5572582713516284e-05,
+      "loss": 0.6265,
+      "step": 6955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557194152346756e-05,
+      "loss": 0.5332,
+      "step": 6956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557130033341883e-05,
+      "loss": 0.5942,
+      "step": 6957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55706591433701e-05,
+      "loss": 0.5542,
+      "step": 6958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5570017953321364e-05,
+      "loss": 0.6581,
+      "step": 6959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5569376763272636e-05,
+      "loss": 0.5727,
+      "step": 6960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556873557322391e-05,
+      "loss": 0.64,
+      "step": 6961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556809438317517e-05,
+      "loss": 0.5921,
+      "step": 6962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5567453193126445e-05,
+      "loss": 0.5666,
+      "step": 6963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5566812003077716e-05,
+      "loss": 0.5033,
+      "step": 6964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556617081302899e-05,
+      "loss": 0.648,
+      "step": 6965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556552962298025e-05,
+      "loss": 0.4895,
+      "step": 6966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5564888432931525e-05,
+      "loss": 0.6547,
+      "step": 6967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556424724288279e-05,
+      "loss": 0.5148,
+      "step": 6968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556360605283406e-05,
+      "loss": 0.6223,
+      "step": 6969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556296486278533e-05,
+      "loss": 0.6428,
+      "step": 6970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55623236727366e-05,
+      "loss": 0.4996,
+      "step": 6971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556168248268787e-05,
+      "loss": 0.5875,
+      "step": 6972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556104129263914e-05,
+      "loss": 0.5632,
+      "step": 6973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556040010259041e-05,
+      "loss": 0.581,
+      "step": 6974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555975891254168e-05,
+      "loss": 0.5551,
+      "step": 6975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555911772249295e-05,
+      "loss": 0.6666,
+      "step": 6976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5558476532444215e-05,
+      "loss": 0.5591,
+      "step": 6977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5557835342395486e-05,
+      "loss": 0.537,
+      "step": 6978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555719415234676e-05,
+      "loss": 0.4858,
+      "step": 6979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555655296229803e-05,
+      "loss": 0.7543,
+      "step": 6980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55559117722493e-05,
+      "loss": 0.6151,
+      "step": 6981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5555270582200566e-05,
+      "loss": 0.6214,
+      "step": 6982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555462939215184e-05,
+      "loss": 0.5573,
+      "step": 6983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55539882021031e-05,
+      "loss": 0.4926,
+      "step": 6984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5553347012054375e-05,
+      "loss": 0.5411,
+      "step": 6985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555270582200564e-05,
+      "loss": 0.5845,
+      "step": 6986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555206463195691e-05,
+      "loss": 0.7337,
+      "step": 6987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555142344190818e-05,
+      "loss": 0.5984,
+      "step": 6988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5550782251859455e-05,
+      "loss": 0.5382,
+      "step": 6989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555014106181073e-05,
+      "loss": 0.4964,
+      "step": 6990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554949987176199e-05,
+      "loss": 0.6744,
+      "step": 6991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554885868171326e-05,
+      "loss": 0.685,
+      "step": 6992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554821749166453e-05,
+      "loss": 0.5188,
+      "step": 6993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55475763016158e-05,
+      "loss": 0.5403,
+      "step": 6994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554693511156707e-05,
+      "loss": 0.5501,
+      "step": 6995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5546293921518343e-05,
+      "loss": 0.5717,
+      "step": 6996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5545652731469615e-05,
+      "loss": 0.5265,
+      "step": 6997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554501154142088e-05,
+      "loss": 0.5683,
+      "step": 6998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554437035137215e-05,
+      "loss": 0.5895,
+      "step": 6999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554372916132342e-05,
+      "loss": 0.5543,
+      "step": 7000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554308797127469e-05,
+      "loss": 0.5596,
+      "step": 7001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554244678122595e-05,
+      "loss": 0.5354,
+      "step": 7002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5541805591177225e-05,
+      "loss": 0.5468,
+      "step": 7003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55411644011285e-05,
+      "loss": 0.5754,
+      "step": 7004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554052321107977e-05,
+      "loss": 0.4953,
+      "step": 7005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553988202103104e-05,
+      "loss": 0.4824,
+      "step": 7006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5539240830982305e-05,
+      "loss": 0.666,
+      "step": 7007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553859964093358e-05,
+      "loss": 0.6375,
+      "step": 7008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553795845088484e-05,
+      "loss": 0.6024,
+      "step": 7009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5537317260836114e-05,
+      "loss": 0.5548,
+      "step": 7010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553667607078738e-05,
+      "loss": 0.6234,
+      "step": 7011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553603488073866e-05,
+      "loss": 0.5383,
+      "step": 7012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553539369068992e-05,
+      "loss": 0.6828,
+      "step": 7013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5534752500641194e-05,
+      "loss": 0.6409,
+      "step": 7014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5534111310592465e-05,
+      "loss": 0.6699,
+      "step": 7015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553347012054373e-05,
+      "loss": 0.5641,
+      "step": 7016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5532828930495e-05,
+      "loss": 0.5641,
+      "step": 7017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553218774044627e-05,
+      "loss": 0.6824,
+      "step": 7018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553154655039754e-05,
+      "loss": 0.5664,
+      "step": 7019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553090536034881e-05,
+      "loss": 0.5482,
+      "step": 7020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553026417030008e-05,
+      "loss": 0.5812,
+      "step": 7021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552962298025135e-05,
+      "loss": 0.5591,
+      "step": 7022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552898179020262e-05,
+      "loss": 0.5764,
+      "step": 7023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552834060015389e-05,
+      "loss": 0.572,
+      "step": 7024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5527699410105155e-05,
+      "loss": 0.5523,
+      "step": 7025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552705822005643e-05,
+      "loss": 0.6028,
+      "step": 7026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552641703000769e-05,
+      "loss": 0.5751,
+      "step": 7027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552577583995897e-05,
+      "loss": 0.6174,
+      "step": 7028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5525134649910236e-05,
+      "loss": 0.5846,
+      "step": 7029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552449345986151e-05,
+      "loss": 0.5816,
+      "step": 7030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552385226981277e-05,
+      "loss": 0.6053,
+      "step": 7031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5523211079764044e-05,
+      "loss": 0.5805,
+      "step": 7032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5522569889715316e-05,
+      "loss": 0.556,
+      "step": 7033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552192869966658e-05,
+      "loss": 0.6198,
+      "step": 7034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552128750961785e-05,
+      "loss": 0.6297,
+      "step": 7035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5520646319569124e-05,
+      "loss": 0.5,
+      "step": 7036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5520005129520396e-05,
+      "loss": 0.5285,
+      "step": 7037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551936393947166e-05,
+      "loss": 0.5683,
+      "step": 7038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551872274942293e-05,
+      "loss": 0.6333,
+      "step": 7039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55180815593742e-05,
+      "loss": 0.5335,
+      "step": 7040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551744036932547e-05,
+      "loss": 0.5545,
+      "step": 7041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551679917927674e-05,
+      "loss": 0.5367,
+      "step": 7042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5516157989228006e-05,
+      "loss": 0.5936,
+      "step": 7043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5515516799179284e-05,
+      "loss": 0.5226,
+      "step": 7044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551487560913055e-05,
+      "loss": 0.4598,
+      "step": 7045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551423441908182e-05,
+      "loss": 0.5769,
+      "step": 7046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5513593229033086e-05,
+      "loss": 0.5963,
+      "step": 7047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551295203898436e-05,
+      "loss": 0.4878,
+      "step": 7048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551231084893562e-05,
+      "loss": 0.6382,
+      "step": 7049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5511669658886894e-05,
+      "loss": 0.5279,
+      "step": 7050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5511028468838166e-05,
+      "loss": 0.5453,
+      "step": 7051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.551038727878944e-05,
+      "loss": 0.5591,
+      "step": 7052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550974608874071e-05,
+      "loss": 0.5784,
+      "step": 7053
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5509104898691974e-05,
+      "loss": 0.3719,
+      "step": 7054
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5508463708643246e-05,
+      "loss": 0.574,
+      "step": 7055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550782251859451e-05,
+      "loss": 0.6026,
+      "step": 7056
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550718132854578e-05,
+      "loss": 0.5303,
+      "step": 7057
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550654013849705e-05,
+      "loss": 0.5146,
+      "step": 7058
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550589894844832e-05,
+      "loss": 0.549,
+      "step": 7059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550525775839959e-05,
+      "loss": 0.4975,
+      "step": 7060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550461656835086e-05,
+      "loss": 0.4747,
+      "step": 7061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5503975378302134e-05,
+      "loss": 0.6723,
+      "step": 7062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55033341882534e-05,
+      "loss": 0.5859,
+      "step": 7063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550269299820467e-05,
+      "loss": 0.5512,
+      "step": 7064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5502051808155936e-05,
+      "loss": 0.4922,
+      "step": 7065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550141061810721e-05,
+      "loss": 0.5731,
+      "step": 7066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550076942805848e-05,
+      "loss": 0.5629,
+      "step": 7067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.550012823800975e-05,
+      "loss": 0.6036,
+      "step": 7068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549948704796102e-05,
+      "loss": 0.6346,
+      "step": 7069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549884585791229e-05,
+      "loss": 0.6036,
+      "step": 7070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549820466786356e-05,
+      "loss": 0.7241,
+      "step": 7071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5497563477814824e-05,
+      "loss": 0.6259,
+      "step": 7072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5496922287766096e-05,
+      "loss": 0.5513,
+      "step": 7073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549628109771736e-05,
+      "loss": 0.6174,
+      "step": 7074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549563990766863e-05,
+      "loss": 0.5735,
+      "step": 7075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5494998717619905e-05,
+      "loss": 0.4934,
+      "step": 7076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5494357527571176e-05,
+      "loss": 0.5915,
+      "step": 7077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549371633752245e-05,
+      "loss": 0.6383,
+      "step": 7078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549307514747371e-05,
+      "loss": 0.5655,
+      "step": 7079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5492433957424985e-05,
+      "loss": 0.5298,
+      "step": 7080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549179276737625e-05,
+      "loss": 0.5513,
+      "step": 7081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549115157732752e-05,
+      "loss": 0.6554,
+      "step": 7082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549051038727879e-05,
+      "loss": 0.5686,
+      "step": 7083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5489869197230065e-05,
+      "loss": 0.6024,
+      "step": 7084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.548922800718133e-05,
+      "loss": 0.4935,
+      "step": 7085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.54885868171326e-05,
+      "loss": 0.6251,
+      "step": 7086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.548794562708387e-05,
+      "loss": 0.5964,
+      "step": 7087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.548730443703514e-05,
+      "loss": 0.5243,
+      "step": 7088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.548666324698641e-05,
+      "loss": 0.5087,
+      "step": 7089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5486022056937675e-05,
+      "loss": 0.6071,
+      "step": 7090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5485380866888946e-05,
+      "loss": 0.5545,
+      "step": 7091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.548473967684022e-05,
+      "loss": 0.5669,
+      "step": 7092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.548409848679149e-05,
+      "loss": 0.5792,
+      "step": 7093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5483457296742755e-05,
+      "loss": 0.5434,
+      "step": 7094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5482816106694027e-05,
+      "loss": 0.4819,
+      "step": 7095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.54821749166453e-05,
+      "loss": 0.6296,
+      "step": 7096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.548153372659656e-05,
+      "loss": 0.7193,
+      "step": 7097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5480892536547835e-05,
+      "loss": 0.5997,
+      "step": 7098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.54802513464991e-05,
+      "loss": 0.4898,
+      "step": 7099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547961015645038e-05,
+      "loss": 0.4908,
+      "step": 7100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547896896640164e-05,
+      "loss": 0.5866,
+      "step": 7101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5478327776352915e-05,
+      "loss": 0.5966,
+      "step": 7102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547768658630418e-05,
+      "loss": 0.5355,
+      "step": 7103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547704539625545e-05,
+      "loss": 0.5712,
+      "step": 7104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547640420620672e-05,
+      "loss": 0.5587,
+      "step": 7105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547576301615799e-05,
+      "loss": 0.643,
+      "step": 7106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547512182610926e-05,
+      "loss": 0.6829,
+      "step": 7107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547448063606053e-05,
+      "loss": 0.5112,
+      "step": 7108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5473839446011803e-05,
+      "loss": 0.5157,
+      "step": 7109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547319825596307e-05,
+      "loss": 0.532,
+      "step": 7110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547255706591434e-05,
+      "loss": 0.5377,
+      "step": 7111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5471915875865605e-05,
+      "loss": 0.5906,
+      "step": 7112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547127468581688e-05,
+      "loss": 0.6311,
+      "step": 7113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547063349576815e-05,
+      "loss": 0.6408,
+      "step": 7114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5469992305719413e-05,
+      "loss": 0.5885,
+      "step": 7115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546935111567069e-05,
+      "loss": 0.4992,
+      "step": 7116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546870992562196e-05,
+      "loss": 0.5379,
+      "step": 7117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546806873557323e-05,
+      "loss": 0.5671,
+      "step": 7118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5467427545524494e-05,
+      "loss": 0.6092,
+      "step": 7119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5466786355475765e-05,
+      "loss": 0.5542,
+      "step": 7120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546614516542703e-05,
+      "loss": 0.5227,
+      "step": 7121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.54655039753783e-05,
+      "loss": 0.481,
+      "step": 7122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5464862785329574e-05,
+      "loss": 0.5879,
+      "step": 7123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5464221595280845e-05,
+      "loss": 0.5372,
+      "step": 7124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546358040523212e-05,
+      "loss": 0.6259,
+      "step": 7125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546293921518338e-05,
+      "loss": 0.5625,
+      "step": 7126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5462298025134654e-05,
+      "loss": 0.6364,
+      "step": 7127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546165683508592e-05,
+      "loss": 0.6546,
+      "step": 7128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546101564503719e-05,
+      "loss": 0.5892,
+      "step": 7129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5460374454988455e-05,
+      "loss": 0.5171,
+      "step": 7130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545973326493973e-05,
+      "loss": 0.5684,
+      "step": 7131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5459092074891e-05,
+      "loss": 0.578,
+      "step": 7132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545845088484227e-05,
+      "loss": 0.5475,
+      "step": 7133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545780969479354e-05,
+      "loss": 0.525,
+      "step": 7134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545716850474481e-05,
+      "loss": 0.5252,
+      "step": 7135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545652731469608e-05,
+      "loss": 0.5376,
+      "step": 7136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5455886124647344e-05,
+      "loss": 0.6053,
+      "step": 7137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5455244934598615e-05,
+      "loss": 0.5119,
+      "step": 7138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545460374454989e-05,
+      "loss": 0.5845,
+      "step": 7139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545396255450116e-05,
+      "loss": 0.5464,
+      "step": 7140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545332136445243e-05,
+      "loss": 0.601,
+      "step": 7141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5452680174403696e-05,
+      "loss": 0.4966,
+      "step": 7142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545203898435497e-05,
+      "loss": 0.5985,
+      "step": 7143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545139779430623e-05,
+      "loss": 0.5781,
+      "step": 7144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5450756604257504e-05,
+      "loss": 0.4998,
+      "step": 7145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.545011541420877e-05,
+      "loss": 0.5469,
+      "step": 7146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.544947422416004e-05,
+      "loss": 0.5185,
+      "step": 7147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.544883303411131e-05,
+      "loss": 0.4856,
+      "step": 7148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5448191844062584e-05,
+      "loss": 0.598,
+      "step": 7149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5447550654013856e-05,
+      "loss": 0.618,
+      "step": 7150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.544690946396512e-05,
+      "loss": 0.5955,
+      "step": 7151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.544626827391639e-05,
+      "loss": 0.6584,
+      "step": 7152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544562708386766e-05,
+      "loss": 0.5028,
+      "step": 7153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544498589381893e-05,
+      "loss": 0.6866,
+      "step": 7154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54443447037702e-05,
+      "loss": 0.4972,
+      "step": 7155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544370351372147e-05,
+      "loss": 0.5695,
+      "step": 7156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5443062323672744e-05,
+      "loss": 0.5974,
+      "step": 7157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544242113362401e-05,
+      "loss": 0.5913,
+      "step": 7158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544177994357528e-05,
+      "loss": 0.6014,
+      "step": 7159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5441138753526546e-05,
+      "loss": 0.6133,
+      "step": 7160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.544049756347782e-05,
+      "loss": 0.5168,
+      "step": 7161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543985637342908e-05,
+      "loss": 0.5338,
+      "step": 7162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5439215183380354e-05,
+      "loss": 0.5688,
+      "step": 7163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5438573993331626e-05,
+      "loss": 0.5939,
+      "step": 7164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54379328032829e-05,
+      "loss": 0.5789,
+      "step": 7165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543729161323417e-05,
+      "loss": 0.6248,
+      "step": 7166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5436650423185434e-05,
+      "loss": 0.5643,
+      "step": 7167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5436009233136706e-05,
+      "loss": 0.6014,
+      "step": 7168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543536804308797e-05,
+      "loss": 0.5442,
+      "step": 7169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543472685303924e-05,
+      "loss": 0.5649,
+      "step": 7170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543408566299051e-05,
+      "loss": 0.581,
+      "step": 7171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5433444472941786e-05,
+      "loss": 0.6564,
+      "step": 7172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543280328289305e-05,
+      "loss": 0.5541,
+      "step": 7173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543216209284432e-05,
+      "loss": 0.6141,
+      "step": 7174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5431520902795594e-05,
+      "loss": 0.5035,
+      "step": 7175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543087971274686e-05,
+      "loss": 0.4537,
+      "step": 7176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543023852269813e-05,
+      "loss": 0.4102,
+      "step": 7177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5429597332649396e-05,
+      "loss": 0.5637,
+      "step": 7178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542895614260067e-05,
+      "loss": 0.5416,
+      "step": 7179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542831495255194e-05,
+      "loss": 0.6245,
+      "step": 7180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542767376250321e-05,
+      "loss": 0.5866,
+      "step": 7181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5427032572454476e-05,
+      "loss": 0.623,
+      "step": 7182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542639138240575e-05,
+      "loss": 0.594,
+      "step": 7183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542575019235702e-05,
+      "loss": 0.5881,
+      "step": 7184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5425109002308285e-05,
+      "loss": 0.543,
+      "step": 7185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5424467812259556e-05,
+      "loss": 0.4257,
+      "step": 7186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542382662221082e-05,
+      "loss": 0.5923,
+      "step": 7187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54231854321621e-05,
+      "loss": 0.5653,
+      "step": 7188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5422544242113365e-05,
+      "loss": 0.4866,
+      "step": 7189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5421903052064636e-05,
+      "loss": 0.5343,
+      "step": 7190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54212618620159e-05,
+      "loss": 0.5307,
+      "step": 7191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542062067196717e-05,
+      "loss": 0.6182,
+      "step": 7192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5419979481918445e-05,
+      "loss": 0.5837,
+      "step": 7193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541933829186971e-05,
+      "loss": 0.6543,
+      "step": 7194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541869710182098e-05,
+      "loss": 0.5315,
+      "step": 7195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541805591177225e-05,
+      "loss": 0.6447,
+      "step": 7196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5417414721723525e-05,
+      "loss": 0.598,
+      "step": 7197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541677353167479e-05,
+      "loss": 0.5991,
+      "step": 7198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541613234162606e-05,
+      "loss": 0.678,
+      "step": 7199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5415491151577326e-05,
+      "loss": 0.7154,
+      "step": 7200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54148499615286e-05,
+      "loss": 0.6278,
+      "step": 7201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541420877147987e-05,
+      "loss": 0.6056,
+      "step": 7202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5413567581431135e-05,
+      "loss": 0.5628,
+      "step": 7203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541292639138241e-05,
+      "loss": 0.5719,
+      "step": 7204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541228520133368e-05,
+      "loss": 0.4631,
+      "step": 7205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541164401128495e-05,
+      "loss": 0.6487,
+      "step": 7206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5411002821236215e-05,
+      "loss": 0.5557,
+      "step": 7207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5410361631187487e-05,
+      "loss": 0.5144,
+      "step": 7208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540972044113875e-05,
+      "loss": 0.5348,
+      "step": 7209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540907925109002e-05,
+      "loss": 0.5354,
+      "step": 7210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5408438061041295e-05,
+      "loss": 0.5238,
+      "step": 7211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540779687099257e-05,
+      "loss": 0.5949,
+      "step": 7212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540715568094384e-05,
+      "loss": 0.6535,
+      "step": 7213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54065144908951e-05,
+      "loss": 0.6067,
+      "step": 7214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5405873300846375e-05,
+      "loss": 0.5901,
+      "step": 7215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540523211079764e-05,
+      "loss": 0.5868,
+      "step": 7216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540459092074891e-05,
+      "loss": 0.7009,
+      "step": 7217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540394973070018e-05,
+      "loss": 0.5512,
+      "step": 7218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540330854065145e-05,
+      "loss": 0.6185,
+      "step": 7219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540266735060272e-05,
+      "loss": 0.5238,
+      "step": 7220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540202616055399e-05,
+      "loss": 0.5706,
+      "step": 7221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5401384970505263e-05,
+      "loss": 0.71,
+      "step": 7222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540074378045653e-05,
+      "loss": 0.5516,
+      "step": 7223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54001025904078e-05,
+      "loss": 0.6057,
+      "step": 7224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5399461400359065e-05,
+      "loss": 0.5123,
+      "step": 7225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539882021031034e-05,
+      "loss": 0.6006,
+      "step": 7226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539817902026161e-05,
+      "loss": 0.656,
+      "step": 7227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539753783021288e-05,
+      "loss": 0.4964,
+      "step": 7228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539689664016415e-05,
+      "loss": 0.6367,
+      "step": 7229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539625545011542e-05,
+      "loss": 0.5457,
+      "step": 7230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539561426006669e-05,
+      "loss": 0.5942,
+      "step": 7231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5394973070017954e-05,
+      "loss": 0.5808,
+      "step": 7232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5394331879969225e-05,
+      "loss": 0.559,
+      "step": 7233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539369068992049e-05,
+      "loss": 0.53,
+      "step": 7234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539304949987176e-05,
+      "loss": 0.6106,
+      "step": 7235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5392408309823034e-05,
+      "loss": 0.6069,
+      "step": 7236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5391767119774305e-05,
+      "loss": 0.5302,
+      "step": 7237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539112592972558e-05,
+      "loss": 0.6562,
+      "step": 7238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539048473967684e-05,
+      "loss": 0.5659,
+      "step": 7239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5389843549628114e-05,
+      "loss": 0.4893,
+      "step": 7240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538920235957938e-05,
+      "loss": 0.5064,
+      "step": 7241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538856116953065e-05,
+      "loss": 0.5024,
+      "step": 7242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538791997948192e-05,
+      "loss": 0.5183,
+      "step": 7243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5387278789433194e-05,
+      "loss": 0.6433,
+      "step": 7244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538663759938446e-05,
+      "loss": 0.5972,
+      "step": 7245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538599640933573e-05,
+      "loss": 0.5406,
+      "step": 7246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5385355219287e-05,
+      "loss": 0.5528,
+      "step": 7247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538471402923827e-05,
+      "loss": 0.6045,
+      "step": 7248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538407283918954e-05,
+      "loss": 0.608,
+      "step": 7249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5383431649140804e-05,
+      "loss": 0.5358,
+      "step": 7250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5382790459092076e-05,
+      "loss": 0.5819,
+      "step": 7251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538214926904335e-05,
+      "loss": 0.4508,
+      "step": 7252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538150807899462e-05,
+      "loss": 0.6494,
+      "step": 7253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5380866888945884e-05,
+      "loss": 0.7168,
+      "step": 7254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5380225698897156e-05,
+      "loss": 0.5483,
+      "step": 7255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537958450884843e-05,
+      "loss": 0.6005,
+      "step": 7256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537894331879969e-05,
+      "loss": 0.5309,
+      "step": 7257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5378302128750964e-05,
+      "loss": 0.7358,
+      "step": 7258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537766093870223e-05,
+      "loss": 0.5189,
+      "step": 7259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537701974865351e-05,
+      "loss": 0.5226,
+      "step": 7260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537637855860477e-05,
+      "loss": 0.5522,
+      "step": 7261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5375737368556044e-05,
+      "loss": 0.6329,
+      "step": 7262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537509617850731e-05,
+      "loss": 0.6328,
+      "step": 7263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537445498845858e-05,
+      "loss": 0.6598,
+      "step": 7264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537381379840985e-05,
+      "loss": 0.5376,
+      "step": 7265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537317260836112e-05,
+      "loss": 0.6139,
+      "step": 7266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537253141831239e-05,
+      "loss": 0.5427,
+      "step": 7267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537189022826366e-05,
+      "loss": 0.5962,
+      "step": 7268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537124903821493e-05,
+      "loss": 0.6982,
+      "step": 7269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53706078481662e-05,
+      "loss": 0.4955,
+      "step": 7270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536996665811747e-05,
+      "loss": 0.5683,
+      "step": 7271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5369325468068734e-05,
+      "loss": 0.5666,
+      "step": 7272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5368684278020006e-05,
+      "loss": 0.6161,
+      "step": 7273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536804308797128e-05,
+      "loss": 0.5775,
+      "step": 7274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536740189792254e-05,
+      "loss": 0.5483,
+      "step": 7275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536676070787382e-05,
+      "loss": 0.543,
+      "step": 7276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5366119517825086e-05,
+      "loss": 0.5874,
+      "step": 7277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536547832777636e-05,
+      "loss": 0.6751,
+      "step": 7278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536483713772762e-05,
+      "loss": 0.5656,
+      "step": 7279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5364195947678894e-05,
+      "loss": 0.5923,
+      "step": 7280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536355475763016e-05,
+      "loss": 0.5253,
+      "step": 7281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536291356758143e-05,
+      "loss": 0.53,
+      "step": 7282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53622723775327e-05,
+      "loss": 0.5556,
+      "step": 7283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5361631187483974e-05,
+      "loss": 0.5476,
+      "step": 7284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5360989997435246e-05,
+      "loss": 0.6413,
+      "step": 7285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536034880738651e-05,
+      "loss": 0.5202,
+      "step": 7286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535970761733778e-05,
+      "loss": 0.5718,
+      "step": 7287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535906642728905e-05,
+      "loss": 0.5313,
+      "step": 7288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535842523724032e-05,
+      "loss": 0.5985,
+      "step": 7289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5357784047191584e-05,
+      "loss": 0.5327,
+      "step": 7290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5357142857142856e-05,
+      "loss": 0.6899,
+      "step": 7291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535650166709413e-05,
+      "loss": 0.5384,
+      "step": 7292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53558604770454e-05,
+      "loss": 0.5043,
+      "step": 7293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535521928699667e-05,
+      "loss": 0.619,
+      "step": 7294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5354578096947936e-05,
+      "loss": 0.6472,
+      "step": 7295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535393690689921e-05,
+      "loss": 0.5948,
+      "step": 7296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535329571685047e-05,
+      "loss": 0.5814,
+      "step": 7297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5352654526801745e-05,
+      "loss": 0.5535,
+      "step": 7298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5352013336753016e-05,
+      "loss": 0.5865,
+      "step": 7299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535137214670429e-05,
+      "loss": 0.5805,
+      "step": 7300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535073095665556e-05,
+      "loss": 0.5825,
+      "step": 7301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5350089766606825e-05,
+      "loss": 0.4864,
+      "step": 7302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5349448576558096e-05,
+      "loss": 0.5393,
+      "step": 7303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534880738650936e-05,
+      "loss": 0.5416,
+      "step": 7304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534816619646063e-05,
+      "loss": 0.5713,
+      "step": 7305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53475250064119e-05,
+      "loss": 0.479,
+      "step": 7306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534688381636317e-05,
+      "loss": 0.5752,
+      "step": 7307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534624262631444e-05,
+      "loss": 0.5703,
+      "step": 7308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534560143626571e-05,
+      "loss": 0.6387,
+      "step": 7309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5344960246216985e-05,
+      "loss": 0.5797,
+      "step": 7310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534431905616825e-05,
+      "loss": 0.563,
+      "step": 7311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534367786611952e-05,
+      "loss": 0.6327,
+      "step": 7312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5343036676070786e-05,
+      "loss": 0.6546,
+      "step": 7313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534239548602206e-05,
+      "loss": 0.5625,
+      "step": 7314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534175429597333e-05,
+      "loss": 0.5712,
+      "step": 7315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53411131059246e-05,
+      "loss": 0.4828,
+      "step": 7316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5340471915875866e-05,
+      "loss": 0.6361,
+      "step": 7317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533983072582714e-05,
+      "loss": 0.5785,
+      "step": 7318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533918953577841e-05,
+      "loss": 0.6235,
+      "step": 7319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5338548345729675e-05,
+      "loss": 0.4809,
+      "step": 7320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5337907155680947e-05,
+      "loss": 0.5284,
+      "step": 7321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533726596563221e-05,
+      "loss": 0.5713,
+      "step": 7322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533662477558348e-05,
+      "loss": 0.584,
+      "step": 7323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5335983585534755e-05,
+      "loss": 0.5526,
+      "step": 7324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533534239548603e-05,
+      "loss": 0.5924,
+      "step": 7325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533470120543729e-05,
+      "loss": 0.4309,
+      "step": 7326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533406001538856e-05,
+      "loss": 0.5369,
+      "step": 7327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5333418825339835e-05,
+      "loss": 0.5124,
+      "step": 7328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53327776352911e-05,
+      "loss": 0.4848,
+      "step": 7329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533213644524237e-05,
+      "loss": 0.5628,
+      "step": 7330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5331495255193643e-05,
+      "loss": 0.6139,
+      "step": 7331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5330854065144915e-05,
+      "loss": 0.6209,
+      "step": 7332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.533021287509618e-05,
+      "loss": 0.5658,
+      "step": 7333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532957168504745e-05,
+      "loss": 0.6258,
+      "step": 7334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532893049499872e-05,
+      "loss": 0.4958,
+      "step": 7335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532828930494999e-05,
+      "loss": 0.5573,
+      "step": 7336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532764811490126e-05,
+      "loss": 0.5281,
+      "step": 7337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5327006924852525e-05,
+      "loss": 0.568,
+      "step": 7338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53263657348038e-05,
+      "loss": 0.5271,
+      "step": 7339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532572454475507e-05,
+      "loss": 0.5829,
+      "step": 7340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532508335470634e-05,
+      "loss": 0.5978,
+      "step": 7341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5324442164657605e-05,
+      "loss": 0.5254,
+      "step": 7342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532380097460888e-05,
+      "loss": 0.5207,
+      "step": 7343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532315978456015e-05,
+      "loss": 0.5954,
+      "step": 7344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5322518594511414e-05,
+      "loss": 0.5934,
+      "step": 7345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5321877404462685e-05,
+      "loss": 0.6337,
+      "step": 7346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532123621441395e-05,
+      "loss": 0.5366,
+      "step": 7347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.532059502436523e-05,
+      "loss": 0.6181,
+      "step": 7348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5319953834316494e-05,
+      "loss": 0.5372,
+      "step": 7349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5319312644267765e-05,
+      "loss": 0.5067,
+      "step": 7350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531867145421903e-05,
+      "loss": 0.566,
+      "step": 7351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53180302641703e-05,
+      "loss": 0.5917,
+      "step": 7352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5317389074121574e-05,
+      "loss": 0.5975,
+      "step": 7353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531674788407284e-05,
+      "loss": 0.6303,
+      "step": 7354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531610669402411e-05,
+      "loss": 0.6555,
+      "step": 7355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531546550397538e-05,
+      "loss": 0.5823,
+      "step": 7356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5314824313926654e-05,
+      "loss": 0.6033,
+      "step": 7357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531418312387792e-05,
+      "loss": 0.613,
+      "step": 7358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531354193382919e-05,
+      "loss": 0.5161,
+      "step": 7359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5312900743780455e-05,
+      "loss": 0.6071,
+      "step": 7360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531225955373173e-05,
+      "loss": 0.6292,
+      "step": 7361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5311618363683e-05,
+      "loss": 0.5817,
+      "step": 7362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5310977173634264e-05,
+      "loss": 0.5767,
+      "step": 7363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531033598358554e-05,
+      "loss": 0.5294,
+      "step": 7364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530969479353681e-05,
+      "loss": 0.5453,
+      "step": 7365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530905360348808e-05,
+      "loss": 0.5922,
+      "step": 7366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5308412413439344e-05,
+      "loss": 0.6194,
+      "step": 7367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5307771223390616e-05,
+      "loss": 0.5634,
+      "step": 7368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530713003334188e-05,
+      "loss": 0.4897,
+      "step": 7369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530648884329315e-05,
+      "loss": 0.588,
+      "step": 7370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5305847653244424e-05,
+      "loss": 0.5713,
+      "step": 7371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5305206463195696e-05,
+      "loss": 0.535,
+      "step": 7372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530456527314697e-05,
+      "loss": 0.5542,
+      "step": 7373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530392408309823e-05,
+      "loss": 0.4981,
+      "step": 7374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5303282893049504e-05,
+      "loss": 0.5515,
+      "step": 7375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530264170300077e-05,
+      "loss": 0.557,
+      "step": 7376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530200051295204e-05,
+      "loss": 0.575,
+      "step": 7377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5301359322903306e-05,
+      "loss": 0.6222,
+      "step": 7378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530071813285458e-05,
+      "loss": 0.4371,
+      "step": 7379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530007694280585e-05,
+      "loss": 0.6317,
+      "step": 7380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529943575275712e-05,
+      "loss": 0.6773,
+      "step": 7381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529879456270839e-05,
+      "loss": 0.5117,
+      "step": 7382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529815337265966e-05,
+      "loss": 0.5504,
+      "step": 7383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529751218261093e-05,
+      "loss": 0.526,
+      "step": 7384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5296870992562194e-05,
+      "loss": 0.5792,
+      "step": 7385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5296229802513466e-05,
+      "loss": 0.6137,
+      "step": 7386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529558861246474e-05,
+      "loss": 0.6762,
+      "step": 7387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529494742241601e-05,
+      "loss": 0.5406,
+      "step": 7388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529430623236728e-05,
+      "loss": 0.6943,
+      "step": 7389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5293665042318546e-05,
+      "loss": 0.6579,
+      "step": 7390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529302385226982e-05,
+      "loss": 0.5847,
+      "step": 7391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529238266222108e-05,
+      "loss": 0.6231,
+      "step": 7392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5291741472172354e-05,
+      "loss": 0.6056,
+      "step": 7393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529110028212362e-05,
+      "loss": 0.7549,
+      "step": 7394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.529045909207489e-05,
+      "loss": 0.5563,
+      "step": 7395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528981790202616e-05,
+      "loss": 0.603,
+      "step": 7396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5289176711977434e-05,
+      "loss": 0.634,
+      "step": 7397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5288535521928706e-05,
+      "loss": 0.5438,
+      "step": 7398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528789433187997e-05,
+      "loss": 0.5278,
+      "step": 7399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528725314183124e-05,
+      "loss": 0.5222,
+      "step": 7400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528661195178251e-05,
+      "loss": 0.5344,
+      "step": 7401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528597076173378e-05,
+      "loss": 0.6177,
+      "step": 7402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528532957168505e-05,
+      "loss": 0.6158,
+      "step": 7403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528468838163632e-05,
+      "loss": 0.5984,
+      "step": 7404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528404719158759e-05,
+      "loss": 0.5476,
+      "step": 7405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528340600153886e-05,
+      "loss": 0.5719,
+      "step": 7406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528276481149013e-05,
+      "loss": 0.6045,
+      "step": 7407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5282123621441396e-05,
+      "loss": 0.6157,
+      "step": 7408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528148243139267e-05,
+      "loss": 0.5987,
+      "step": 7409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.528084124134393e-05,
+      "loss": 0.5609,
+      "step": 7410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5280200051295205e-05,
+      "loss": 0.5997,
+      "step": 7411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5279558861246476e-05,
+      "loss": 0.5539,
+      "step": 7412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527891767119775e-05,
+      "loss": 0.5688,
+      "step": 7413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527827648114901e-05,
+      "loss": 0.6197,
+      "step": 7414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5277635291100285e-05,
+      "loss": 0.6477,
+      "step": 7415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5276994101051556e-05,
+      "loss": 0.6178,
+      "step": 7416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527635291100282e-05,
+      "loss": 0.5289,
+      "step": 7417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527571172095409e-05,
+      "loss": 0.549,
+      "step": 7418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527507053090536e-05,
+      "loss": 0.5786,
+      "step": 7419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5274429340856636e-05,
+      "loss": 0.5774,
+      "step": 7420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52737881508079e-05,
+      "loss": 0.6161,
+      "step": 7421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527314696075917e-05,
+      "loss": 0.5844,
+      "step": 7422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527250577071044e-05,
+      "loss": 0.6773,
+      "step": 7423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527186458066171e-05,
+      "loss": 0.6065,
+      "step": 7424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527122339061298e-05,
+      "loss": 0.5592,
+      "step": 7425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5270582200564246e-05,
+      "loss": 0.6306,
+      "step": 7426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526994101051552e-05,
+      "loss": 0.5115,
+      "step": 7427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526929982046679e-05,
+      "loss": 0.5315,
+      "step": 7428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526865863041806e-05,
+      "loss": 0.6457,
+      "step": 7429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5268017440369327e-05,
+      "loss": 0.576,
+      "step": 7430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52673762503206e-05,
+      "loss": 0.5489,
+      "step": 7431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526673506027186e-05,
+      "loss": 0.6051,
+      "step": 7432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5266093870223135e-05,
+      "loss": 0.6207,
+      "step": 7433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526545268017441e-05,
+      "loss": 0.5748,
+      "step": 7434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526481149012567e-05,
+      "loss": 0.6501,
+      "step": 7435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526417030007695e-05,
+      "loss": 0.5544,
+      "step": 7436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5263529110028215e-05,
+      "loss": 0.5741,
+      "step": 7437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526288791997949e-05,
+      "loss": 0.5289,
+      "step": 7438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526224672993075e-05,
+      "loss": 0.6188,
+      "step": 7439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526160553988202e-05,
+      "loss": 0.636,
+      "step": 7440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526096434983329e-05,
+      "loss": 0.607,
+      "step": 7441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526032315978456e-05,
+      "loss": 0.5807,
+      "step": 7442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525968196973583e-05,
+      "loss": 0.5293,
+      "step": 7443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5259040779687103e-05,
+      "loss": 0.6146,
+      "step": 7444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5258399589638375e-05,
+      "loss": 0.4781,
+      "step": 7445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525775839958964e-05,
+      "loss": 0.6047,
+      "step": 7446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525711720954091e-05,
+      "loss": 0.5748,
+      "step": 7447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525647601949218e-05,
+      "loss": 0.5843,
+      "step": 7448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525583482944345e-05,
+      "loss": 0.5922,
+      "step": 7449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5255193639394713e-05,
+      "loss": 0.5603,
+      "step": 7450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5254552449345985e-05,
+      "loss": 0.5082,
+      "step": 7451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5253911259297264e-05,
+      "loss": 0.5638,
+      "step": 7452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525327006924853e-05,
+      "loss": 0.6153,
+      "step": 7453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52526288791998e-05,
+      "loss": 0.5776,
+      "step": 7454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5251987689151065e-05,
+      "loss": 0.5122,
+      "step": 7455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525134649910234e-05,
+      "loss": 0.6915,
+      "step": 7456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52507053090536e-05,
+      "loss": 0.5891,
+      "step": 7457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5250064119004874e-05,
+      "loss": 0.5045,
+      "step": 7458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5249422928956145e-05,
+      "loss": 0.6013,
+      "step": 7459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524878173890742e-05,
+      "loss": 0.6201,
+      "step": 7460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524814054885869e-05,
+      "loss": 0.6604,
+      "step": 7461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5247499358809954e-05,
+      "loss": 0.5317,
+      "step": 7462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5246858168761225e-05,
+      "loss": 0.6233,
+      "step": 7463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524621697871249e-05,
+      "loss": 0.6313,
+      "step": 7464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524557578866376e-05,
+      "loss": 0.6023,
+      "step": 7465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524493459861503e-05,
+      "loss": 0.5765,
+      "step": 7466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52442934085663e-05,
+      "loss": 0.5698,
+      "step": 7467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524365221851757e-05,
+      "loss": 0.5995,
+      "step": 7468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524301102846884e-05,
+      "loss": 0.5608,
+      "step": 7469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5242369838420114e-05,
+      "loss": 0.5373,
+      "step": 7470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524172864837138e-05,
+      "loss": 0.5766,
+      "step": 7471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524108745832265e-05,
+      "loss": 0.5292,
+      "step": 7472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5240446268273915e-05,
+      "loss": 0.5876,
+      "step": 7473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523980507822519e-05,
+      "loss": 0.5386,
+      "step": 7474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523916388817646e-05,
+      "loss": 0.4582,
+      "step": 7475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523852269812773e-05,
+      "loss": 0.5699,
+      "step": 7476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5237881508078996e-05,
+      "loss": 0.6113,
+      "step": 7477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523724031803027e-05,
+      "loss": 0.6432,
+      "step": 7478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523659912798154e-05,
+      "loss": 0.5711,
+      "step": 7479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5235957937932804e-05,
+      "loss": 0.5476,
+      "step": 7480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5235316747884076e-05,
+      "loss": 0.5977,
+      "step": 7481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523467555783534e-05,
+      "loss": 0.4998,
+      "step": 7482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523403436778661e-05,
+      "loss": 0.5362,
+      "step": 7483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5233393177737884e-05,
+      "loss": 0.5864,
+      "step": 7484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5232751987689156e-05,
+      "loss": 0.532,
+      "step": 7485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523211079764042e-05,
+      "loss": 0.4993,
+      "step": 7486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523146960759169e-05,
+      "loss": 0.5607,
+      "step": 7487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5230828417542964e-05,
+      "loss": 0.5531,
+      "step": 7488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523018722749423e-05,
+      "loss": 0.5453,
+      "step": 7489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52295460374455e-05,
+      "loss": 0.6364,
+      "step": 7490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522890484739677e-05,
+      "loss": 0.5772,
+      "step": 7491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5228263657348044e-05,
+      "loss": 0.5467,
+      "step": 7492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522762246729931e-05,
+      "loss": 0.6432,
+      "step": 7493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522698127725058e-05,
+      "loss": 0.4884,
+      "step": 7494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5226340087201846e-05,
+      "loss": 0.5072,
+      "step": 7495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522569889715312e-05,
+      "loss": 0.6557,
+      "step": 7496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522505770710439e-05,
+      "loss": 0.6432,
+      "step": 7497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5224416517055654e-05,
+      "loss": 0.5682,
+      "step": 7498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5223775327006926e-05,
+      "loss": 0.7032,
+      "step": 7499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52231341369582e-05,
+      "loss": 0.6176,
+      "step": 7500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522249294690947e-05,
+      "loss": 0.4809,
+      "step": 7501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5221851756860734e-05,
+      "loss": 0.5598,
+      "step": 7502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5221210566812006e-05,
+      "loss": 0.5179,
+      "step": 7503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522056937676327e-05,
+      "loss": 0.5907,
+      "step": 7504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521992818671454e-05,
+      "loss": 0.5699,
+      "step": 7505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5219286996665814e-05,
+      "loss": 0.6174,
+      "step": 7506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521864580661708e-05,
+      "loss": 0.5661,
+      "step": 7507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521800461656836e-05,
+      "loss": 0.5422,
+      "step": 7508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521736342651962e-05,
+      "loss": 0.5729,
+      "step": 7509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5216722236470894e-05,
+      "loss": 0.5652,
+      "step": 7510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521608104642216e-05,
+      "loss": 0.5764,
+      "step": 7511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521543985637343e-05,
+      "loss": 0.6216,
+      "step": 7512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5214798666324696e-05,
+      "loss": 0.4521,
+      "step": 7513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521415747627597e-05,
+      "loss": 0.5265,
+      "step": 7514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521351628622724e-05,
+      "loss": 0.5639,
+      "step": 7515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521287509617851e-05,
+      "loss": 0.5778,
+      "step": 7516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521223390612978e-05,
+      "loss": 0.5738,
+      "step": 7517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521159271608105e-05,
+      "loss": 0.5381,
+      "step": 7518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521095152603232e-05,
+      "loss": 0.6019,
+      "step": 7519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5210310335983585e-05,
+      "loss": 0.5419,
+      "step": 7520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5209669145934856e-05,
+      "loss": 0.6869,
+      "step": 7521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520902795588612e-05,
+      "loss": 0.5857,
+      "step": 7522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520838676583739e-05,
+      "loss": 0.556,
+      "step": 7523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520774557578867e-05,
+      "loss": 0.5629,
+      "step": 7524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5207104385739936e-05,
+      "loss": 0.5838,
+      "step": 7525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520646319569121e-05,
+      "loss": 0.5421,
+      "step": 7526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520582200564247e-05,
+      "loss": 0.5193,
+      "step": 7527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5205180815593745e-05,
+      "loss": 0.5429,
+      "step": 7528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520453962554501e-05,
+      "loss": 0.6593,
+      "step": 7529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520389843549628e-05,
+      "loss": 0.5061,
+      "step": 7530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520325724544755e-05,
+      "loss": 0.5199,
+      "step": 7531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5202616055398825e-05,
+      "loss": 0.6128,
+      "step": 7532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5201974865350097e-05,
+      "loss": 0.6016,
+      "step": 7533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520133367530136e-05,
+      "loss": 0.669,
+      "step": 7534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520069248525263e-05,
+      "loss": 0.4733,
+      "step": 7535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52000512952039e-05,
+      "loss": 0.5435,
+      "step": 7536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519941010515517e-05,
+      "loss": 0.6166,
+      "step": 7537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5198768915106435e-05,
+      "loss": 0.6092,
+      "step": 7538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5198127725057706e-05,
+      "loss": 0.5631,
+      "step": 7539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519748653500898e-05,
+      "loss": 0.5731,
+      "step": 7540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519684534496025e-05,
+      "loss": 0.6107,
+      "step": 7541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519620415491152e-05,
+      "loss": 0.6498,
+      "step": 7542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5195562964862787e-05,
+      "loss": 0.6083,
+      "step": 7543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519492177481406e-05,
+      "loss": 0.5226,
+      "step": 7544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519428058476532e-05,
+      "loss": 0.5712,
+      "step": 7545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5193639394716595e-05,
+      "loss": 0.604,
+      "step": 7546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519299820466787e-05,
+      "loss": 0.541,
+      "step": 7547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519235701461914e-05,
+      "loss": 0.4961,
+      "step": 7548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519171582457041e-05,
+      "loss": 0.5905,
+      "step": 7549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5191074634521675e-05,
+      "loss": 0.5726,
+      "step": 7550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519043344447295e-05,
+      "loss": 0.4696,
+      "step": 7551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518979225442421e-05,
+      "loss": 0.553,
+      "step": 7552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5189151064375483e-05,
+      "loss": 0.4972,
+      "step": 7553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518850987432675e-05,
+      "loss": 0.681,
+      "step": 7554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518786868427802e-05,
+      "loss": 0.5728,
+      "step": 7555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518722749422929e-05,
+      "loss": 0.5606,
+      "step": 7556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5186586304180563e-05,
+      "loss": 0.5725,
+      "step": 7557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5185945114131835e-05,
+      "loss": 0.5286,
+      "step": 7558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51853039240831e-05,
+      "loss": 0.4907,
+      "step": 7559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518466273403437e-05,
+      "loss": 0.517,
+      "step": 7560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518402154398564e-05,
+      "loss": 0.5663,
+      "step": 7561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518338035393691e-05,
+      "loss": 0.5588,
+      "step": 7562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518273916388818e-05,
+      "loss": 0.6801,
+      "step": 7563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518209797383945e-05,
+      "loss": 0.5481,
+      "step": 7564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518145678379072e-05,
+      "loss": 0.5764,
+      "step": 7565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518081559374199e-05,
+      "loss": 0.5634,
+      "step": 7566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518017440369326e-05,
+      "loss": 0.5464,
+      "step": 7567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5179533213644525e-05,
+      "loss": 0.6377,
+      "step": 7568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51788920235958e-05,
+      "loss": 0.5315,
+      "step": 7569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.517825083354706e-05,
+      "loss": 0.5312,
+      "step": 7570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5177609643498334e-05,
+      "loss": 0.5794,
+      "step": 7571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5176968453449605e-05,
+      "loss": 0.5341,
+      "step": 7572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.517632726340088e-05,
+      "loss": 0.5838,
+      "step": 7573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.517568607335214e-05,
+      "loss": 0.6528,
+      "step": 7574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5175044883303414e-05,
+      "loss": 0.5832,
+      "step": 7575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5174403693254685e-05,
+      "loss": 0.5341,
+      "step": 7576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.517376250320595e-05,
+      "loss": 0.5143,
+      "step": 7577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.517312131315722e-05,
+      "loss": 0.5375,
+      "step": 7578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.517248012310849e-05,
+      "loss": 0.5811,
+      "step": 7579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5171838933059766e-05,
+      "loss": 0.5185,
+      "step": 7580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.517119774301103e-05,
+      "loss": 0.5773,
+      "step": 7581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51705565529623e-05,
+      "loss": 0.5602,
+      "step": 7582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516991536291357e-05,
+      "loss": 0.6487,
+      "step": 7583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516927417286484e-05,
+      "loss": 0.5917,
+      "step": 7584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516863298281611e-05,
+      "loss": 0.5924,
+      "step": 7585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5167991792767376e-05,
+      "loss": 0.6413,
+      "step": 7586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516735060271865e-05,
+      "loss": 0.5901,
+      "step": 7587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516670941266992e-05,
+      "loss": 0.61,
+      "step": 7588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516606822262119e-05,
+      "loss": 0.5343,
+      "step": 7589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5165427032572456e-05,
+      "loss": 0.5875,
+      "step": 7590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516478584252373e-05,
+      "loss": 0.6984,
+      "step": 7591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516414465247499e-05,
+      "loss": 0.548,
+      "step": 7592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5163503462426264e-05,
+      "loss": 0.6072,
+      "step": 7593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5162862272377536e-05,
+      "loss": 0.567,
+      "step": 7594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51622210823288e-05,
+      "loss": 0.6068,
+      "step": 7595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516157989228008e-05,
+      "loss": 0.6293,
+      "step": 7596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5160938702231344e-05,
+      "loss": 0.5836,
+      "step": 7597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5160297512182616e-05,
+      "loss": 0.5774,
+      "step": 7598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515965632213388e-05,
+      "loss": 0.5421,
+      "step": 7599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515901513208515e-05,
+      "loss": 0.6596,
+      "step": 7600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515837394203642e-05,
+      "loss": 0.5338,
+      "step": 7601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515773275198769e-05,
+      "loss": 0.5438,
+      "step": 7602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515709156193896e-05,
+      "loss": 0.6307,
+      "step": 7603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515645037189023e-05,
+      "loss": 0.6251,
+      "step": 7604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5155809181841504e-05,
+      "loss": 0.6246,
+      "step": 7605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515516799179277e-05,
+      "loss": 0.6227,
+      "step": 7606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515452680174404e-05,
+      "loss": 0.5766,
+      "step": 7607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5153885611695306e-05,
+      "loss": 0.6224,
+      "step": 7608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515324442164658e-05,
+      "loss": 0.6034,
+      "step": 7609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515260323159784e-05,
+      "loss": 0.6151,
+      "step": 7610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5151962041549114e-05,
+      "loss": 0.6615,
+      "step": 7611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515132085150039e-05,
+      "loss": 0.5753,
+      "step": 7612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515067966145166e-05,
+      "loss": 0.5646,
+      "step": 7613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515003847140293e-05,
+      "loss": 0.6294,
+      "step": 7614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5149397281354194e-05,
+      "loss": 0.5658,
+      "step": 7615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5148756091305466e-05,
+      "loss": 0.5717,
+      "step": 7616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.514811490125673e-05,
+      "loss": 0.5,
+      "step": 7617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5147473711208e-05,
+      "loss": 0.5202,
+      "step": 7618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5146832521159274e-05,
+      "loss": 0.5707,
+      "step": 7619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5146191331110546e-05,
+      "loss": 0.5886,
+      "step": 7620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.514555014106182e-05,
+      "loss": 0.6102,
+      "step": 7621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.514490895101308e-05,
+      "loss": 0.6254,
+      "step": 7622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5144267760964354e-05,
+      "loss": 0.5817,
+      "step": 7623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.514362657091562e-05,
+      "loss": 0.4662,
+      "step": 7624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.514298538086689e-05,
+      "loss": 0.6006,
+      "step": 7625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5142344190818156e-05,
+      "loss": 0.5578,
+      "step": 7626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.514170300076943e-05,
+      "loss": 0.4879,
+      "step": 7627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51410618107207e-05,
+      "loss": 0.5553,
+      "step": 7628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.514042062067197e-05,
+      "loss": 0.5595,
+      "step": 7629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513977943062324e-05,
+      "loss": 0.4845,
+      "step": 7630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513913824057451e-05,
+      "loss": 0.5403,
+      "step": 7631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513849705052578e-05,
+      "loss": 0.5465,
+      "step": 7632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5137855860477045e-05,
+      "loss": 0.5193,
+      "step": 7633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5137214670428316e-05,
+      "loss": 0.6094,
+      "step": 7634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513657348037959e-05,
+      "loss": 0.6528,
+      "step": 7635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513593229033086e-05,
+      "loss": 0.6356,
+      "step": 7636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5135291100282125e-05,
+      "loss": 0.6965,
+      "step": 7637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5134649910233396e-05,
+      "loss": 0.6417,
+      "step": 7638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513400872018467e-05,
+      "loss": 0.5453,
+      "step": 7639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513336753013593e-05,
+      "loss": 0.5627,
+      "step": 7640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5132726340087205e-05,
+      "loss": 0.5754,
+      "step": 7641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513208515003847e-05,
+      "loss": 0.5669,
+      "step": 7642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513144395998974e-05,
+      "loss": 0.5776,
+      "step": 7643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.513080276994101e-05,
+      "loss": 0.5178,
+      "step": 7644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5130161579892285e-05,
+      "loss": 0.5825,
+      "step": 7645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512952038984355e-05,
+      "loss": 0.5,
+      "step": 7646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512887919979482e-05,
+      "loss": 0.617,
+      "step": 7647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512823800974609e-05,
+      "loss": 0.6205,
+      "step": 7648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512759681969736e-05,
+      "loss": 0.5868,
+      "step": 7649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512695562964863e-05,
+      "loss": 0.5759,
+      "step": 7650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51263144395999e-05,
+      "loss": 0.5681,
+      "step": 7651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512567324955117e-05,
+      "loss": 0.6393,
+      "step": 7652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512503205950244e-05,
+      "loss": 0.5596,
+      "step": 7653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512439086945371e-05,
+      "loss": 0.5712,
+      "step": 7654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5123749679404975e-05,
+      "loss": 0.594,
+      "step": 7655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5123108489356247e-05,
+      "loss": 0.6228,
+      "step": 7656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512246729930752e-05,
+      "loss": 0.5251,
+      "step": 7657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512182610925878e-05,
+      "loss": 0.5235,
+      "step": 7658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5121184919210055e-05,
+      "loss": 0.4976,
+      "step": 7659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.512054372916133e-05,
+      "loss": 0.6316,
+      "step": 7660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51199025391126e-05,
+      "loss": 0.5936,
+      "step": 7661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511926134906386e-05,
+      "loss": 0.5319,
+      "step": 7662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5118620159015135e-05,
+      "loss": 0.5905,
+      "step": 7663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51179789689664e-05,
+      "loss": 0.5543,
+      "step": 7664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511733777891767e-05,
+      "loss": 0.5797,
+      "step": 7665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5116696588868943e-05,
+      "loss": 0.5476,
+      "step": 7666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511605539882021e-05,
+      "loss": 0.5491,
+      "step": 7667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511541420877149e-05,
+      "loss": 0.5482,
+      "step": 7668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511477301872275e-05,
+      "loss": 0.5023,
+      "step": 7669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5114131828674024e-05,
+      "loss": 0.632,
+      "step": 7670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511349063862529e-05,
+      "loss": 0.5881,
+      "step": 7671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511284944857656e-05,
+      "loss": 0.5012,
+      "step": 7672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5112208258527825e-05,
+      "loss": 0.6656,
+      "step": 7673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.51115670684791e-05,
+      "loss": 0.6535,
+      "step": 7674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.511092587843037e-05,
+      "loss": 0.5665,
+      "step": 7675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.511028468838164e-05,
+      "loss": 0.4989,
+      "step": 7676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510964349833291e-05,
+      "loss": 0.6781,
+      "step": 7677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510900230828418e-05,
+      "loss": 0.6836,
+      "step": 7678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510836111823545e-05,
+      "loss": 0.5343,
+      "step": 7679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5107719928186714e-05,
+      "loss": 0.5467,
+      "step": 7680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5107078738137985e-05,
+      "loss": 0.5971,
+      "step": 7681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510643754808925e-05,
+      "loss": 0.5982,
+      "step": 7682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510579635804052e-05,
+      "loss": 0.6139,
+      "step": 7683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.51051551679918e-05,
+      "loss": 0.4958,
+      "step": 7684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5104513977943065e-05,
+      "loss": 0.6407,
+      "step": 7685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510387278789434e-05,
+      "loss": 0.4724,
+      "step": 7686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.51032315978456e-05,
+      "loss": 0.5998,
+      "step": 7687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5102590407796874e-05,
+      "loss": 0.5989,
+      "step": 7688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510194921774814e-05,
+      "loss": 0.6534,
+      "step": 7689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510130802769941e-05,
+      "loss": 0.5418,
+      "step": 7690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510066683765068e-05,
+      "loss": 0.6211,
+      "step": 7691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5100025647601954e-05,
+      "loss": 0.4659,
+      "step": 7692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5099384457553226e-05,
+      "loss": 0.5351,
+      "step": 7693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509874326750449e-05,
+      "loss": 0.7076,
+      "step": 7694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509810207745576e-05,
+      "loss": 0.5136,
+      "step": 7695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509746088740703e-05,
+      "loss": 0.6126,
+      "step": 7696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50968196973583e-05,
+      "loss": 0.5033,
+      "step": 7697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5096178507309564e-05,
+      "loss": 0.4999,
+      "step": 7698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5095537317260836e-05,
+      "loss": 0.6423,
+      "step": 7699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509489612721211e-05,
+      "loss": 0.5322,
+      "step": 7700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509425493716338e-05,
+      "loss": 0.5587,
+      "step": 7701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509361374711465e-05,
+      "loss": 0.6017,
+      "step": 7702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5092972557065916e-05,
+      "loss": 0.6241,
+      "step": 7703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509233136701719e-05,
+      "loss": 0.6332,
+      "step": 7704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509169017696845e-05,
+      "loss": 0.587,
+      "step": 7705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5091048986919724e-05,
+      "loss": 0.5733,
+      "step": 7706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5090407796870996e-05,
+      "loss": 0.5358,
+      "step": 7707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508976660682227e-05,
+      "loss": 0.5866,
+      "step": 7708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508912541677353e-05,
+      "loss": 0.6481,
+      "step": 7709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5088484226724804e-05,
+      "loss": 0.5565,
+      "step": 7710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5087843036676076e-05,
+      "loss": 0.6222,
+      "step": 7711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508720184662734e-05,
+      "loss": 0.5636,
+      "step": 7712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508656065657861e-05,
+      "loss": 0.4953,
+      "step": 7713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508591946652988e-05,
+      "loss": 0.5767,
+      "step": 7714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508527827648115e-05,
+      "loss": 0.6457,
+      "step": 7715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508463708643242e-05,
+      "loss": 0.5432,
+      "step": 7716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508399589638369e-05,
+      "loss": 0.5706,
+      "step": 7717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5083354706334964e-05,
+      "loss": 0.5754,
+      "step": 7718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508271351628623e-05,
+      "loss": 0.6299,
+      "step": 7719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50820723262375e-05,
+      "loss": 0.5773,
+      "step": 7720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5081431136188766e-05,
+      "loss": 0.5622,
+      "step": 7721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508078994614004e-05,
+      "loss": 0.571,
+      "step": 7722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508014875609131e-05,
+      "loss": 0.5398,
+      "step": 7723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507950756604258e-05,
+      "loss": 0.547,
+      "step": 7724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5078866375993846e-05,
+      "loss": 0.6336,
+      "step": 7725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507822518594512e-05,
+      "loss": 0.7175,
+      "step": 7726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507758399589639e-05,
+      "loss": 0.6011,
+      "step": 7727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5076942805847654e-05,
+      "loss": 0.5673,
+      "step": 7728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5076301615798926e-05,
+      "loss": 0.5275,
+      "step": 7729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507566042575019e-05,
+      "loss": 0.5746,
+      "step": 7730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507501923570146e-05,
+      "loss": 0.5661,
+      "step": 7731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5074378045652734e-05,
+      "loss": 0.5965,
+      "step": 7732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5073736855604006e-05,
+      "loss": 0.5797,
+      "step": 7733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507309566555527e-05,
+      "loss": 0.5534,
+      "step": 7734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507245447550654e-05,
+      "loss": 0.5518,
+      "step": 7735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5071813285457815e-05,
+      "loss": 0.5815,
+      "step": 7736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507117209540908e-05,
+      "loss": 0.574,
+      "step": 7737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507053090536035e-05,
+      "loss": 0.608,
+      "step": 7738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506988971531162e-05,
+      "loss": 0.6093,
+      "step": 7739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5069248525262895e-05,
+      "loss": 0.6456,
+      "step": 7740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506860733521416e-05,
+      "loss": 0.6949,
+      "step": 7741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506796614516543e-05,
+      "loss": 0.5891,
+      "step": 7742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5067324955116696e-05,
+      "loss": 0.6156,
+      "step": 7743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506668376506797e-05,
+      "loss": 0.5391,
+      "step": 7744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506604257501924e-05,
+      "loss": 0.6012,
+      "step": 7745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5065401384970505e-05,
+      "loss": 0.5361,
+      "step": 7746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5064760194921776e-05,
+      "loss": 0.6679,
+      "step": 7747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506411900487305e-05,
+      "loss": 0.5385,
+      "step": 7748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506347781482432e-05,
+      "loss": 0.5495,
+      "step": 7749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5062836624775585e-05,
+      "loss": 0.6316,
+      "step": 7750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5062195434726856e-05,
+      "loss": 0.5137,
+      "step": 7751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506155424467812e-05,
+      "loss": 0.6002,
+      "step": 7752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506091305462939e-05,
+      "loss": 0.564,
+      "step": 7753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5060271864580665e-05,
+      "loss": 0.6313,
+      "step": 7754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505963067453193e-05,
+      "loss": 0.5934,
+      "step": 7755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505898948448321e-05,
+      "loss": 0.5516,
+      "step": 7756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505834829443447e-05,
+      "loss": 0.5273,
+      "step": 7757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5057707104385745e-05,
+      "loss": 0.5311,
+      "step": 7758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505706591433701e-05,
+      "loss": 0.642,
+      "step": 7759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505642472428828e-05,
+      "loss": 0.5537,
+      "step": 7760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5055783534239546e-05,
+      "loss": 0.5501,
+      "step": 7761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505514234419082e-05,
+      "loss": 0.5207,
+      "step": 7762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505450115414209e-05,
+      "loss": 0.5477,
+      "step": 7763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505385996409336e-05,
+      "loss": 0.5729,
+      "step": 7764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505321877404463e-05,
+      "loss": 0.556,
+      "step": 7765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50525775839959e-05,
+      "loss": 0.5966,
+      "step": 7766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505193639394717e-05,
+      "loss": 0.4895,
+      "step": 7767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5051295203898435e-05,
+      "loss": 0.5299,
+      "step": 7768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505065401384971e-05,
+      "loss": 0.5438,
+      "step": 7769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505001282380097e-05,
+      "loss": 0.6141,
+      "step": 7770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504937163375224e-05,
+      "loss": 0.4537,
+      "step": 7771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504873044370352e-05,
+      "loss": 0.6299,
+      "step": 7772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504808925365479e-05,
+      "loss": 0.5838,
+      "step": 7773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504744806360606e-05,
+      "loss": 0.592,
+      "step": 7774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504680687355732e-05,
+      "loss": 0.6139,
+      "step": 7775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5046165683508595e-05,
+      "loss": 0.553,
+      "step": 7776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504552449345986e-05,
+      "loss": 0.6087,
+      "step": 7777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504488330341113e-05,
+      "loss": 0.638,
+      "step": 7778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5044242113362403e-05,
+      "loss": 0.5516,
+      "step": 7779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5043600923313675e-05,
+      "loss": 0.5997,
+      "step": 7780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504295973326495e-05,
+      "loss": 0.5116,
+      "step": 7781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504231854321621e-05,
+      "loss": 0.5588,
+      "step": 7782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5041677353167484e-05,
+      "loss": 0.5877,
+      "step": 7783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504103616311875e-05,
+      "loss": 0.5459,
+      "step": 7784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504039497307002e-05,
+      "loss": 0.5099,
+      "step": 7785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5039753783021285e-05,
+      "loss": 0.6083,
+      "step": 7786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503911259297256e-05,
+      "loss": 0.5647,
+      "step": 7787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503847140292383e-05,
+      "loss": 0.4622,
+      "step": 7788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50378302128751e-05,
+      "loss": 0.5593,
+      "step": 7789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503718902282637e-05,
+      "loss": 0.6578,
+      "step": 7790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503654783277764e-05,
+      "loss": 0.5401,
+      "step": 7791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503590664272891e-05,
+      "loss": 0.4737,
+      "step": 7792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5035265452680174e-05,
+      "loss": 0.6197,
+      "step": 7793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5034624262631445e-05,
+      "loss": 0.6404,
+      "step": 7794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503398307258272e-05,
+      "loss": 0.5915,
+      "step": 7795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503334188253399e-05,
+      "loss": 0.5214,
+      "step": 7796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5032700692485254e-05,
+      "loss": 0.5813,
+      "step": 7797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5032059502436525e-05,
+      "loss": 0.5565,
+      "step": 7798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50314183123878e-05,
+      "loss": 0.5271,
+      "step": 7799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503077712233906e-05,
+      "loss": 0.5244,
+      "step": 7800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5030135932290334e-05,
+      "loss": 0.5099,
+      "step": 7801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50294947422416e-05,
+      "loss": 0.6401,
+      "step": 7802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502885355219287e-05,
+      "loss": 0.5556,
+      "step": 7803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502821236214414e-05,
+      "loss": 0.5349,
+      "step": 7804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5027571172095414e-05,
+      "loss": 0.5534,
+      "step": 7805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502692998204668e-05,
+      "loss": 0.6423,
+      "step": 7806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502628879199795e-05,
+      "loss": 0.5945,
+      "step": 7807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502564760194922e-05,
+      "loss": 0.5892,
+      "step": 7808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502500641190049e-05,
+      "loss": 0.6572,
+      "step": 7809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502436522185176e-05,
+      "loss": 0.5986,
+      "step": 7810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502372403180303e-05,
+      "loss": 0.5458,
+      "step": 7811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50230828417543e-05,
+      "loss": 0.4454,
+      "step": 7812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502244165170557e-05,
+      "loss": 0.6451,
+      "step": 7813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502180046165684e-05,
+      "loss": 0.5116,
+      "step": 7814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5021159271608104e-05,
+      "loss": 0.583,
+      "step": 7815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5020518081559376e-05,
+      "loss": 0.6309,
+      "step": 7816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501987689151065e-05,
+      "loss": 0.5686,
+      "step": 7817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501923570146191e-05,
+      "loss": 0.6243,
+      "step": 7818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5018594511413184e-05,
+      "loss": 0.6289,
+      "step": 7819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5017953321364456e-05,
+      "loss": 0.6032,
+      "step": 7820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501731213131573e-05,
+      "loss": 0.5682,
+      "step": 7821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501667094126699e-05,
+      "loss": 0.5908,
+      "step": 7822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5016029751218264e-05,
+      "loss": 0.573,
+      "step": 7823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501538856116953e-05,
+      "loss": 0.5341,
+      "step": 7824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50147473711208e-05,
+      "loss": 0.6375,
+      "step": 7825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501410618107207e-05,
+      "loss": 0.6236,
+      "step": 7826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501346499102334e-05,
+      "loss": 0.641,
+      "step": 7827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5012823800974616e-05,
+      "loss": 0.5287,
+      "step": 7828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501218261092588e-05,
+      "loss": 0.5714,
+      "step": 7829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501154142087715e-05,
+      "loss": 0.5708,
+      "step": 7830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501090023082842e-05,
+      "loss": 0.6054,
+      "step": 7831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501025904077969e-05,
+      "loss": 0.5814,
+      "step": 7832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5009617850730954e-05,
+      "loss": 0.5378,
+      "step": 7833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5008976660682226e-05,
+      "loss": 0.556,
+      "step": 7834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50083354706335e-05,
+      "loss": 0.5395,
+      "step": 7835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.500769428058477e-05,
+      "loss": 0.5419,
+      "step": 7836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.500705309053604e-05,
+      "loss": 0.5674,
+      "step": 7837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5006411900487306e-05,
+      "loss": 0.5612,
+      "step": 7838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.500577071043858e-05,
+      "loss": 0.5707,
+      "step": 7839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.500512952038984e-05,
+      "loss": 0.5864,
+      "step": 7840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5004488330341114e-05,
+      "loss": 0.6316,
+      "step": 7841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.500384714029238e-05,
+      "loss": 0.5821,
+      "step": 7842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.500320595024365e-05,
+      "loss": 0.601,
+      "step": 7843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.500256476019493e-05,
+      "loss": 0.6437,
+      "step": 7844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5001923570146194e-05,
+      "loss": 0.4942,
+      "step": 7845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5001282380097466e-05,
+      "loss": 0.5789,
+      "step": 7846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.500064119004873e-05,
+      "loss": 0.6079,
+      "step": 7847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5e-05,
+      "loss": 0.506,
+      "step": 7848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499935880995127e-05,
+      "loss": 0.4994,
+      "step": 7849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499871761990254e-05,
+      "loss": 0.6539,
+      "step": 7850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499807642985381e-05,
+      "loss": 0.5337,
+      "step": 7851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499743523980508e-05,
+      "loss": 0.5801,
+      "step": 7852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4996794049756355e-05,
+      "loss": 0.5868,
+      "step": 7853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499615285970762e-05,
+      "loss": 0.4864,
+      "step": 7854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499551166965889e-05,
+      "loss": 0.6435,
+      "step": 7855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4994870479610156e-05,
+      "loss": 0.612,
+      "step": 7856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499422928956143e-05,
+      "loss": 0.6401,
+      "step": 7857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499358809951269e-05,
+      "loss": 0.5703,
+      "step": 7858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4992946909463965e-05,
+      "loss": 0.5394,
+      "step": 7859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4992305719415236e-05,
+      "loss": 0.5609,
+      "step": 7860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499166452936651e-05,
+      "loss": 0.5253,
+      "step": 7861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499102333931778e-05,
+      "loss": 0.5343,
+      "step": 7862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4990382149269045e-05,
+      "loss": 0.6038,
+      "step": 7863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4989740959220316e-05,
+      "loss": 0.6487,
+      "step": 7864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498909976917158e-05,
+      "loss": 0.5842,
+      "step": 7865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498845857912285e-05,
+      "loss": 0.601,
+      "step": 7866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4987817389074125e-05,
+      "loss": 0.6425,
+      "step": 7867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4987176199025397e-05,
+      "loss": 0.6695,
+      "step": 7868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498653500897666e-05,
+      "loss": 0.5942,
+      "step": 7869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498589381892793e-05,
+      "loss": 0.6026,
+      "step": 7870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4985252628879205e-05,
+      "loss": 0.5981,
+      "step": 7871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498461143883047e-05,
+      "loss": 0.5804,
+      "step": 7872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498397024878174e-05,
+      "loss": 0.5739,
+      "step": 7873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4983329058733006e-05,
+      "loss": 0.5229,
+      "step": 7874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498268786868428e-05,
+      "loss": 0.5213,
+      "step": 7875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498204667863555e-05,
+      "loss": 0.5447,
+      "step": 7876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498140548858682e-05,
+      "loss": 0.555,
+      "step": 7877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4980764298538087e-05,
+      "loss": 0.7162,
+      "step": 7878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498012310848936e-05,
+      "loss": 0.6087,
+      "step": 7879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497948191844063e-05,
+      "loss": 0.6668,
+      "step": 7880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4978840728391895e-05,
+      "loss": 0.5863,
+      "step": 7881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497819953834317e-05,
+      "loss": 0.5789,
+      "step": 7882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497755834829444e-05,
+      "loss": 0.7341,
+      "step": 7883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497691715824571e-05,
+      "loss": 0.6407,
+      "step": 7884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4976275968196975e-05,
+      "loss": 0.6117,
+      "step": 7885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497563477814825e-05,
+      "loss": 0.609,
+      "step": 7886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497499358809951e-05,
+      "loss": 0.5364,
+      "step": 7887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4974352398050783e-05,
+      "loss": 0.57,
+      "step": 7888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4973711208002055e-05,
+      "loss": 0.5432,
+      "step": 7889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497307001795332e-05,
+      "loss": 0.6147,
+      "step": 7890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.497242882790459e-05,
+      "loss": 0.5315,
+      "step": 7891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4971787637855863e-05,
+      "loss": 0.5839,
+      "step": 7892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4971146447807135e-05,
+      "loss": 0.6723,
+      "step": 7893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.49705052577584e-05,
+      "loss": 0.5123,
+      "step": 7894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496986406770967e-05,
+      "loss": 0.6027,
+      "step": 7895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496922287766094e-05,
+      "loss": 0.6591,
+      "step": 7896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496858168761221e-05,
+      "loss": 0.5579,
+      "step": 7897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496794049756348e-05,
+      "loss": 0.5352,
+      "step": 7898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496729930751475e-05,
+      "loss": 0.5487,
+      "step": 7899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4966658117466024e-05,
+      "loss": 0.5274,
+      "step": 7900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496601692741729e-05,
+      "loss": 0.7005,
+      "step": 7901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496537573736856e-05,
+      "loss": 0.6054,
+      "step": 7902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4964734547319825e-05,
+      "loss": 0.5386,
+      "step": 7903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.49640933572711e-05,
+      "loss": 0.5626,
+      "step": 7904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496345216722237e-05,
+      "loss": 0.5553,
+      "step": 7905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4962810977173634e-05,
+      "loss": 0.6202,
+      "step": 7906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4962169787124905e-05,
+      "loss": 0.5881,
+      "step": 7907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496152859707618e-05,
+      "loss": 0.4908,
+      "step": 7908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496088740702745e-05,
+      "loss": 0.5443,
+      "step": 7909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4960246216978714e-05,
+      "loss": 0.5562,
+      "step": 7910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4959605026929985e-05,
+      "loss": 0.5983,
+      "step": 7911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495896383688125e-05,
+      "loss": 0.5698,
+      "step": 7912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495832264683252e-05,
+      "loss": 0.6393,
+      "step": 7913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4957681456783794e-05,
+      "loss": 0.5049,
+      "step": 7914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495704026673506e-05,
+      "loss": 0.575,
+      "step": 7915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495639907668634e-05,
+      "loss": 0.6949,
+      "step": 7916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.49557578866376e-05,
+      "loss": 0.5606,
+      "step": 7917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4955116696588874e-05,
+      "loss": 0.5794,
+      "step": 7918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495447550654014e-05,
+      "loss": 0.5683,
+      "step": 7919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495383431649141e-05,
+      "loss": 0.5046,
+      "step": 7920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4953193126442676e-05,
+      "loss": 0.4995,
+      "step": 7921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495255193639395e-05,
+      "loss": 0.6432,
+      "step": 7922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495191074634522e-05,
+      "loss": 0.5529,
+      "step": 7923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495126955629649e-05,
+      "loss": 0.6133,
+      "step": 7924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.495062836624776e-05,
+      "loss": 0.5728,
+      "step": 7925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.494998717619903e-05,
+      "loss": 0.5106,
+      "step": 7926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.49493459861503e-05,
+      "loss": 0.5407,
+      "step": 7927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4948704796101564e-05,
+      "loss": 0.4948,
+      "step": 7928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4948063606052836e-05,
+      "loss": 0.5528,
+      "step": 7929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.49474224160041e-05,
+      "loss": 0.6755,
+      "step": 7930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.494678122595537e-05,
+      "loss": 0.6352,
+      "step": 7931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.494614003590665e-05,
+      "loss": 0.6138,
+      "step": 7932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4945498845857916e-05,
+      "loss": 0.5008,
+      "step": 7933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494485765580919e-05,
+      "loss": 0.6474,
+      "step": 7934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494421646576045e-05,
+      "loss": 0.582,
+      "step": 7935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4943575275711724e-05,
+      "loss": 0.4619,
+      "step": 7936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494293408566299e-05,
+      "loss": 0.5617,
+      "step": 7937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494229289561426e-05,
+      "loss": 0.5488,
+      "step": 7938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494165170556553e-05,
+      "loss": 0.6317,
+      "step": 7939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4941010515516804e-05,
+      "loss": 0.5418,
+      "step": 7940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4940369325468076e-05,
+      "loss": 0.604,
+      "step": 7941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493972813541934e-05,
+      "loss": 0.5178,
+      "step": 7942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493908694537061e-05,
+      "loss": 0.6271,
+      "step": 7943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493844575532188e-05,
+      "loss": 0.5517,
+      "step": 7944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493780456527315e-05,
+      "loss": 0.5636,
+      "step": 7945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4937163375224414e-05,
+      "loss": 0.6249,
+      "step": 7946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4936522185175686e-05,
+      "loss": 0.5454,
+      "step": 7947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493588099512696e-05,
+      "loss": 0.6215,
+      "step": 7948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493523980507823e-05,
+      "loss": 0.577,
+      "step": 7949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.49345986150295e-05,
+      "loss": 0.5921,
+      "step": 7950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4933957424980766e-05,
+      "loss": 0.5934,
+      "step": 7951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493331623493204e-05,
+      "loss": 0.6501,
+      "step": 7952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.49326750448833e-05,
+      "loss": 0.5895,
+      "step": 7953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4932033854834574e-05,
+      "loss": 0.644,
+      "step": 7954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4931392664785846e-05,
+      "loss": 0.5604,
+      "step": 7955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493075147473712e-05,
+      "loss": 0.5724,
+      "step": 7956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493011028468838e-05,
+      "loss": 0.5848,
+      "step": 7957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4929469094639654e-05,
+      "loss": 0.5728,
+      "step": 7958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4928827904590926e-05,
+      "loss": 0.6116,
+      "step": 7959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492818671454219e-05,
+      "loss": 0.5341,
+      "step": 7960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492754552449346e-05,
+      "loss": 0.6541,
+      "step": 7961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492690433444473e-05,
+      "loss": 0.6302,
+      "step": 7962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4926263144396e-05,
+      "loss": 0.52,
+      "step": 7963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492562195434727e-05,
+      "loss": 0.5998,
+      "step": 7964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492498076429854e-05,
+      "loss": 0.6483,
+      "step": 7965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492433957424981e-05,
+      "loss": 0.6551,
+      "step": 7966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492369838420108e-05,
+      "loss": 0.5273,
+      "step": 7967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492305719415235e-05,
+      "loss": 0.5429,
+      "step": 7968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4922416004103616e-05,
+      "loss": 0.5496,
+      "step": 7969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492177481405489e-05,
+      "loss": 0.6348,
+      "step": 7970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492113362400616e-05,
+      "loss": 0.5014,
+      "step": 7971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492049243395743e-05,
+      "loss": 0.5445,
+      "step": 7972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4919851243908696e-05,
+      "loss": 0.6208,
+      "step": 7973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491921005385997e-05,
+      "loss": 0.5268,
+      "step": 7974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491856886381123e-05,
+      "loss": 0.5489,
+      "step": 7975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4917927673762505e-05,
+      "loss": 0.6103,
+      "step": 7976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4917286483713776e-05,
+      "loss": 0.5239,
+      "step": 7977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491664529366504e-05,
+      "loss": 0.6258,
+      "step": 7978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491600410361631e-05,
+      "loss": 0.4866,
+      "step": 7979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4915362913567585e-05,
+      "loss": 0.5275,
+      "step": 7980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4914721723518857e-05,
+      "loss": 0.5947,
+      "step": 7981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491408053347012e-05,
+      "loss": 0.5141,
+      "step": 7982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491343934342139e-05,
+      "loss": 0.6188,
+      "step": 7983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491279815337266e-05,
+      "loss": 0.5082,
+      "step": 7984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491215696332393e-05,
+      "loss": 0.5365,
+      "step": 7985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.49115157732752e-05,
+      "loss": 0.5742,
+      "step": 7986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491087458322647e-05,
+      "loss": 0.4871,
+      "step": 7987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4910233393177745e-05,
+      "loss": 0.6408,
+      "step": 7988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490959220312901e-05,
+      "loss": 0.5577,
+      "step": 7989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490895101308028e-05,
+      "loss": 0.5956,
+      "step": 7990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4908309823031547e-05,
+      "loss": 0.633,
+      "step": 7991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490766863298282e-05,
+      "loss": 0.6256,
+      "step": 7992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490702744293408e-05,
+      "loss": 0.5414,
+      "step": 7993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4906386252885355e-05,
+      "loss": 0.5722,
+      "step": 7994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490574506283663e-05,
+      "loss": 0.5787,
+      "step": 7995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.49051038727879e-05,
+      "loss": 0.652,
+      "step": 7996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490446268273917e-05,
+      "loss": 0.4981,
+      "step": 7997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4903821492690435e-05,
+      "loss": 0.6042,
+      "step": 7998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490318030264171e-05,
+      "loss": 0.5907,
+      "step": 7999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490253911259297e-05,
+      "loss": 0.6757,
+      "step": 8000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4901897922544243e-05,
+      "loss": 0.6341,
+      "step": 8001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490125673249551e-05,
+      "loss": 0.5404,
+      "step": 8002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490061554244678e-05,
+      "loss": 0.6753,
+      "step": 8003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489997435239806e-05,
+      "loss": 0.6099,
+      "step": 8004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4899333162349324e-05,
+      "loss": 0.6006,
+      "step": 8005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4898691972300595e-05,
+      "loss": 0.4657,
+      "step": 8006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489805078225186e-05,
+      "loss": 0.5104,
+      "step": 8007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489740959220313e-05,
+      "loss": 0.5161,
+      "step": 8008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48967684021544e-05,
+      "loss": 0.6295,
+      "step": 8009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489612721210567e-05,
+      "loss": 0.5003,
+      "step": 8010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489548602205694e-05,
+      "loss": 0.5675,
+      "step": 8011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489484483200821e-05,
+      "loss": 0.6212,
+      "step": 8012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4894203641959484e-05,
+      "loss": 0.54,
+      "step": 8013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489356245191075e-05,
+      "loss": 0.6086,
+      "step": 8014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489292126186202e-05,
+      "loss": 0.4813,
+      "step": 8015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4892280071813285e-05,
+      "loss": 0.563,
+      "step": 8016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489163888176456e-05,
+      "loss": 0.5371,
+      "step": 8017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489099769171582e-05,
+      "loss": 0.5178,
+      "step": 8018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4890356501667094e-05,
+      "loss": 0.5464,
+      "step": 8019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4889715311618365e-05,
+      "loss": 0.6123,
+      "step": 8020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488907412156964e-05,
+      "loss": 0.6079,
+      "step": 8021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488843293152091e-05,
+      "loss": 0.7197,
+      "step": 8022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4887791741472174e-05,
+      "loss": 0.5317,
+      "step": 8023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4887150551423445e-05,
+      "loss": 0.6835,
+      "step": 8024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488650936137471e-05,
+      "loss": 0.5951,
+      "step": 8025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488586817132598e-05,
+      "loss": 0.571,
+      "step": 8026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4885226981277254e-05,
+      "loss": 0.6278,
+      "step": 8027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4884585791228526e-05,
+      "loss": 0.4994,
+      "step": 8028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488394460117979e-05,
+      "loss": 0.6269,
+      "step": 8029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488330341113106e-05,
+      "loss": 0.6056,
+      "step": 8030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4882662221082334e-05,
+      "loss": 0.6192,
+      "step": 8031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48820210310336e-05,
+      "loss": 0.5737,
+      "step": 8032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488137984098487e-05,
+      "loss": 0.6227,
+      "step": 8033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4880738650936136e-05,
+      "loss": 0.5843,
+      "step": 8034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488009746088741e-05,
+      "loss": 0.6222,
+      "step": 8035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487945627083868e-05,
+      "loss": 0.5366,
+      "step": 8036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487881508078995e-05,
+      "loss": 0.593,
+      "step": 8037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4878173890741216e-05,
+      "loss": 0.5774,
+      "step": 8038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487753270069249e-05,
+      "loss": 0.5598,
+      "step": 8039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487689151064376e-05,
+      "loss": 0.5484,
+      "step": 8040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4876250320595024e-05,
+      "loss": 0.598,
+      "step": 8041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4875609130546296e-05,
+      "loss": 0.5789,
+      "step": 8042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487496794049757e-05,
+      "loss": 0.5757,
+      "step": 8043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487432675044884e-05,
+      "loss": 0.5971,
+      "step": 8044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4873685560400104e-05,
+      "loss": 0.4949,
+      "step": 8045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4873044370351376e-05,
+      "loss": 0.5898,
+      "step": 8046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487240318030264e-05,
+      "loss": 0.4917,
+      "step": 8047
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487176199025391e-05,
+      "loss": 0.5574,
+      "step": 8048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4871120800205184e-05,
+      "loss": 0.584,
+      "step": 8049
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487047961015645e-05,
+      "loss": 0.5325,
+      "step": 8050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486983842010772e-05,
+      "loss": 0.5479,
+      "step": 8051
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486919723005899e-05,
+      "loss": 0.6836,
+      "step": 8052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4868556040010264e-05,
+      "loss": 0.6763,
+      "step": 8053
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486791484996153e-05,
+      "loss": 0.5854,
+      "step": 8054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48672736599128e-05,
+      "loss": 0.5723,
+      "step": 8055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4866632469864066e-05,
+      "loss": 0.4701,
+      "step": 8056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486599127981534e-05,
+      "loss": 0.6007,
+      "step": 8057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486535008976661e-05,
+      "loss": 0.516,
+      "step": 8058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486470889971788e-05,
+      "loss": 0.5491,
+      "step": 8059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486406770966915e-05,
+      "loss": 0.7217,
+      "step": 8060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486342651962042e-05,
+      "loss": 0.6107,
+      "step": 8061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486278532957169e-05,
+      "loss": 0.4746,
+      "step": 8062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4862144139522954e-05,
+      "loss": 0.6077,
+      "step": 8063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4861502949474226e-05,
+      "loss": 0.5666,
+      "step": 8064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486086175942549e-05,
+      "loss": 0.5012,
+      "step": 8065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486022056937676e-05,
+      "loss": 0.5418,
+      "step": 8066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4859579379328034e-05,
+      "loss": 0.5232,
+      "step": 8067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4858938189279306e-05,
+      "loss": 0.4972,
+      "step": 8068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.485829699923058e-05,
+      "loss": 0.5056,
+      "step": 8069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.485765580918184e-05,
+      "loss": 0.545,
+      "step": 8070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4857014619133115e-05,
+      "loss": 0.5964,
+      "step": 8071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.485637342908438e-05,
+      "loss": 0.5258,
+      "step": 8072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.485573223903565e-05,
+      "loss": 0.5935,
+      "step": 8073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4855091048986916e-05,
+      "loss": 0.5552,
+      "step": 8074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.485444985893819e-05,
+      "loss": 0.5373,
+      "step": 8075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4853808668889466e-05,
+      "loss": 0.5589,
+      "step": 8076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.485316747884073e-05,
+      "loss": 0.5742,
+      "step": 8077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4852526288792e-05,
+      "loss": 0.6514,
+      "step": 8078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.485188509874327e-05,
+      "loss": 0.6942,
+      "step": 8079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.485124390869454e-05,
+      "loss": 0.5813,
+      "step": 8080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4850602718645805e-05,
+      "loss": 0.5412,
+      "step": 8081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4849961528597076e-05,
+      "loss": 0.5254,
+      "step": 8082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484932033854835e-05,
+      "loss": 0.5163,
+      "step": 8083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484867914849962e-05,
+      "loss": 0.6308,
+      "step": 8084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484803795845089e-05,
+      "loss": 0.604,
+      "step": 8085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4847396768402156e-05,
+      "loss": 0.6547,
+      "step": 8086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484675557835343e-05,
+      "loss": 0.5074,
+      "step": 8087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484611438830469e-05,
+      "loss": 0.5178,
+      "step": 8088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4845473198255965e-05,
+      "loss": 0.6542,
+      "step": 8089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484483200820723e-05,
+      "loss": 0.5826,
+      "step": 8090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48441908181585e-05,
+      "loss": 0.5467,
+      "step": 8091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484354962810977e-05,
+      "loss": 0.6082,
+      "step": 8092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4842908438061045e-05,
+      "loss": 0.6188,
+      "step": 8093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4842267248012317e-05,
+      "loss": 0.6089,
+      "step": 8094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484162605796358e-05,
+      "loss": 0.5371,
+      "step": 8095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484098486791485e-05,
+      "loss": 0.5451,
+      "step": 8096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484034367786612e-05,
+      "loss": 0.5527,
+      "step": 8097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483970248781739e-05,
+      "loss": 0.5771,
+      "step": 8098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483906129776866e-05,
+      "loss": 0.5769,
+      "step": 8099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483842010771993e-05,
+      "loss": 0.6203,
+      "step": 8100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4837778917671205e-05,
+      "loss": 0.5271,
+      "step": 8101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483713772762247e-05,
+      "loss": 0.5474,
+      "step": 8102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483649653757374e-05,
+      "loss": 0.6173,
+      "step": 8103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483585534752501e-05,
+      "loss": 0.6387,
+      "step": 8104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483521415747628e-05,
+      "loss": 0.6678,
+      "step": 8105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483457296742754e-05,
+      "loss": 0.5991,
+      "step": 8106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4833931777378815e-05,
+      "loss": 0.4804,
+      "step": 8107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483329058733009e-05,
+      "loss": 0.5481,
+      "step": 8108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483264939728136e-05,
+      "loss": 0.5444,
+      "step": 8109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483200820723263e-05,
+      "loss": 0.5514,
+      "step": 8110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4831367017183895e-05,
+      "loss": 0.5943,
+      "step": 8111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483072582713517e-05,
+      "loss": 0.6629,
+      "step": 8112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483008463708643e-05,
+      "loss": 0.5644,
+      "step": 8113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4829443447037703e-05,
+      "loss": 0.5942,
+      "step": 8114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4828802256988975e-05,
+      "loss": 0.6698,
+      "step": 8115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482816106694025e-05,
+      "loss": 0.4247,
+      "step": 8116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482751987689151e-05,
+      "loss": 0.5993,
+      "step": 8117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4826878686842784e-05,
+      "loss": 0.5871,
+      "step": 8118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4826237496794055e-05,
+      "loss": 0.5905,
+      "step": 8119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482559630674532e-05,
+      "loss": 0.6865,
+      "step": 8120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482495511669659e-05,
+      "loss": 0.5768,
+      "step": 8121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482431392664786e-05,
+      "loss": 0.5253,
+      "step": 8122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482367273659913e-05,
+      "loss": 0.5333,
+      "step": 8123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48230315465504e-05,
+      "loss": 0.5559,
+      "step": 8124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482239035650167e-05,
+      "loss": 0.5565,
+      "step": 8125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482174916645294e-05,
+      "loss": 0.5857,
+      "step": 8126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482110797640421e-05,
+      "loss": 0.5733,
+      "step": 8127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.482046678635548e-05,
+      "loss": 0.611,
+      "step": 8128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4819825596306745e-05,
+      "loss": 0.6061,
+      "step": 8129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481918440625802e-05,
+      "loss": 0.6762,
+      "step": 8130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481854321620929e-05,
+      "loss": 0.5641,
+      "step": 8131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481790202616056e-05,
+      "loss": 0.6341,
+      "step": 8132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4817260836111825e-05,
+      "loss": 0.5858,
+      "step": 8133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48166196460631e-05,
+      "loss": 0.4921,
+      "step": 8134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481597845601436e-05,
+      "loss": 0.521,
+      "step": 8135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4815337265965634e-05,
+      "loss": 0.4781,
+      "step": 8136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4814696075916906e-05,
+      "loss": 0.5219,
+      "step": 8137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481405488586817e-05,
+      "loss": 0.5842,
+      "step": 8138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481341369581944e-05,
+      "loss": 0.5746,
+      "step": 8139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4812772505770714e-05,
+      "loss": 0.6069,
+      "step": 8140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4812131315721986e-05,
+      "loss": 0.4972,
+      "step": 8141
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481149012567325e-05,
+      "loss": 0.5348,
+      "step": 8142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481084893562452e-05,
+      "loss": 0.5891,
+      "step": 8143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.481020774557579e-05,
+      "loss": 0.5876,
+      "step": 8144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.480956655552706e-05,
+      "loss": 0.5551,
+      "step": 8145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.480892536547833e-05,
+      "loss": 0.5624,
+      "step": 8146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48082841754296e-05,
+      "loss": 0.6297,
+      "step": 8147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4807642985380874e-05,
+      "loss": 0.5742,
+      "step": 8148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.480700179533214e-05,
+      "loss": 0.4821,
+      "step": 8149
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.480636060528341e-05,
+      "loss": 0.6965,
+      "step": 8150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4805719415234676e-05,
+      "loss": 0.5739,
+      "step": 8151
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.480507822518595e-05,
+      "loss": 0.6721,
+      "step": 8152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.480443703513721e-05,
+      "loss": 0.5875,
+      "step": 8153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4803795845088484e-05,
+      "loss": 0.569,
+      "step": 8154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4803154655039756e-05,
+      "loss": 0.6399,
+      "step": 8155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.480251346499103e-05,
+      "loss": 0.5666,
+      "step": 8156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48018722749423e-05,
+      "loss": 0.5987,
+      "step": 8157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4801231084893564e-05,
+      "loss": 0.6114,
+      "step": 8158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4800589894844836e-05,
+      "loss": 0.6721,
+      "step": 8159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.47999487047961e-05,
+      "loss": 0.5721,
+      "step": 8160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479930751474737e-05,
+      "loss": 0.5556,
+      "step": 8161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479866632469864e-05,
+      "loss": 0.5873,
+      "step": 8162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479802513464991e-05,
+      "loss": 0.5691,
+      "step": 8163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479738394460119e-05,
+      "loss": 0.6191,
+      "step": 8164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479674275455245e-05,
+      "loss": 0.6163,
+      "step": 8165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4796101564503724e-05,
+      "loss": 0.5712,
+      "step": 8166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479546037445499e-05,
+      "loss": 0.5794,
+      "step": 8167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479481918440626e-05,
+      "loss": 0.5425,
+      "step": 8168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4794177994357526e-05,
+      "loss": 0.4934,
+      "step": 8169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.47935368043088e-05,
+      "loss": 0.5818,
+      "step": 8170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479289561426007e-05,
+      "loss": 0.6712,
+      "step": 8171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479225442421134e-05,
+      "loss": 0.5214,
+      "step": 8172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479161323416261e-05,
+      "loss": 0.5627,
+      "step": 8173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479097204411388e-05,
+      "loss": 0.5664,
+      "step": 8174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479033085406515e-05,
+      "loss": 0.5956,
+      "step": 8175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4789689664016414e-05,
+      "loss": 0.5244,
+      "step": 8176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4789048473967686e-05,
+      "loss": 0.5792,
+      "step": 8177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478840728391895e-05,
+      "loss": 0.5414,
+      "step": 8178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478776609387022e-05,
+      "loss": 0.4889,
+      "step": 8179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4787124903821494e-05,
+      "loss": 0.6444,
+      "step": 8180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4786483713772766e-05,
+      "loss": 0.5717,
+      "step": 8181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478584252372404e-05,
+      "loss": 0.5337,
+      "step": 8182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.47852013336753e-05,
+      "loss": 0.4915,
+      "step": 8183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4784560143626575e-05,
+      "loss": 0.5958,
+      "step": 8184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478391895357784e-05,
+      "loss": 0.5193,
+      "step": 8185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478327776352911e-05,
+      "loss": 0.5253,
+      "step": 8186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478263657348038e-05,
+      "loss": 0.6022,
+      "step": 8187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4781995383431655e-05,
+      "loss": 0.5115,
+      "step": 8188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478135419338292e-05,
+      "loss": 0.486,
+      "step": 8189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478071300333419e-05,
+      "loss": 0.538,
+      "step": 8190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.478007181328546e-05,
+      "loss": 0.5051,
+      "step": 8191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.477943062323673e-05,
+      "loss": 0.5134,
+      "step": 8192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4778789433188e-05,
+      "loss": 0.4847,
+      "step": 8193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4778148243139265e-05,
+      "loss": 0.5398,
+      "step": 8194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4777507053090536e-05,
+      "loss": 0.5702,
+      "step": 8195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477686586304181e-05,
+      "loss": 0.5464,
+      "step": 8196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477622467299308e-05,
+      "loss": 0.7215,
+      "step": 8197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4775583482944345e-05,
+      "loss": 0.61,
+      "step": 8198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4774942292895616e-05,
+      "loss": 0.5528,
+      "step": 8199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477430110284689e-05,
+      "loss": 0.4328,
+      "step": 8200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477365991279815e-05,
+      "loss": 0.6461,
+      "step": 8201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4773018722749425e-05,
+      "loss": 0.6188,
+      "step": 8202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4772377532700697e-05,
+      "loss": 0.5965,
+      "step": 8203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477173634265197e-05,
+      "loss": 0.5592,
+      "step": 8204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.477109515260323e-05,
+      "loss": 0.6207,
+      "step": 8205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4770453962554505e-05,
+      "loss": 0.5645,
+      "step": 8206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476981277250577e-05,
+      "loss": 0.5914,
+      "step": 8207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476917158245704e-05,
+      "loss": 0.6889,
+      "step": 8208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476853039240831e-05,
+      "loss": 0.6069,
+      "step": 8209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476788920235958e-05,
+      "loss": 0.5072,
+      "step": 8210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476724801231085e-05,
+      "loss": 0.6217,
+      "step": 8211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476660682226212e-05,
+      "loss": 0.507,
+      "step": 8212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476596563221339e-05,
+      "loss": 0.6289,
+      "step": 8213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476532444216466e-05,
+      "loss": 0.5962,
+      "step": 8214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476468325211593e-05,
+      "loss": 0.5978,
+      "step": 8215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4764042062067195e-05,
+      "loss": 0.5786,
+      "step": 8216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476340087201847e-05,
+      "loss": 0.6396,
+      "step": 8217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476275968196974e-05,
+      "loss": 0.5596,
+      "step": 8218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476211849192101e-05,
+      "loss": 0.6311,
+      "step": 8219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476147730187228e-05,
+      "loss": 0.5974,
+      "step": 8220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476083611182355e-05,
+      "loss": 0.5765,
+      "step": 8221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476019492177482e-05,
+      "loss": 0.5754,
+      "step": 8222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4759553731726083e-05,
+      "loss": 0.5315,
+      "step": 8223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4758912541677355e-05,
+      "loss": 0.5359,
+      "step": 8224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475827135162862e-05,
+      "loss": 0.6352,
+      "step": 8225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475763016157989e-05,
+      "loss": 0.5912,
+      "step": 8226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4756988971531164e-05,
+      "loss": 0.5405,
+      "step": 8227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4756347781482435e-05,
+      "loss": 0.6025,
+      "step": 8228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475570659143371e-05,
+      "loss": 0.5438,
+      "step": 8229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475506540138497e-05,
+      "loss": 0.6106,
+      "step": 8230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4754424211336244e-05,
+      "loss": 0.563,
+      "step": 8231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475378302128751e-05,
+      "loss": 0.5386,
+      "step": 8232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475314183123878e-05,
+      "loss": 0.5133,
+      "step": 8233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4752500641190045e-05,
+      "loss": 0.623,
+      "step": 8234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475185945114132e-05,
+      "loss": 0.4978,
+      "step": 8235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4751218261092595e-05,
+      "loss": 0.5696,
+      "step": 8236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475057707104386e-05,
+      "loss": 0.5322,
+      "step": 8237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474993588099513e-05,
+      "loss": 0.4814,
+      "step": 8238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47492946909464e-05,
+      "loss": 0.5864,
+      "step": 8239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474865350089767e-05,
+      "loss": 0.5157,
+      "step": 8240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4748012310848934e-05,
+      "loss": 0.5481,
+      "step": 8241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4747371120800205e-05,
+      "loss": 0.5176,
+      "step": 8242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474672993075148e-05,
+      "loss": 0.5171,
+      "step": 8243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474608874070275e-05,
+      "loss": 0.6355,
+      "step": 8244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474544755065402e-05,
+      "loss": 0.6683,
+      "step": 8245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4744806360605285e-05,
+      "loss": 0.6355,
+      "step": 8246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474416517055656e-05,
+      "loss": 0.7829,
+      "step": 8247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474352398050782e-05,
+      "loss": 0.5261,
+      "step": 8248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4742882790459094e-05,
+      "loss": 0.49,
+      "step": 8249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474224160041036e-05,
+      "loss": 0.5628,
+      "step": 8250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474160041036163e-05,
+      "loss": 0.6591,
+      "step": 8251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47409592203129e-05,
+      "loss": 0.5568,
+      "step": 8252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4740318030264174e-05,
+      "loss": 0.562,
+      "step": 8253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4739676840215446e-05,
+      "loss": 0.4864,
+      "step": 8254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473903565016671e-05,
+      "loss": 0.5893,
+      "step": 8255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473839446011798e-05,
+      "loss": 0.5052,
+      "step": 8256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473775327006925e-05,
+      "loss": 0.4749,
+      "step": 8257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473711208002052e-05,
+      "loss": 0.565,
+      "step": 8258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473647088997179e-05,
+      "loss": 0.6189,
+      "step": 8259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473582969992306e-05,
+      "loss": 0.6117,
+      "step": 8260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473518850987433e-05,
+      "loss": 0.5794,
+      "step": 8261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47345473198256e-05,
+      "loss": 0.4981,
+      "step": 8262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473390612977687e-05,
+      "loss": 0.6414,
+      "step": 8263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4733264939728136e-05,
+      "loss": 0.5316,
+      "step": 8264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473262374967941e-05,
+      "loss": 0.4871,
+      "step": 8265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473198255963067e-05,
+      "loss": 0.5807,
+      "step": 8266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4731341369581944e-05,
+      "loss": 0.5608,
+      "step": 8267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4730700179533216e-05,
+      "loss": 0.6149,
+      "step": 8268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473005898948449e-05,
+      "loss": 0.5393,
+      "step": 8269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472941779943575e-05,
+      "loss": 0.5513,
+      "step": 8270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4728776609387024e-05,
+      "loss": 0.5518,
+      "step": 8271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4728135419338296e-05,
+      "loss": 0.587,
+      "step": 8272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472749422928956e-05,
+      "loss": 0.4899,
+      "step": 8273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472685303924083e-05,
+      "loss": 0.5483,
+      "step": 8274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4726211849192104e-05,
+      "loss": 0.544,
+      "step": 8275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4725570659143376e-05,
+      "loss": 0.5785,
+      "step": 8276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472492946909464e-05,
+      "loss": 0.6495,
+      "step": 8277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472428827904591e-05,
+      "loss": 0.649,
+      "step": 8278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472364708899718e-05,
+      "loss": 0.4872,
+      "step": 8279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472300589894845e-05,
+      "loss": 0.6099,
+      "step": 8280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472236470889972e-05,
+      "loss": 0.6097,
+      "step": 8281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4721723518850986e-05,
+      "loss": 0.6401,
+      "step": 8282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472108232880226e-05,
+      "loss": 0.5427,
+      "step": 8283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472044113875353e-05,
+      "loss": 0.5718,
+      "step": 8284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47197999487048e-05,
+      "loss": 0.5692,
+      "step": 8285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4719158758656066e-05,
+      "loss": 0.5302,
+      "step": 8286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471851756860734e-05,
+      "loss": 0.5331,
+      "step": 8287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471787637855861e-05,
+      "loss": 0.5269,
+      "step": 8288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4717235188509874e-05,
+      "loss": 0.5554,
+      "step": 8289
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4716593998461146e-05,
+      "loss": 0.5875,
+      "step": 8290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471595280841242e-05,
+      "loss": 0.6319,
+      "step": 8291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471531161836369e-05,
+      "loss": 0.5254,
+      "step": 8292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4714670428314955e-05,
+      "loss": 0.61,
+      "step": 8293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4714029238266226e-05,
+      "loss": 0.5337,
+      "step": 8294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471338804821749e-05,
+      "loss": 0.5591,
+      "step": 8295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471274685816876e-05,
+      "loss": 0.4948,
+      "step": 8296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4712105668120035e-05,
+      "loss": 0.592,
+      "step": 8297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47114644780713e-05,
+      "loss": 0.5184,
+      "step": 8298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471082328802257e-05,
+      "loss": 0.7385,
+      "step": 8299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471018209797384e-05,
+      "loss": 0.4295,
+      "step": 8300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4709540907925115e-05,
+      "loss": 0.5951,
+      "step": 8301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470889971787638e-05,
+      "loss": 0.5515,
+      "step": 8302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470825852782765e-05,
+      "loss": 0.5732,
+      "step": 8303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4707617337778916e-05,
+      "loss": 0.5841,
+      "step": 8304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470697614773019e-05,
+      "loss": 0.5996,
+      "step": 8305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470633495768146e-05,
+      "loss": 0.483,
+      "step": 8306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470569376763273e-05,
+      "loss": 0.5282,
+      "step": 8307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4705052577584e-05,
+      "loss": 0.6048,
+      "step": 8308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470441138753527e-05,
+      "loss": 0.5042,
+      "step": 8309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470377019748654e-05,
+      "loss": 0.6443,
+      "step": 8310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4703129007437805e-05,
+      "loss": 0.5357,
+      "step": 8311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4702487817389076e-05,
+      "loss": 0.4597,
+      "step": 8312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470184662734034e-05,
+      "loss": 0.5435,
+      "step": 8313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470120543729161e-05,
+      "loss": 0.6034,
+      "step": 8314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4700564247242885e-05,
+      "loss": 0.5322,
+      "step": 8315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4699923057194157e-05,
+      "loss": 0.5879,
+      "step": 8316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469928186714543e-05,
+      "loss": 0.7021,
+      "step": 8317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469864067709669e-05,
+      "loss": 0.6025,
+      "step": 8318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4697999487047965e-05,
+      "loss": 0.6142,
+      "step": 8319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469735829699923e-05,
+      "loss": 0.5601,
+      "step": 8320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46967171069505e-05,
+      "loss": 0.4916,
+      "step": 8321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4696075916901767e-05,
+      "loss": 0.5905,
+      "step": 8322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469543472685304e-05,
+      "loss": 0.5166,
+      "step": 8323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469479353680432e-05,
+      "loss": 0.5596,
+      "step": 8324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469415234675558e-05,
+      "loss": 0.6768,
+      "step": 8325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469351115670685e-05,
+      "loss": 0.4988,
+      "step": 8326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469286996665812e-05,
+      "loss": 0.5136,
+      "step": 8327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469222877660939e-05,
+      "loss": 0.506,
+      "step": 8328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4691587586560655e-05,
+      "loss": 0.5979,
+      "step": 8329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469094639651193e-05,
+      "loss": 0.5761,
+      "step": 8330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46903052064632e-05,
+      "loss": 0.5813,
+      "step": 8331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468966401641447e-05,
+      "loss": 0.5455,
+      "step": 8332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468902282636574e-05,
+      "loss": 0.4698,
+      "step": 8333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468838163631701e-05,
+      "loss": 0.6155,
+      "step": 8334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468774044626828e-05,
+      "loss": 0.5747,
+      "step": 8335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4687099256219543e-05,
+      "loss": 0.5128,
+      "step": 8336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4686458066170815e-05,
+      "loss": 0.6983,
+      "step": 8337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468581687612208e-05,
+      "loss": 0.7667,
+      "step": 8338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468517568607335e-05,
+      "loss": 0.5372,
+      "step": 8339
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4684534496024624e-05,
+      "loss": 0.5216,
+      "step": 8340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4683893305975895e-05,
+      "loss": 0.5262,
+      "step": 8341
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468325211592717e-05,
+      "loss": 0.5357,
+      "step": 8342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468261092587843e-05,
+      "loss": 0.5657,
+      "step": 8343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4681969735829704e-05,
+      "loss": 0.5757,
+      "step": 8344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468132854578097e-05,
+      "loss": 0.5648,
+      "step": 8345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468068735573224e-05,
+      "loss": 0.6005,
+      "step": 8346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.468004616568351e-05,
+      "loss": 0.5642,
+      "step": 8347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4679404975634784e-05,
+      "loss": 0.4727,
+      "step": 8348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467876378558605e-05,
+      "loss": 0.4809,
+      "step": 8349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467812259553732e-05,
+      "loss": 0.4907,
+      "step": 8350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467748140548859e-05,
+      "loss": 0.4778,
+      "step": 8351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467684021543986e-05,
+      "loss": 0.4571,
+      "step": 8352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467619902539113e-05,
+      "loss": 0.5397,
+      "step": 8353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4675557835342394e-05,
+      "loss": 0.6227,
+      "step": 8354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4674916645293665e-05,
+      "loss": 0.471,
+      "step": 8355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467427545524494e-05,
+      "loss": 0.6201,
+      "step": 8356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467363426519621e-05,
+      "loss": 0.6289,
+      "step": 8357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4672993075147474e-05,
+      "loss": 0.539,
+      "step": 8358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4672351885098745e-05,
+      "loss": 0.5883,
+      "step": 8359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467171069505002e-05,
+      "loss": 0.676,
+      "step": 8360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467106950500128e-05,
+      "loss": 0.5489,
+      "step": 8361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4670428314952554e-05,
+      "loss": 0.5722,
+      "step": 8362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4669787124903826e-05,
+      "loss": 0.6174,
+      "step": 8363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46691459348551e-05,
+      "loss": 0.5459,
+      "step": 8364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466850474480636e-05,
+      "loss": 0.521,
+      "step": 8365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4667863554757634e-05,
+      "loss": 0.5368,
+      "step": 8366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46672223647089e-05,
+      "loss": 0.6514,
+      "step": 8367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466658117466017e-05,
+      "loss": 0.5218,
+      "step": 8368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466593998461144e-05,
+      "loss": 0.6029,
+      "step": 8369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466529879456271e-05,
+      "loss": 0.5652,
+      "step": 8370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466465760451398e-05,
+      "loss": 0.707,
+      "step": 8371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466401641446525e-05,
+      "loss": 0.5996,
+      "step": 8372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466337522441652e-05,
+      "loss": 0.5575,
+      "step": 8373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466273403436779e-05,
+      "loss": 0.4527,
+      "step": 8374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466209284431906e-05,
+      "loss": 0.6506,
+      "step": 8375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4661451654270324e-05,
+      "loss": 0.6583,
+      "step": 8376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4660810464221596e-05,
+      "loss": 0.5663,
+      "step": 8377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466016927417287e-05,
+      "loss": 0.6902,
+      "step": 8378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465952808412414e-05,
+      "loss": 0.6875,
+      "step": 8379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465888689407541e-05,
+      "loss": 0.4993,
+      "step": 8380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4658245704026676e-05,
+      "loss": 0.5254,
+      "step": 8381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465760451397795e-05,
+      "loss": 0.6597,
+      "step": 8382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465696332392921e-05,
+      "loss": 0.6047,
+      "step": 8383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4656322133880484e-05,
+      "loss": 0.6689,
+      "step": 8384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465568094383175e-05,
+      "loss": 0.5464,
+      "step": 8385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465503975378302e-05,
+      "loss": 0.5406,
+      "step": 8386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465439856373429e-05,
+      "loss": 0.6198,
+      "step": 8387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4653757373685564e-05,
+      "loss": 0.5518,
+      "step": 8388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4653116183636836e-05,
+      "loss": 0.5864,
+      "step": 8389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46524749935881e-05,
+      "loss": 0.5955,
+      "step": 8390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465183380353937e-05,
+      "loss": 0.5913,
+      "step": 8391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465119261349064e-05,
+      "loss": 0.575,
+      "step": 8392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465055142344191e-05,
+      "loss": 0.6273,
+      "step": 8393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4649910233393174e-05,
+      "loss": 0.4683,
+      "step": 8394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.464926904334445e-05,
+      "loss": 0.614,
+      "step": 8395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4648627853295724e-05,
+      "loss": 0.5684,
+      "step": 8396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.464798666324699e-05,
+      "loss": 0.6482,
+      "step": 8397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.464734547319826e-05,
+      "loss": 0.6162,
+      "step": 8398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4646704283149526e-05,
+      "loss": 0.646,
+      "step": 8399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46460630931008e-05,
+      "loss": 0.5829,
+      "step": 8400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.464542190305206e-05,
+      "loss": 0.5941,
+      "step": 8401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4644780713003334e-05,
+      "loss": 0.5996,
+      "step": 8402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4644139522954606e-05,
+      "loss": 0.6146,
+      "step": 8403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.464349833290588e-05,
+      "loss": 0.5628,
+      "step": 8404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.5065,
+      "step": 8405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4642215952808415e-05,
+      "loss": 0.6123,
+      "step": 8406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4641574762759686e-05,
+      "loss": 0.5467,
+      "step": 8407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.464093357271095e-05,
+      "loss": 0.554,
+      "step": 8408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.464029238266222e-05,
+      "loss": 0.4757,
+      "step": 8409
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463965119261349e-05,
+      "loss": 0.5519,
+      "step": 8410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463901000256476e-05,
+      "loss": 0.5572,
+      "step": 8411
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463836881251603e-05,
+      "loss": 0.6472,
+      "step": 8412
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46377276224673e-05,
+      "loss": 0.5497,
+      "step": 8413
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4637086432418575e-05,
+      "loss": 0.5282,
+      "step": 8414
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463644524236984e-05,
+      "loss": 0.5764,
+      "step": 8415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463580405232111e-05,
+      "loss": 0.67,
+      "step": 8416
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4635162862272376e-05,
+      "loss": 0.5955,
+      "step": 8417
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463452167222365e-05,
+      "loss": 0.5117,
+      "step": 8418
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463388048217492e-05,
+      "loss": 0.5054,
+      "step": 8419
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463323929212619e-05,
+      "loss": 0.589,
+      "step": 8420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4632598102077456e-05,
+      "loss": 0.5851,
+      "step": 8421
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463195691202873e-05,
+      "loss": 0.5579,
+      "step": 8422
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.463131572198e-05,
+      "loss": 0.5568,
+      "step": 8423
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4630674531931265e-05,
+      "loss": 0.5432,
+      "step": 8424
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4630033341882536e-05,
+      "loss": 0.5968,
+      "step": 8425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46293921518338e-05,
+      "loss": 0.4908,
+      "step": 8426
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462875096178507e-05,
+      "loss": 0.6334,
+      "step": 8427
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4628109771736345e-05,
+      "loss": 0.6447,
+      "step": 8428
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4627468581687617e-05,
+      "loss": 0.5777,
+      "step": 8429
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462682739163888e-05,
+      "loss": 0.5463,
+      "step": 8430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462618620159015e-05,
+      "loss": 0.5494,
+      "step": 8431
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4625545011541425e-05,
+      "loss": 0.5007,
+      "step": 8432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462490382149269e-05,
+      "loss": 0.697,
+      "step": 8433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462426263144396e-05,
+      "loss": 0.8105,
+      "step": 8434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462362144139523e-05,
+      "loss": 0.6072,
+      "step": 8435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4622980251346505e-05,
+      "loss": 0.5348,
+      "step": 8436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462233906129777e-05,
+      "loss": 0.5558,
+      "step": 8437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462169787124904e-05,
+      "loss": 0.5994,
+      "step": 8438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462105668120031e-05,
+      "loss": 0.5724,
+      "step": 8439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462041549115158e-05,
+      "loss": 0.5034,
+      "step": 8440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461977430110285e-05,
+      "loss": 0.5386,
+      "step": 8441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4619133111054115e-05,
+      "loss": 0.6507,
+      "step": 8442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461849192100539e-05,
+      "loss": 0.5968,
+      "step": 8443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461785073095666e-05,
+      "loss": 0.5734,
+      "step": 8444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461720954090793e-05,
+      "loss": 0.6165,
+      "step": 8445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4616568350859195e-05,
+      "loss": 0.5785,
+      "step": 8446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461592716081047e-05,
+      "loss": 0.6404,
+      "step": 8447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461528597076173e-05,
+      "loss": 0.5599,
+      "step": 8448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4614644780713003e-05,
+      "loss": 0.5289,
+      "step": 8449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4614003590664275e-05,
+      "loss": 0.6064,
+      "step": 8450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461336240061555e-05,
+      "loss": 0.5112,
+      "step": 8451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461272121056682e-05,
+      "loss": 0.5393,
+      "step": 8452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4612080020518084e-05,
+      "loss": 0.4976,
+      "step": 8453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4611438830469355e-05,
+      "loss": 0.5244,
+      "step": 8454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.461079764042062e-05,
+      "loss": 0.6119,
+      "step": 8455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.461015645037189e-05,
+      "loss": 0.586,
+      "step": 8456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460951526032316e-05,
+      "loss": 0.5326,
+      "step": 8457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460887407027443e-05,
+      "loss": 0.6281,
+      "step": 8458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.46082328802257e-05,
+      "loss": 0.4957,
+      "step": 8459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460759169017697e-05,
+      "loss": 0.5856,
+      "step": 8460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4606950500128244e-05,
+      "loss": 0.5929,
+      "step": 8461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460630931007951e-05,
+      "loss": 0.6137,
+      "step": 8462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460566812003078e-05,
+      "loss": 0.5301,
+      "step": 8463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4605026929982045e-05,
+      "loss": 0.5196,
+      "step": 8464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460438573993332e-05,
+      "loss": 0.5996,
+      "step": 8465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460374454988458e-05,
+      "loss": 0.5088,
+      "step": 8466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460310335983586e-05,
+      "loss": 0.542,
+      "step": 8467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460246216978713e-05,
+      "loss": 0.7661,
+      "step": 8468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.46018209797384e-05,
+      "loss": 0.5526,
+      "step": 8469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460117978968967e-05,
+      "loss": 0.7504,
+      "step": 8470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4600538599640934e-05,
+      "loss": 0.8382,
+      "step": 8471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4599897409592206e-05,
+      "loss": 0.6157,
+      "step": 8472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459925621954347e-05,
+      "loss": 0.5429,
+      "step": 8473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459861502949474e-05,
+      "loss": 0.5574,
+      "step": 8474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4597973839446014e-05,
+      "loss": 0.547,
+      "step": 8475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4597332649397286e-05,
+      "loss": 0.5222,
+      "step": 8476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459669145934856e-05,
+      "loss": 0.6044,
+      "step": 8477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459605026929982e-05,
+      "loss": 0.567,
+      "step": 8478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4595409079251094e-05,
+      "loss": 0.6187,
+      "step": 8479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459476788920236e-05,
+      "loss": 0.6082,
+      "step": 8480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459412669915363e-05,
+      "loss": 0.6045,
+      "step": 8481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4593485509104896e-05,
+      "loss": 0.5425,
+      "step": 8482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459284431905617e-05,
+      "loss": 0.5788,
+      "step": 8483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4592203129007446e-05,
+      "loss": 0.6039,
+      "step": 8484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459156193895871e-05,
+      "loss": 0.5453,
+      "step": 8485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459092074890998e-05,
+      "loss": 0.6296,
+      "step": 8486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459027955886125e-05,
+      "loss": 0.5769,
+      "step": 8487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458963836881252e-05,
+      "loss": 0.4666,
+      "step": 8488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4588997178763784e-05,
+      "loss": 0.5986,
+      "step": 8489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4588355988715056e-05,
+      "loss": 0.6026,
+      "step": 8490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458771479866633e-05,
+      "loss": 0.569,
+      "step": 8491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45870736086176e-05,
+      "loss": 0.6584,
+      "step": 8492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458643241856887e-05,
+      "loss": 0.5207,
+      "step": 8493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4585791228520136e-05,
+      "loss": 0.5926,
+      "step": 8494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458515003847141e-05,
+      "loss": 0.5518,
+      "step": 8495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458450884842267e-05,
+      "loss": 0.5685,
+      "step": 8496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4583867658373944e-05,
+      "loss": 0.6054,
+      "step": 8497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458322646832521e-05,
+      "loss": 0.592,
+      "step": 8498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458258527827648e-05,
+      "loss": 0.5857,
+      "step": 8499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458194408822775e-05,
+      "loss": 0.5644,
+      "step": 8500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4581302898179024e-05,
+      "loss": 0.609,
+      "step": 8501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4580661708130296e-05,
+      "loss": 0.526,
+      "step": 8502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458002051808156e-05,
+      "loss": 0.5252,
+      "step": 8503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457937932803283e-05,
+      "loss": 0.546,
+      "step": 8504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45787381379841e-05,
+      "loss": 0.5879,
+      "step": 8505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457809694793537e-05,
+      "loss": 0.5987,
+      "step": 8506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457745575788664e-05,
+      "loss": 0.6078,
+      "step": 8507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457681456783791e-05,
+      "loss": 0.5735,
+      "step": 8508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457617337778918e-05,
+      "loss": 0.5509,
+      "step": 8509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457553218774045e-05,
+      "loss": 0.4713,
+      "step": 8510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457489099769172e-05,
+      "loss": 0.5488,
+      "step": 8511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4574249807642986e-05,
+      "loss": 0.4776,
+      "step": 8512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457360861759426e-05,
+      "loss": 0.6395,
+      "step": 8513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457296742754552e-05,
+      "loss": 0.5624,
+      "step": 8514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4572326237496794e-05,
+      "loss": 0.5347,
+      "step": 8515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4571685047448066e-05,
+      "loss": 0.5646,
+      "step": 8516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457104385739934e-05,
+      "loss": 0.5876,
+      "step": 8517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45704026673506e-05,
+      "loss": 0.5156,
+      "step": 8518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4569761477301875e-05,
+      "loss": 0.6344,
+      "step": 8519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4569120287253146e-05,
+      "loss": 0.5436,
+      "step": 8520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456847909720441e-05,
+      "loss": 0.513,
+      "step": 8521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456783790715568e-05,
+      "loss": 0.6599,
+      "step": 8522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4567196717106955e-05,
+      "loss": 0.6396,
+      "step": 8523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4566555527058226e-05,
+      "loss": 0.512,
+      "step": 8524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456591433700949e-05,
+      "loss": 0.4499,
+      "step": 8525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456527314696076e-05,
+      "loss": 0.5084,
+      "step": 8526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456463195691203e-05,
+      "loss": 0.5946,
+      "step": 8527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45639907668633e-05,
+      "loss": 0.564,
+      "step": 8528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456334957681457e-05,
+      "loss": 0.6641,
+      "step": 8529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4562708386765836e-05,
+      "loss": 0.5371,
+      "step": 8530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456206719671711e-05,
+      "loss": 0.5412,
+      "step": 8531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456142600666838e-05,
+      "loss": 0.4866,
+      "step": 8532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456078481661965e-05,
+      "loss": 0.5538,
+      "step": 8533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4560143626570916e-05,
+      "loss": 0.6568,
+      "step": 8534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455950243652219e-05,
+      "loss": 0.5408,
+      "step": 8535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455886124647345e-05,
+      "loss": 0.5664,
+      "step": 8536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4558220056424725e-05,
+      "loss": 0.6182,
+      "step": 8537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4557578866375997e-05,
+      "loss": 0.543,
+      "step": 8538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455693767632727e-05,
+      "loss": 0.6033,
+      "step": 8539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455629648627854e-05,
+      "loss": 0.5936,
+      "step": 8540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4555655296229805e-05,
+      "loss": 0.5191,
+      "step": 8541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455501410618108e-05,
+      "loss": 0.6485,
+      "step": 8542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455437291613234e-05,
+      "loss": 0.5846,
+      "step": 8543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455373172608361e-05,
+      "loss": 0.56,
+      "step": 8544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455309053603488e-05,
+      "loss": 0.6173,
+      "step": 8545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455244934598615e-05,
+      "loss": 0.5275,
+      "step": 8546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455180815593742e-05,
+      "loss": 0.568,
+      "step": 8547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455116696588869e-05,
+      "loss": 0.6102,
+      "step": 8548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4550525775839965e-05,
+      "loss": 0.4958,
+      "step": 8549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454988458579123e-05,
+      "loss": 0.5811,
+      "step": 8550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45492433957425e-05,
+      "loss": 0.5928,
+      "step": 8551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454860220569377e-05,
+      "loss": 0.6326,
+      "step": 8552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454796101564504e-05,
+      "loss": 0.5356,
+      "step": 8553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45473198255963e-05,
+      "loss": 0.5512,
+      "step": 8554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454667863554758e-05,
+      "loss": 0.5513,
+      "step": 8555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4546037445498854e-05,
+      "loss": 0.5236,
+      "step": 8556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454539625545012e-05,
+      "loss": 0.5549,
+      "step": 8557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454475506540139e-05,
+      "loss": 0.601,
+      "step": 8558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4544113875352655e-05,
+      "loss": 0.5881,
+      "step": 8559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454347268530393e-05,
+      "loss": 0.5898,
+      "step": 8560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454283149525519e-05,
+      "loss": 0.503,
+      "step": 8561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4542190305206464e-05,
+      "loss": 0.4955,
+      "step": 8562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4541549115157735e-05,
+      "loss": 0.6298,
+      "step": 8563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454090792510901e-05,
+      "loss": 0.5059,
+      "step": 8564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454026673506028e-05,
+      "loss": 0.5371,
+      "step": 8565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4539625545011544e-05,
+      "loss": 0.6468,
+      "step": 8566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4538984354962815e-05,
+      "loss": 0.6016,
+      "step": 8567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453834316491408e-05,
+      "loss": 0.5248,
+      "step": 8568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453770197486535e-05,
+      "loss": 0.5268,
+      "step": 8569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453706078481662e-05,
+      "loss": 0.4643,
+      "step": 8570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453641959476789e-05,
+      "loss": 0.4879,
+      "step": 8571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453577840471916e-05,
+      "loss": 0.6953,
+      "step": 8572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453513721467043e-05,
+      "loss": 0.7358,
+      "step": 8573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4534496024621704e-05,
+      "loss": 0.56,
+      "step": 8574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453385483457297e-05,
+      "loss": 0.5507,
+      "step": 8575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453321364452424e-05,
+      "loss": 0.4916,
+      "step": 8576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4532572454475505e-05,
+      "loss": 0.5772,
+      "step": 8577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453193126442678e-05,
+      "loss": 0.5232,
+      "step": 8578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453129007437805e-05,
+      "loss": 0.5868,
+      "step": 8579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453064888432932e-05,
+      "loss": 0.499,
+      "step": 8580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4530007694280585e-05,
+      "loss": 0.5896,
+      "step": 8581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452936650423186e-05,
+      "loss": 0.5295,
+      "step": 8582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452872531418313e-05,
+      "loss": 0.6621,
+      "step": 8583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4528084124134394e-05,
+      "loss": 0.62,
+      "step": 8584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4527442934085666e-05,
+      "loss": 0.5527,
+      "step": 8585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452680174403693e-05,
+      "loss": 0.5928,
+      "step": 8586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45261605539882e-05,
+      "loss": 0.5056,
+      "step": 8587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4525519363939474e-05,
+      "loss": 0.6687,
+      "step": 8588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4524878173890746e-05,
+      "loss": 0.6651,
+      "step": 8589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452423698384201e-05,
+      "loss": 0.514,
+      "step": 8590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452359579379328e-05,
+      "loss": 0.569,
+      "step": 8591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4522954603744554e-05,
+      "loss": 0.5633,
+      "step": 8592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452231341369582e-05,
+      "loss": 0.5758,
+      "step": 8593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452167222364709e-05,
+      "loss": 0.481,
+      "step": 8594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452103103359836e-05,
+      "loss": 0.6111,
+      "step": 8595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4520389843549634e-05,
+      "loss": 0.52,
+      "step": 8596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45197486535009e-05,
+      "loss": 0.5581,
+      "step": 8597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451910746345217e-05,
+      "loss": 0.6316,
+      "step": 8598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4518466273403436e-05,
+      "loss": 0.4735,
+      "step": 8599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451782508335471e-05,
+      "loss": 0.5669,
+      "step": 8600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451718389330598e-05,
+      "loss": 0.6258,
+      "step": 8601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4516542703257244e-05,
+      "loss": 0.6813,
+      "step": 8602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4515901513208516e-05,
+      "loss": 0.5675,
+      "step": 8603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451526032315979e-05,
+      "loss": 0.6272,
+      "step": 8604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451461913311106e-05,
+      "loss": 0.6091,
+      "step": 8605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4513977943062324e-05,
+      "loss": 0.4786,
+      "step": 8606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4513336753013596e-05,
+      "loss": 0.576,
+      "step": 8607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451269556296486e-05,
+      "loss": 0.488,
+      "step": 8608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451205437291613e-05,
+      "loss": 0.584,
+      "step": 8609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4511413182867404e-05,
+      "loss": 0.4877,
+      "step": 8610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4510771992818676e-05,
+      "loss": 0.5283,
+      "step": 8611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451013080276995e-05,
+      "loss": 0.5545,
+      "step": 8612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450948961272121e-05,
+      "loss": 0.5301,
+      "step": 8613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4508848422672484e-05,
+      "loss": 0.5562,
+      "step": 8614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450820723262375e-05,
+      "loss": 0.4665,
+      "step": 8615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450756604257502e-05,
+      "loss": 0.5947,
+      "step": 8616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4506924852526286e-05,
+      "loss": 0.5629,
+      "step": 8617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450628366247756e-05,
+      "loss": 0.6121,
+      "step": 8618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450564247242883e-05,
+      "loss": 0.505,
+      "step": 8619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45050012823801e-05,
+      "loss": 0.6102,
+      "step": 8620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450436009233137e-05,
+      "loss": 0.5233,
+      "step": 8621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450371890228264e-05,
+      "loss": 0.6834,
+      "step": 8622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450307771223391e-05,
+      "loss": 0.6189,
+      "step": 8623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4502436522185174e-05,
+      "loss": 0.5953,
+      "step": 8624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4501795332136446e-05,
+      "loss": 0.4646,
+      "step": 8625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450115414208771e-05,
+      "loss": 0.4995,
+      "step": 8626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450051295203899e-05,
+      "loss": 0.6056,
+      "step": 8627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449987176199026e-05,
+      "loss": 0.5437,
+      "step": 8628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4499230571941526e-05,
+      "loss": 0.5747,
+      "step": 8629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.44985893818928e-05,
+      "loss": 0.5623,
+      "step": 8630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449794819184406e-05,
+      "loss": 0.629,
+      "step": 8631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4497307001795335e-05,
+      "loss": 0.5062,
+      "step": 8632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.44966658117466e-05,
+      "loss": 0.527,
+      "step": 8633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449602462169787e-05,
+      "loss": 0.5914,
+      "step": 8634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449538343164914e-05,
+      "loss": 0.6227,
+      "step": 8635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4494742241600415e-05,
+      "loss": 0.6255,
+      "step": 8636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4494101051551686e-05,
+      "loss": 0.5885,
+      "step": 8637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449345986150295e-05,
+      "loss": 0.4969,
+      "step": 8638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449281867145422e-05,
+      "loss": 0.5665,
+      "step": 8639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449217748140549e-05,
+      "loss": 0.5367,
+      "step": 8640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.449153629135676e-05,
+      "loss": 0.5566,
+      "step": 8641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4490895101308025e-05,
+      "loss": 0.5312,
+      "step": 8642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4490253911259296e-05,
+      "loss": 0.5489,
+      "step": 8643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448961272121057e-05,
+      "loss": 0.5783,
+      "step": 8644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448897153116184e-05,
+      "loss": 0.47,
+      "step": 8645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448833034111311e-05,
+      "loss": 0.6031,
+      "step": 8646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4487689151064376e-05,
+      "loss": 0.5451,
+      "step": 8647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448704796101565e-05,
+      "loss": 0.6982,
+      "step": 8648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448640677096691e-05,
+      "loss": 0.5791,
+      "step": 8649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4485765580918185e-05,
+      "loss": 0.5708,
+      "step": 8650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4485124390869457e-05,
+      "loss": 0.5494,
+      "step": 8651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448448320082073e-05,
+      "loss": 0.5811,
+      "step": 8652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448384201077199e-05,
+      "loss": 0.5815,
+      "step": 8653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4483200820723265e-05,
+      "loss": 0.6,
+      "step": 8654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448255963067454e-05,
+      "loss": 0.6354,
+      "step": 8655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.44819184406258e-05,
+      "loss": 0.4936,
+      "step": 8656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448127725057707e-05,
+      "loss": 0.5552,
+      "step": 8657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448063606052834e-05,
+      "loss": 0.6033,
+      "step": 8658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447999487047961e-05,
+      "loss": 0.5372,
+      "step": 8659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447935368043088e-05,
+      "loss": 0.5252,
+      "step": 8660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4478712490382153e-05,
+      "loss": 0.5616,
+      "step": 8661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4478071300333425e-05,
+      "loss": 0.6027,
+      "step": 8662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447743011028469e-05,
+      "loss": 0.6886,
+      "step": 8663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447678892023596e-05,
+      "loss": 0.5648,
+      "step": 8664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447614773018723e-05,
+      "loss": 0.5501,
+      "step": 8665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.44755065401385e-05,
+      "loss": 0.5457,
+      "step": 8666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447486535008977e-05,
+      "loss": 0.5983,
+      "step": 8667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447422416004104e-05,
+      "loss": 0.6195,
+      "step": 8668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447358296999231e-05,
+      "loss": 0.5424,
+      "step": 8669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447294177994358e-05,
+      "loss": 0.6192,
+      "step": 8670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447230058989485e-05,
+      "loss": 0.5799,
+      "step": 8671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4471659399846115e-05,
+      "loss": 0.5444,
+      "step": 8672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447101820979739e-05,
+      "loss": 0.6123,
+      "step": 8673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447037701974865e-05,
+      "loss": 0.6593,
+      "step": 8674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4469735829699924e-05,
+      "loss": 0.6279,
+      "step": 8675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4469094639651195e-05,
+      "loss": 0.5951,
+      "step": 8676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446845344960247e-05,
+      "loss": 0.5151,
+      "step": 8677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446781225955373e-05,
+      "loss": 0.6075,
+      "step": 8678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4467171069505004e-05,
+      "loss": 0.5753,
+      "step": 8679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4466529879456275e-05,
+      "loss": 0.6379,
+      "step": 8680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446588868940754e-05,
+      "loss": 0.6054,
+      "step": 8681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446524749935881e-05,
+      "loss": 0.5593,
+      "step": 8682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4464606309310084e-05,
+      "loss": 0.5612,
+      "step": 8683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4463965119261355e-05,
+      "loss": 0.5745,
+      "step": 8684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446332392921262e-05,
+      "loss": 0.5894,
+      "step": 8685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446268273916389e-05,
+      "loss": 0.5891,
+      "step": 8686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446204154911516e-05,
+      "loss": 0.6131,
+      "step": 8687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446140035906643e-05,
+      "loss": 0.5895,
+      "step": 8688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.44607591690177e-05,
+      "loss": 0.5779,
+      "step": 8689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4460117978968965e-05,
+      "loss": 0.5145,
+      "step": 8690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445947678892024e-05,
+      "loss": 0.6314,
+      "step": 8691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445883559887151e-05,
+      "loss": 0.5687,
+      "step": 8692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445819440882278e-05,
+      "loss": 0.5881,
+      "step": 8693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4457553218774046e-05,
+      "loss": 0.6712,
+      "step": 8694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445691202872532e-05,
+      "loss": 0.5677,
+      "step": 8695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445627083867658e-05,
+      "loss": 0.6094,
+      "step": 8696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4455629648627854e-05,
+      "loss": 0.6274,
+      "step": 8697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4454988458579126e-05,
+      "loss": 0.5708,
+      "step": 8698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.44543472685304e-05,
+      "loss": 0.571,
+      "step": 8699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445370607848167e-05,
+      "loss": 0.565,
+      "step": 8700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4453064888432934e-05,
+      "loss": 0.5172,
+      "step": 8701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4452423698384206e-05,
+      "loss": 0.6369,
+      "step": 8702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445178250833547e-05,
+      "loss": 0.6064,
+      "step": 8703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445114131828674e-05,
+      "loss": 0.6832,
+      "step": 8704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445050012823801e-05,
+      "loss": 0.545,
+      "step": 8705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.444985893818928e-05,
+      "loss": 0.5337,
+      "step": 8706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.444921774814055e-05,
+      "loss": 0.5004,
+      "step": 8707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.444857655809182e-05,
+      "loss": 0.5554,
+      "step": 8708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4447935368043094e-05,
+      "loss": 0.5299,
+      "step": 8709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.444729417799436e-05,
+      "loss": 0.6594,
+      "step": 8710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.444665298794563e-05,
+      "loss": 0.5537,
+      "step": 8711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4446011797896896e-05,
+      "loss": 0.5927,
+      "step": 8712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.444537060784817e-05,
+      "loss": 0.5401,
+      "step": 8713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444472941779943e-05,
+      "loss": 0.6057,
+      "step": 8714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444408822775071e-05,
+      "loss": 0.4679,
+      "step": 8715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444344703770198e-05,
+      "loss": 0.6532,
+      "step": 8716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444280584765325e-05,
+      "loss": 0.5308,
+      "step": 8717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444216465760452e-05,
+      "loss": 0.5832,
+      "step": 8718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4441523467555784e-05,
+      "loss": 0.4797,
+      "step": 8719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4440882277507056e-05,
+      "loss": 0.5775,
+      "step": 8720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444024108745832e-05,
+      "loss": 0.5155,
+      "step": 8721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443959989740959e-05,
+      "loss": 0.5288,
+      "step": 8722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4438958707360864e-05,
+      "loss": 0.5125,
+      "step": 8723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4438317517312136e-05,
+      "loss": 0.5652,
+      "step": 8724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443767632726341e-05,
+      "loss": 0.5388,
+      "step": 8725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443703513721467e-05,
+      "loss": 0.5803,
+      "step": 8726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4436393947165944e-05,
+      "loss": 0.4764,
+      "step": 8727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443575275711721e-05,
+      "loss": 0.6986,
+      "step": 8728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443511156706848e-05,
+      "loss": 0.5308,
+      "step": 8729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4434470377019746e-05,
+      "loss": 0.4395,
+      "step": 8730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443382918697102e-05,
+      "loss": 0.6091,
+      "step": 8731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443318799692229e-05,
+      "loss": 0.5866,
+      "step": 8732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443254680687356e-05,
+      "loss": 0.6178,
+      "step": 8733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443190561682483e-05,
+      "loss": 0.6319,
+      "step": 8734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44312644267761e-05,
+      "loss": 0.6602,
+      "step": 8735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443062323672737e-05,
+      "loss": 0.6313,
+      "step": 8736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4429982046678634e-05,
+      "loss": 0.6359,
+      "step": 8737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4429340856629906e-05,
+      "loss": 0.5891,
+      "step": 8738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442869966658118e-05,
+      "loss": 0.6262,
+      "step": 8739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442805847653245e-05,
+      "loss": 0.6282,
+      "step": 8740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4427417286483715e-05,
+      "loss": 0.4623,
+      "step": 8741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4426776096434986e-05,
+      "loss": 0.602,
+      "step": 8742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442613490638626e-05,
+      "loss": 0.6928,
+      "step": 8743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442549371633752e-05,
+      "loss": 0.5637,
+      "step": 8744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4424852526288795e-05,
+      "loss": 0.5023,
+      "step": 8745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442421133624006e-05,
+      "loss": 0.6135,
+      "step": 8746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442357014619133e-05,
+      "loss": 0.5071,
+      "step": 8747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44229289561426e-05,
+      "loss": 0.5028,
+      "step": 8748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4422287766093875e-05,
+      "loss": 0.6358,
+      "step": 8749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442164657604514e-05,
+      "loss": 0.586,
+      "step": 8750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442100538599641e-05,
+      "loss": 0.5267,
+      "step": 8751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442036419594768e-05,
+      "loss": 0.5325,
+      "step": 8752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441972300589895e-05,
+      "loss": 0.529,
+      "step": 8753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441908181585022e-05,
+      "loss": 0.5653,
+      "step": 8754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441844062580149e-05,
+      "loss": 0.6391,
+      "step": 8755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441779943575276e-05,
+      "loss": 0.582,
+      "step": 8756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441715824570403e-05,
+      "loss": 0.5306,
+      "step": 8757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44165170556553e-05,
+      "loss": 0.6226,
+      "step": 8758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4415875865606565e-05,
+      "loss": 0.6711,
+      "step": 8759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4415234675557837e-05,
+      "loss": 0.5711,
+      "step": 8760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441459348550911e-05,
+      "loss": 0.5608,
+      "step": 8761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441395229546037e-05,
+      "loss": 0.6079,
+      "step": 8762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4413311105411645e-05,
+      "loss": 0.6067,
+      "step": 8763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4412669915362917e-05,
+      "loss": 0.5683,
+      "step": 8764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441202872531419e-05,
+      "loss": 0.6122,
+      "step": 8765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441138753526545e-05,
+      "loss": 0.5795,
+      "step": 8766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4410746345216725e-05,
+      "loss": 0.5721,
+      "step": 8767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441010515516799e-05,
+      "loss": 0.6031,
+      "step": 8768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440946396511926e-05,
+      "loss": 0.5145,
+      "step": 8769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440882277507053e-05,
+      "loss": 0.6036,
+      "step": 8770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4408181585021805e-05,
+      "loss": 0.5342,
+      "step": 8771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440754039497308e-05,
+      "loss": 0.5377,
+      "step": 8772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440689920492434e-05,
+      "loss": 0.6469,
+      "step": 8773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4406258014875613e-05,
+      "loss": 0.5312,
+      "step": 8774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440561682482688e-05,
+      "loss": 0.5765,
+      "step": 8775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440497563477815e-05,
+      "loss": 0.5734,
+      "step": 8776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4404334444729415e-05,
+      "loss": 0.562,
+      "step": 8777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440369325468069e-05,
+      "loss": 0.484,
+      "step": 8778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440305206463196e-05,
+      "loss": 0.5952,
+      "step": 8779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440241087458323e-05,
+      "loss": 0.5669,
+      "step": 8780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44017696845345e-05,
+      "loss": 0.5917,
+      "step": 8781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440112849448577e-05,
+      "loss": 0.5991,
+      "step": 8782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440048730443704e-05,
+      "loss": 0.5956,
+      "step": 8783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4399846114388303e-05,
+      "loss": 0.5935,
+      "step": 8784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4399204924339575e-05,
+      "loss": 0.5771,
+      "step": 8785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439856373429084e-05,
+      "loss": 0.4787,
+      "step": 8786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439792254424212e-05,
+      "loss": 0.6257,
+      "step": 8787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439728135419339e-05,
+      "loss": 0.4667,
+      "step": 8788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4396640164144655e-05,
+      "loss": 0.5521,
+      "step": 8789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439599897409593e-05,
+      "loss": 0.5951,
+      "step": 8790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439535778404719e-05,
+      "loss": 0.6056,
+      "step": 8791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4394716593998464e-05,
+      "loss": 0.5413,
+      "step": 8792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439407540394973e-05,
+      "loss": 0.6237,
+      "step": 8793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4393434213901e-05,
+      "loss": 0.4904,
+      "step": 8794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439279302385227e-05,
+      "loss": 0.6256,
+      "step": 8795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4392151833803544e-05,
+      "loss": 0.5831,
+      "step": 8796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4391510643754815e-05,
+      "loss": 0.6471,
+      "step": 8797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439086945370608e-05,
+      "loss": 0.6336,
+      "step": 8798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439022826365735e-05,
+      "loss": 0.5065,
+      "step": 8799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438958707360862e-05,
+      "loss": 0.5708,
+      "step": 8800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438894588355989e-05,
+      "loss": 0.6035,
+      "step": 8801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4388304693511154e-05,
+      "loss": 0.51,
+      "step": 8802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438766350346243e-05,
+      "loss": 0.5028,
+      "step": 8803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43870223134137e-05,
+      "loss": 0.5887,
+      "step": 8804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438638112336497e-05,
+      "loss": 0.5578,
+      "step": 8805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438573993331624e-05,
+      "loss": 0.5142,
+      "step": 8806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4385098743267506e-05,
+      "loss": 0.5464,
+      "step": 8807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438445755321878e-05,
+      "loss": 0.5294,
+      "step": 8808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438381636317004e-05,
+      "loss": 0.4524,
+      "step": 8809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4383175173121314e-05,
+      "loss": 0.5761,
+      "step": 8810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4382533983072586e-05,
+      "loss": 0.6029,
+      "step": 8811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438189279302386e-05,
+      "loss": 0.6398,
+      "step": 8812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438125160297512e-05,
+      "loss": 0.6633,
+      "step": 8813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4380610412926394e-05,
+      "loss": 0.6057,
+      "step": 8814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4379969222877666e-05,
+      "loss": 0.625,
+      "step": 8815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437932803282893e-05,
+      "loss": 0.6958,
+      "step": 8816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43786868427802e-05,
+      "loss": 0.6101,
+      "step": 8817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437804565273147e-05,
+      "loss": 0.5973,
+      "step": 8818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437740446268274e-05,
+      "loss": 0.5454,
+      "step": 8819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437676327263401e-05,
+      "loss": 0.5913,
+      "step": 8820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437612208258528e-05,
+      "loss": 0.6017,
+      "step": 8821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437548089253655e-05,
+      "loss": 0.494,
+      "step": 8822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437483970248782e-05,
+      "loss": 0.6007,
+      "step": 8823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437419851243909e-05,
+      "loss": 0.5673,
+      "step": 8824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4373557322390356e-05,
+      "loss": 0.5238,
+      "step": 8825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437291613234163e-05,
+      "loss": 0.6716,
+      "step": 8826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43722749422929e-05,
+      "loss": 0.5319,
+      "step": 8827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437163375224417e-05,
+      "loss": 0.5781,
+      "step": 8828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4370992562195436e-05,
+      "loss": 0.55,
+      "step": 8829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437035137214671e-05,
+      "loss": 0.5078,
+      "step": 8830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436971018209797e-05,
+      "loss": 0.59,
+      "step": 8831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4369068992049244e-05,
+      "loss": 0.6293,
+      "step": 8832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4368427802000516e-05,
+      "loss": 0.6091,
+      "step": 8833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436778661195178e-05,
+      "loss": 0.592,
+      "step": 8834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436714542190305e-05,
+      "loss": 0.5411,
+      "step": 8835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4366504231854324e-05,
+      "loss": 0.4376,
+      "step": 8836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4365863041805596e-05,
+      "loss": 0.6065,
+      "step": 8837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436522185175686e-05,
+      "loss": 0.6921,
+      "step": 8838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436458066170813e-05,
+      "loss": 0.558,
+      "step": 8839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43639394716594e-05,
+      "loss": 0.6565,
+      "step": 8840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436329828161067e-05,
+      "loss": 0.6229,
+      "step": 8841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436265709156194e-05,
+      "loss": 0.5987,
+      "step": 8842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436201590151321e-05,
+      "loss": 0.6404,
+      "step": 8843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4361374711464485e-05,
+      "loss": 0.6127,
+      "step": 8844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436073352141575e-05,
+      "loss": 0.5308,
+      "step": 8845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436009233136702e-05,
+      "loss": 0.5888,
+      "step": 8846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4359451141318286e-05,
+      "loss": 0.5852,
+      "step": 8847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435880995126956e-05,
+      "loss": 0.573,
+      "step": 8848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435816876122082e-05,
+      "loss": 0.587,
+      "step": 8849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4357527571172094e-05,
+      "loss": 0.5337,
+      "step": 8850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4356886381123366e-05,
+      "loss": 0.5503,
+      "step": 8851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435624519107464e-05,
+      "loss": 0.5981,
+      "step": 8852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435560400102591e-05,
+      "loss": 0.6705,
+      "step": 8853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4354962810977175e-05,
+      "loss": 0.589,
+      "step": 8854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4354321620928446e-05,
+      "loss": 0.6211,
+      "step": 8855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435368043087971e-05,
+      "loss": 0.6081,
+      "step": 8856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435303924083098e-05,
+      "loss": 0.633,
+      "step": 8857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4352398050782255e-05,
+      "loss": 0.5802,
+      "step": 8858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4351756860733526e-05,
+      "loss": 0.7109,
+      "step": 8859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43511156706848e-05,
+      "loss": 0.5092,
+      "step": 8860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435047448063606e-05,
+      "loss": 0.5743,
+      "step": 8861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4349833290587335e-05,
+      "loss": 0.5592,
+      "step": 8862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43491921005386e-05,
+      "loss": 0.5535,
+      "step": 8863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434855091048987e-05,
+      "loss": 0.5637,
+      "step": 8864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4347909720441136e-05,
+      "loss": 0.5626,
+      "step": 8865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434726853039241e-05,
+      "loss": 0.528,
+      "step": 8866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434662734034368e-05,
+      "loss": 0.5875,
+      "step": 8867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434598615029495e-05,
+      "loss": 0.6103,
+      "step": 8868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434534496024622e-05,
+      "loss": 0.5939,
+      "step": 8869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434470377019749e-05,
+      "loss": 0.6075,
+      "step": 8870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434406258014876e-05,
+      "loss": 0.5509,
+      "step": 8871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4343421390100025e-05,
+      "loss": 0.5603,
+      "step": 8872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4342780200051297e-05,
+      "loss": 0.6138,
+      "step": 8873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434213901000256e-05,
+      "loss": 0.573,
+      "step": 8874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434149781995384e-05,
+      "loss": 0.5892,
+      "step": 8875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434085662990511e-05,
+      "loss": 0.529,
+      "step": 8876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.434021543985638e-05,
+      "loss": 0.5836,
+      "step": 8877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433957424980765e-05,
+      "loss": 0.5257,
+      "step": 8878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433893305975891e-05,
+      "loss": 0.6047,
+      "step": 8879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4338291869710185e-05,
+      "loss": 0.5706,
+      "step": 8880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433765067966145e-05,
+      "loss": 0.5343,
+      "step": 8881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433700948961272e-05,
+      "loss": 0.6146,
+      "step": 8882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433636829956399e-05,
+      "loss": 0.5617,
+      "step": 8883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4335727109515265e-05,
+      "loss": 0.6151,
+      "step": 8884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433508591946654e-05,
+      "loss": 0.5351,
+      "step": 8885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43344447294178e-05,
+      "loss": 0.5274,
+      "step": 8886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4333803539369073e-05,
+      "loss": 0.6626,
+      "step": 8887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433316234932034e-05,
+      "loss": 0.6112,
+      "step": 8888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433252115927161e-05,
+      "loss": 0.5311,
+      "step": 8889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4331879969222875e-05,
+      "loss": 0.5306,
+      "step": 8890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433123877917415e-05,
+      "loss": 0.5374,
+      "step": 8891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.433059758912542e-05,
+      "loss": 0.5619,
+      "step": 8892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432995639907669e-05,
+      "loss": 0.5548,
+      "step": 8893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432931520902796e-05,
+      "loss": 0.5704,
+      "step": 8894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432867401897923e-05,
+      "loss": 0.5817,
+      "step": 8895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43280328289305e-05,
+      "loss": 0.5817,
+      "step": 8896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4327391638881764e-05,
+      "loss": 0.5322,
+      "step": 8897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4326750448833035e-05,
+      "loss": 0.5936,
+      "step": 8898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432610925878431e-05,
+      "loss": 0.6623,
+      "step": 8899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432546806873558e-05,
+      "loss": 0.5917,
+      "step": 8900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4324826878686844e-05,
+      "loss": 0.5367,
+      "step": 8901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4324185688638115e-05,
+      "loss": 0.5396,
+      "step": 8902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432354449858939e-05,
+      "loss": 0.6593,
+      "step": 8903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432290330854065e-05,
+      "loss": 0.5788,
+      "step": 8904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4322262118491924e-05,
+      "loss": 0.5923,
+      "step": 8905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432162092844319e-05,
+      "loss": 0.5958,
+      "step": 8906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432097973839446e-05,
+      "loss": 0.6045,
+      "step": 8907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.432033854834573e-05,
+      "loss": 0.5737,
+      "step": 8908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4319697358297004e-05,
+      "loss": 0.5282,
+      "step": 8909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431905616824827e-05,
+      "loss": 0.5896,
+      "step": 8910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431841497819954e-05,
+      "loss": 0.5236,
+      "step": 8911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431777378815081e-05,
+      "loss": 0.5974,
+      "step": 8912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431713259810208e-05,
+      "loss": 0.5656,
+      "step": 8913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431649140805335e-05,
+      "loss": 0.6029,
+      "step": 8914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431585021800462e-05,
+      "loss": 0.5781,
+      "step": 8915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431520902795589e-05,
+      "loss": 0.5721,
+      "step": 8916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431456783790716e-05,
+      "loss": 0.5506,
+      "step": 8917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431392664785843e-05,
+      "loss": 0.5428,
+      "step": 8918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4313285457809694e-05,
+      "loss": 0.5391,
+      "step": 8919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4312644267760966e-05,
+      "loss": 0.5348,
+      "step": 8920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431200307771224e-05,
+      "loss": 0.6221,
+      "step": 8921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43113618876635e-05,
+      "loss": 0.6339,
+      "step": 8922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4310720697614774e-05,
+      "loss": 0.5758,
+      "step": 8923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4310079507566046e-05,
+      "loss": 0.6146,
+      "step": 8924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430943831751732e-05,
+      "loss": 0.6444,
+      "step": 8925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430879712746858e-05,
+      "loss": 0.5334,
+      "step": 8926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4308155937419854e-05,
+      "loss": 0.595,
+      "step": 8927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430751474737112e-05,
+      "loss": 0.6691,
+      "step": 8928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430687355732239e-05,
+      "loss": 0.6233,
+      "step": 8929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430623236727366e-05,
+      "loss": 0.5665,
+      "step": 8930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4305591177224934e-05,
+      "loss": 0.5113,
+      "step": 8931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4304949987176206e-05,
+      "loss": 0.6325,
+      "step": 8932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430430879712747e-05,
+      "loss": 0.5382,
+      "step": 8933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430366760707874e-05,
+      "loss": 0.6223,
+      "step": 8934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430302641703001e-05,
+      "loss": 0.5443,
+      "step": 8935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430238522698128e-05,
+      "loss": 0.5993,
+      "step": 8936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4301744036932544e-05,
+      "loss": 0.5736,
+      "step": 8937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4301102846883816e-05,
+      "loss": 0.5404,
+      "step": 8938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.430046165683509e-05,
+      "loss": 0.5824,
+      "step": 8939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429982046678636e-05,
+      "loss": 0.5929,
+      "step": 8940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429917927673763e-05,
+      "loss": 0.5532,
+      "step": 8941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4298538086688896e-05,
+      "loss": 0.6189,
+      "step": 8942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429789689664017e-05,
+      "loss": 0.6164,
+      "step": 8943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429725570659143e-05,
+      "loss": 0.5661,
+      "step": 8944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4296614516542704e-05,
+      "loss": 0.5422,
+      "step": 8945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429597332649397e-05,
+      "loss": 0.5947,
+      "step": 8946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429533213644525e-05,
+      "loss": 0.5458,
+      "step": 8947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429469094639652e-05,
+      "loss": 0.6398,
+      "step": 8948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4294049756347784e-05,
+      "loss": 0.6646,
+      "step": 8949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4293408566299056e-05,
+      "loss": 0.5751,
+      "step": 8950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429276737625032e-05,
+      "loss": 0.6475,
+      "step": 8951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429212618620159e-05,
+      "loss": 0.5174,
+      "step": 8952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429148499615286e-05,
+      "loss": 0.6546,
+      "step": 8953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429084380610413e-05,
+      "loss": 0.5993,
+      "step": 8954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.42902026160554e-05,
+      "loss": 0.5987,
+      "step": 8955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428956142600667e-05,
+      "loss": 0.5784,
+      "step": 8956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4288920235957945e-05,
+      "loss": 0.5869,
+      "step": 8957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428827904590921e-05,
+      "loss": 0.6017,
+      "step": 8958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428763785586048e-05,
+      "loss": 0.4574,
+      "step": 8959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4286996665811746e-05,
+      "loss": 0.5734,
+      "step": 8960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428635547576302e-05,
+      "loss": 0.514,
+      "step": 8961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428571428571428e-05,
+      "loss": 0.5665,
+      "step": 8962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428507309566556e-05,
+      "loss": 0.5148,
+      "step": 8963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4284431905616826e-05,
+      "loss": 0.6107,
+      "step": 8964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.42837907155681e-05,
+      "loss": 0.5437,
+      "step": 8965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428314952551937e-05,
+      "loss": 0.6195,
+      "step": 8966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4282508335470635e-05,
+      "loss": 0.6468,
+      "step": 8967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4281867145421906e-05,
+      "loss": 0.4928,
+      "step": 8968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428122595537317e-05,
+      "loss": 0.5522,
+      "step": 8969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.428058476532444e-05,
+      "loss": 0.6571,
+      "step": 8970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4279943575275715e-05,
+      "loss": 0.4876,
+      "step": 8971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4279302385226986e-05,
+      "loss": 0.5953,
+      "step": 8972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.427866119517825e-05,
+      "loss": 0.5776,
+      "step": 8973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427802000512952e-05,
+      "loss": 0.5726,
+      "step": 8974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4277378815080795e-05,
+      "loss": 0.6528,
+      "step": 8975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427673762503206e-05,
+      "loss": 0.6299,
+      "step": 8976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427609643498333e-05,
+      "loss": 0.6474,
+      "step": 8977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4275455244934596e-05,
+      "loss": 0.6463,
+      "step": 8978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427481405488587e-05,
+      "loss": 0.6528,
+      "step": 8979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427417286483714e-05,
+      "loss": 0.5221,
+      "step": 8980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427353167478841e-05,
+      "loss": 0.5521,
+      "step": 8981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4272890484739676e-05,
+      "loss": 0.5462,
+      "step": 8982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427224929469095e-05,
+      "loss": 0.5861,
+      "step": 8983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427160810464222e-05,
+      "loss": 0.6044,
+      "step": 8984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4270966914593485e-05,
+      "loss": 0.5208,
+      "step": 8985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4270325724544757e-05,
+      "loss": 0.5967,
+      "step": 8986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426968453449603e-05,
+      "loss": 0.6221,
+      "step": 8987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42690433444473e-05,
+      "loss": 0.5583,
+      "step": 8988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4268402154398565e-05,
+      "loss": 0.5523,
+      "step": 8989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426776096434984e-05,
+      "loss": 0.5238,
+      "step": 8990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42671197743011e-05,
+      "loss": 0.5288,
+      "step": 8991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426647858425237e-05,
+      "loss": 0.611,
+      "step": 8992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4265837394203645e-05,
+      "loss": 0.6405,
+      "step": 8993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426519620415491e-05,
+      "loss": 0.4891,
+      "step": 8994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426455501410618e-05,
+      "loss": 0.5962,
+      "step": 8995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4263913824057453e-05,
+      "loss": 0.5259,
+      "step": 8996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4263272634008725e-05,
+      "loss": 0.5666,
+      "step": 8997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426263144395999e-05,
+      "loss": 0.5251,
+      "step": 8998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426199025391126e-05,
+      "loss": 0.6714,
+      "step": 8999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426134906386253e-05,
+      "loss": 0.5852,
+      "step": 9000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42607078738138e-05,
+      "loss": 0.6663,
+      "step": 9001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426006668376507e-05,
+      "loss": 0.5352,
+      "step": 9002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425942549371634e-05,
+      "loss": 0.5897,
+      "step": 9003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4258784303667614e-05,
+      "loss": 0.6067,
+      "step": 9004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425814311361888e-05,
+      "loss": 0.6009,
+      "step": 9005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425750192357015e-05,
+      "loss": 0.5639,
+      "step": 9006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4256860733521415e-05,
+      "loss": 0.6072,
+      "step": 9007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425621954347269e-05,
+      "loss": 0.5925,
+      "step": 9008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425557835342395e-05,
+      "loss": 0.5288,
+      "step": 9009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4254937163375224e-05,
+      "loss": 0.6382,
+      "step": 9010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4254295973326495e-05,
+      "loss": 0.5468,
+      "step": 9011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425365478327777e-05,
+      "loss": 0.5319,
+      "step": 9012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425301359322904e-05,
+      "loss": 0.5536,
+      "step": 9013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4252372403180304e-05,
+      "loss": 0.6033,
+      "step": 9014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4251731213131575e-05,
+      "loss": 0.5423,
+      "step": 9015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425109002308284e-05,
+      "loss": 0.4626,
+      "step": 9016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425044883303411e-05,
+      "loss": 0.6654,
+      "step": 9017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424980764298538e-05,
+      "loss": 0.5885,
+      "step": 9018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4249166452936655e-05,
+      "loss": 0.653,
+      "step": 9019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424852526288793e-05,
+      "loss": 0.5982,
+      "step": 9020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424788407283919e-05,
+      "loss": 0.563,
+      "step": 9021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4247242882790464e-05,
+      "loss": 0.5913,
+      "step": 9022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424660169274173e-05,
+      "loss": 0.5659,
+      "step": 9023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4245960502693e-05,
+      "loss": 0.6231,
+      "step": 9024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4245319312644265e-05,
+      "loss": 0.5678,
+      "step": 9025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424467812259554e-05,
+      "loss": 0.5146,
+      "step": 9026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424403693254681e-05,
+      "loss": 0.5865,
+      "step": 9027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424339574249808e-05,
+      "loss": 0.5339,
+      "step": 9028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424275455244935e-05,
+      "loss": 0.5308,
+      "step": 9029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424211336240062e-05,
+      "loss": 0.6516,
+      "step": 9030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.424147217235189e-05,
+      "loss": 0.5952,
+      "step": 9031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4240830982303154e-05,
+      "loss": 0.6467,
+      "step": 9032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4240189792254426e-05,
+      "loss": 0.582,
+      "step": 9033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423954860220569e-05,
+      "loss": 0.4656,
+      "step": 9034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423890741215697e-05,
+      "loss": 0.5585,
+      "step": 9035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4238266222108234e-05,
+      "loss": 0.4733,
+      "step": 9036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4237625032059506e-05,
+      "loss": 0.5201,
+      "step": 9037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423698384201078e-05,
+      "loss": 0.6286,
+      "step": 9038
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423634265196204e-05,
+      "loss": 0.5455,
+      "step": 9039
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4235701461913314e-05,
+      "loss": 0.6422,
+      "step": 9040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423506027186458e-05,
+      "loss": 0.5765,
+      "step": 9041
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423441908181585e-05,
+      "loss": 0.6116,
+      "step": 9042
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423377789176712e-05,
+      "loss": 0.4873,
+      "step": 9043
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4233136701718394e-05,
+      "loss": 0.6701,
+      "step": 9044
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4232495511669666e-05,
+      "loss": 0.6306,
+      "step": 9045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423185432162093e-05,
+      "loss": 0.5308,
+      "step": 9046
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42312131315722e-05,
+      "loss": 0.5996,
+      "step": 9047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423057194152347e-05,
+      "loss": 0.5734,
+      "step": 9048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422993075147474e-05,
+      "loss": 0.5504,
+      "step": 9049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4229289561426004e-05,
+      "loss": 0.5099,
+      "step": 9050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4228648371377276e-05,
+      "loss": 0.5517,
+      "step": 9051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422800718132855e-05,
+      "loss": 0.5632,
+      "step": 9052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422736599127982e-05,
+      "loss": 0.4644,
+      "step": 9053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422672480123109e-05,
+      "loss": 0.5442,
+      "step": 9054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4226083611182356e-05,
+      "loss": 0.5222,
+      "step": 9055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422544242113363e-05,
+      "loss": 0.5001,
+      "step": 9056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422480123108489e-05,
+      "loss": 0.5483,
+      "step": 9057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4224160041036164e-05,
+      "loss": 0.498,
+      "step": 9058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4223518850987436e-05,
+      "loss": 0.5663,
+      "step": 9059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422287766093871e-05,
+      "loss": 0.6125,
+      "step": 9060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422223647088997e-05,
+      "loss": 0.5647,
+      "step": 9061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4221595280841244e-05,
+      "loss": 0.6316,
+      "step": 9062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4220954090792516e-05,
+      "loss": 0.5232,
+      "step": 9063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422031290074378e-05,
+      "loss": 0.5933,
+      "step": 9064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421967171069505e-05,
+      "loss": 0.6423,
+      "step": 9065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421903052064632e-05,
+      "loss": 0.4689,
+      "step": 9066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421838933059759e-05,
+      "loss": 0.6972,
+      "step": 9067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421774814054886e-05,
+      "loss": 0.5566,
+      "step": 9068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421710695050013e-05,
+      "loss": 0.5578,
+      "step": 9069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42164657604514e-05,
+      "loss": 0.5646,
+      "step": 9070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421582457040267e-05,
+      "loss": 0.5979,
+      "step": 9071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421518338035394e-05,
+      "loss": 0.6261,
+      "step": 9072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4214542190305206e-05,
+      "loss": 0.6917,
+      "step": 9073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421390100025648e-05,
+      "loss": 0.5388,
+      "step": 9074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421325981020775e-05,
+      "loss": 0.5251,
+      "step": 9075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421261862015902e-05,
+      "loss": 0.5706,
+      "step": 9076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4211977430110286e-05,
+      "loss": 0.5575,
+      "step": 9077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421133624006156e-05,
+      "loss": 0.4916,
+      "step": 9078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421069505001282e-05,
+      "loss": 0.4738,
+      "step": 9079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4210053859964095e-05,
+      "loss": 0.5042,
+      "step": 9080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4209412669915366e-05,
+      "loss": 0.6254,
+      "step": 9081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420877147986663e-05,
+      "loss": 0.5618,
+      "step": 9082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42081302898179e-05,
+      "loss": 0.5895,
+      "step": 9083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4207489099769175e-05,
+      "loss": 0.578,
+      "step": 9084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4206847909720446e-05,
+      "loss": 0.6778,
+      "step": 9085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420620671967171e-05,
+      "loss": 0.5139,
+      "step": 9086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420556552962298e-05,
+      "loss": 0.4865,
+      "step": 9087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420492433957425e-05,
+      "loss": 0.6093,
+      "step": 9088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420428314952552e-05,
+      "loss": 0.5306,
+      "step": 9089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420364195947679e-05,
+      "loss": 0.6052,
+      "step": 9090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420300076942806e-05,
+      "loss": 0.4233,
+      "step": 9091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4202359579379335e-05,
+      "loss": 0.5492,
+      "step": 9092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42017183893306e-05,
+      "loss": 0.6347,
+      "step": 9093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420107719928187e-05,
+      "loss": 0.5181,
+      "step": 9094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4200436009233137e-05,
+      "loss": 0.6536,
+      "step": 9095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419979481918441e-05,
+      "loss": 0.6164,
+      "step": 9096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419915362913567e-05,
+      "loss": 0.4769,
+      "step": 9097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4198512439086945e-05,
+      "loss": 0.5121,
+      "step": 9098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4197871249038217e-05,
+      "loss": 0.5362,
+      "step": 9099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419723005898949e-05,
+      "loss": 0.5881,
+      "step": 9100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419658886894076e-05,
+      "loss": 0.5296,
+      "step": 9101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4195947678892025e-05,
+      "loss": 0.4987,
+      "step": 9102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41953064888433e-05,
+      "loss": 0.6525,
+      "step": 9103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419466529879456e-05,
+      "loss": 0.5988,
+      "step": 9104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419402410874583e-05,
+      "loss": 0.5019,
+      "step": 9105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41933829186971e-05,
+      "loss": 0.6468,
+      "step": 9106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419274172864838e-05,
+      "loss": 0.5332,
+      "step": 9107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419210053859965e-05,
+      "loss": 0.5515,
+      "step": 9108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4191459348550913e-05,
+      "loss": 0.6113,
+      "step": 9109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4190818158502185e-05,
+      "loss": 0.4531,
+      "step": 9110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419017696845345e-05,
+      "loss": 0.6312,
+      "step": 9111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418953577840472e-05,
+      "loss": 0.6958,
+      "step": 9112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418889458835599e-05,
+      "loss": 0.6942,
+      "step": 9113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418825339830726e-05,
+      "loss": 0.6334,
+      "step": 9114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418761220825853e-05,
+      "loss": 0.7115,
+      "step": 9115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41869710182098e-05,
+      "loss": 0.5723,
+      "step": 9116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4186329828161074e-05,
+      "loss": 0.5612,
+      "step": 9117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418568863811234e-05,
+      "loss": 0.5672,
+      "step": 9118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418504744806361e-05,
+      "loss": 0.5801,
+      "step": 9119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4184406258014875e-05,
+      "loss": 0.6344,
+      "step": 9120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418376506796615e-05,
+      "loss": 0.592,
+      "step": 9121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418312387791741e-05,
+      "loss": 0.5755,
+      "step": 9122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418248268786869e-05,
+      "loss": 0.5869,
+      "step": 9123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4181841497819955e-05,
+      "loss": 0.5607,
+      "step": 9124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418120030777123e-05,
+      "loss": 0.557,
+      "step": 9125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41805591177225e-05,
+      "loss": 0.6241,
+      "step": 9126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4179917927673764e-05,
+      "loss": 0.499,
+      "step": 9127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4179276737625035e-05,
+      "loss": 0.5718,
+      "step": 9128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41786355475763e-05,
+      "loss": 0.5437,
+      "step": 9129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417799435752757e-05,
+      "loss": 0.6543,
+      "step": 9130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4177353167478844e-05,
+      "loss": 0.5918,
+      "step": 9131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4176711977430115e-05,
+      "loss": 0.6151,
+      "step": 9132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417607078738138e-05,
+      "loss": 0.6329,
+      "step": 9133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417542959733265e-05,
+      "loss": 0.5926,
+      "step": 9134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4174788407283924e-05,
+      "loss": 0.5559,
+      "step": 9135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417414721723519e-05,
+      "loss": 0.5785,
+      "step": 9136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417350602718646e-05,
+      "loss": 0.5926,
+      "step": 9137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4172864837137725e-05,
+      "loss": 0.5805,
+      "step": 9138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4172223647089e-05,
+      "loss": 0.5047,
+      "step": 9139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417158245704027e-05,
+      "loss": 0.5756,
+      "step": 9140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.417094126699154e-05,
+      "loss": 0.5835,
+      "step": 9141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4170300076942806e-05,
+      "loss": 0.5451,
+      "step": 9142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416965888689408e-05,
+      "loss": 0.5246,
+      "step": 9143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416901769684535e-05,
+      "loss": 0.4877,
+      "step": 9144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4168376506796614e-05,
+      "loss": 0.4968,
+      "step": 9145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4167735316747886e-05,
+      "loss": 0.5903,
+      "step": 9146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416709412669916e-05,
+      "loss": 0.4729,
+      "step": 9147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416645293665043e-05,
+      "loss": 0.5702,
+      "step": 9148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4165811746601694e-05,
+      "loss": 0.6017,
+      "step": 9149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4165170556552966e-05,
+      "loss": 0.5627,
+      "step": 9150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416452936650423e-05,
+      "loss": 0.4163,
+      "step": 9151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41638881764555e-05,
+      "loss": 0.6079,
+      "step": 9152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4163246986406774e-05,
+      "loss": 0.5663,
+      "step": 9153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416260579635804e-05,
+      "loss": 0.6231,
+      "step": 9154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416196460630931e-05,
+      "loss": 0.6465,
+      "step": 9155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416132341626058e-05,
+      "loss": 0.5468,
+      "step": 9156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4160682226211854e-05,
+      "loss": 0.5239,
+      "step": 9157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416004103616312e-05,
+      "loss": 0.5425,
+      "step": 9158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415939984611439e-05,
+      "loss": 0.5695,
+      "step": 9159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4158758656065656e-05,
+      "loss": 0.6037,
+      "step": 9160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415811746601693e-05,
+      "loss": 0.5651,
+      "step": 9161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41574762759682e-05,
+      "loss": 0.5638,
+      "step": 9162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415683508591947e-05,
+      "loss": 0.5647,
+      "step": 9163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415619389587074e-05,
+      "loss": 0.5462,
+      "step": 9164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415555270582201e-05,
+      "loss": 0.6058,
+      "step": 9165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415491151577328e-05,
+      "loss": 0.5374,
+      "step": 9166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4154270325724544e-05,
+      "loss": 0.5093,
+      "step": 9167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4153629135675816e-05,
+      "loss": 0.5379,
+      "step": 9168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415298794562708e-05,
+      "loss": 0.6256,
+      "step": 9169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415234675557835e-05,
+      "loss": 0.5655,
+      "step": 9170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4151705565529624e-05,
+      "loss": 0.5761,
+      "step": 9171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4151064375480896e-05,
+      "loss": 0.5975,
+      "step": 9172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.415042318543217e-05,
+      "loss": 0.607,
+      "step": 9173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414978199538343e-05,
+      "loss": 0.5372,
+      "step": 9174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4149140805334704e-05,
+      "loss": 0.5186,
+      "step": 9175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414849961528597e-05,
+      "loss": 0.4994,
+      "step": 9176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414785842523724e-05,
+      "loss": 0.6624,
+      "step": 9177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4147217235188506e-05,
+      "loss": 0.5441,
+      "step": 9178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4146576045139785e-05,
+      "loss": 0.6016,
+      "step": 9179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4145934855091056e-05,
+      "loss": 0.5469,
+      "step": 9180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414529366504232e-05,
+      "loss": 0.5187,
+      "step": 9181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414465247499359e-05,
+      "loss": 0.5723,
+      "step": 9182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414401128494486e-05,
+      "loss": 0.6361,
+      "step": 9183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414337009489613e-05,
+      "loss": 0.474,
+      "step": 9184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4142728904847394e-05,
+      "loss": 0.5982,
+      "step": 9185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4142087714798666e-05,
+      "loss": 0.5752,
+      "step": 9186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414144652474994e-05,
+      "loss": 0.5798,
+      "step": 9187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414080533470121e-05,
+      "loss": 0.5651,
+      "step": 9188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414016414465248e-05,
+      "loss": 0.4885,
+      "step": 9189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4139522954603746e-05,
+      "loss": 0.6284,
+      "step": 9190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413888176455502e-05,
+      "loss": 0.6554,
+      "step": 9191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413824057450628e-05,
+      "loss": 0.504,
+      "step": 9192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4137599384457555e-05,
+      "loss": 0.6471,
+      "step": 9193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413695819440882e-05,
+      "loss": 0.5617,
+      "step": 9194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41363170043601e-05,
+      "loss": 0.5948,
+      "step": 9195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413567581431136e-05,
+      "loss": 0.562,
+      "step": 9196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4135034624262635e-05,
+      "loss": 0.5022,
+      "step": 9197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4134393434213906e-05,
+      "loss": 0.6018,
+      "step": 9198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413375224416517e-05,
+      "loss": 0.492,
+      "step": 9199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413311105411644e-05,
+      "loss": 0.6347,
+      "step": 9200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413246986406771e-05,
+      "loss": 0.5562,
+      "step": 9201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413182867401898e-05,
+      "loss": 0.548,
+      "step": 9202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413118748397025e-05,
+      "loss": 0.6217,
+      "step": 9203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.413054629392152e-05,
+      "loss": 0.6226,
+      "step": 9204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412990510387279e-05,
+      "loss": 0.5251,
+      "step": 9205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412926391382406e-05,
+      "loss": 0.5225,
+      "step": 9206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412862272377533e-05,
+      "loss": 0.5629,
+      "step": 9207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4127981533726597e-05,
+      "loss": 0.5678,
+      "step": 9208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412734034367787e-05,
+      "loss": 0.6065,
+      "step": 9209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412669915362913e-05,
+      "loss": 0.6257,
+      "step": 9210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412605796358041e-05,
+      "loss": 0.5655,
+      "step": 9211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412541677353168e-05,
+      "loss": 0.6963,
+      "step": 9212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412477558348295e-05,
+      "loss": 0.5042,
+      "step": 9213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412413439343421e-05,
+      "loss": 0.5133,
+      "step": 9214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4123493203385485e-05,
+      "loss": 0.6073,
+      "step": 9215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412285201333676e-05,
+      "loss": 0.4683,
+      "step": 9216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412221082328802e-05,
+      "loss": 0.5042,
+      "step": 9217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412156963323929e-05,
+      "loss": 0.5841,
+      "step": 9218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4120928443190565e-05,
+      "loss": 0.5625,
+      "step": 9219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412028725314184e-05,
+      "loss": 0.5934,
+      "step": 9220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41196460630931e-05,
+      "loss": 0.5673,
+      "step": 9221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4119004873044373e-05,
+      "loss": 0.5785,
+      "step": 9222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411836368299564e-05,
+      "loss": 0.5335,
+      "step": 9223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411772249294691e-05,
+      "loss": 0.5535,
+      "step": 9224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411708130289818e-05,
+      "loss": 0.6375,
+      "step": 9225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411644011284945e-05,
+      "loss": 0.6777,
+      "step": 9226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411579892280072e-05,
+      "loss": 0.5568,
+      "step": 9227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411515773275199e-05,
+      "loss": 0.6349,
+      "step": 9228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411451654270326e-05,
+      "loss": 0.6233,
+      "step": 9229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411387535265453e-05,
+      "loss": 0.5441,
+      "step": 9230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41132341626058e-05,
+      "loss": 0.568,
+      "step": 9231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.411259297255707e-05,
+      "loss": 0.5689,
+      "step": 9232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4111951782508335e-05,
+      "loss": 0.6053,
+      "step": 9233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.411131059245961e-05,
+      "loss": 0.5779,
+      "step": 9234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.411066940241088e-05,
+      "loss": 0.5599,
+      "step": 9235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.411002821236215e-05,
+      "loss": 0.596,
+      "step": 9236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4109387022313415e-05,
+      "loss": 0.5414,
+      "step": 9237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410874583226469e-05,
+      "loss": 0.5403,
+      "step": 9238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410810464221595e-05,
+      "loss": 0.5998,
+      "step": 9239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4107463452167224e-05,
+      "loss": 0.5374,
+      "step": 9240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4106822262118495e-05,
+      "loss": 0.5637,
+      "step": 9241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410618107206976e-05,
+      "loss": 0.6064,
+      "step": 9242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410553988202103e-05,
+      "loss": 0.6374,
+      "step": 9243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4104898691972304e-05,
+      "loss": 0.5539,
+      "step": 9244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4104257501923576e-05,
+      "loss": 0.5353,
+      "step": 9245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410361631187484e-05,
+      "loss": 0.5692,
+      "step": 9246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410297512182611e-05,
+      "loss": 0.5501,
+      "step": 9247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410233393177738e-05,
+      "loss": 0.5826,
+      "step": 9248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410169274172865e-05,
+      "loss": 0.5138,
+      "step": 9249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410105155167992e-05,
+      "loss": 0.586,
+      "step": 9250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.410041036163119e-05,
+      "loss": 0.5921,
+      "step": 9251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4099769171582464e-05,
+      "loss": 0.5842,
+      "step": 9252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409912798153373e-05,
+      "loss": 0.6832,
+      "step": 9253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4098486791485e-05,
+      "loss": 0.56,
+      "step": 9254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4097845601436266e-05,
+      "loss": 0.5601,
+      "step": 9255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409720441138754e-05,
+      "loss": 0.578,
+      "step": 9256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40965632213388e-05,
+      "loss": 0.5728,
+      "step": 9257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4095922031290074e-05,
+      "loss": 0.5988,
+      "step": 9258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4095280841241346e-05,
+      "loss": 0.6971,
+      "step": 9259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409463965119262e-05,
+      "loss": 0.5021,
+      "step": 9260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409399846114389e-05,
+      "loss": 0.5845,
+      "step": 9261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4093357271095154e-05,
+      "loss": 0.5933,
+      "step": 9262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4092716081046426e-05,
+      "loss": 0.5751,
+      "step": 9263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409207489099769e-05,
+      "loss": 0.608,
+      "step": 9264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409143370094896e-05,
+      "loss": 0.4675,
+      "step": 9265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409079251090023e-05,
+      "loss": 0.6036,
+      "step": 9266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4090151320851506e-05,
+      "loss": 0.6011,
+      "step": 9267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408951013080278e-05,
+      "loss": 0.531,
+      "step": 9268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408886894075404e-05,
+      "loss": 0.652,
+      "step": 9269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4088227750705314e-05,
+      "loss": 0.5358,
+      "step": 9270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408758656065658e-05,
+      "loss": 0.5678,
+      "step": 9271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408694537060785e-05,
+      "loss": 0.6678,
+      "step": 9272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4086304180559116e-05,
+      "loss": 0.534,
+      "step": 9273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408566299051039e-05,
+      "loss": 0.5792,
+      "step": 9274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408502180046166e-05,
+      "loss": 0.6454,
+      "step": 9275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408438061041293e-05,
+      "loss": 0.5707,
+      "step": 9276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40837394203642e-05,
+      "loss": 0.5194,
+      "step": 9277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408309823031547e-05,
+      "loss": 0.4847,
+      "step": 9278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408245704026674e-05,
+      "loss": 0.5361,
+      "step": 9279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4081815850218004e-05,
+      "loss": 0.5382,
+      "step": 9280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4081174660169276e-05,
+      "loss": 0.635,
+      "step": 9281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408053347012054e-05,
+      "loss": 0.5718,
+      "step": 9282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407989228007182e-05,
+      "loss": 0.5809,
+      "step": 9283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4079251090023084e-05,
+      "loss": 0.6252,
+      "step": 9284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4078609899974356e-05,
+      "loss": 0.4559,
+      "step": 9285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407796870992563e-05,
+      "loss": 0.7329,
+      "step": 9286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407732751987689e-05,
+      "loss": 0.563,
+      "step": 9287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4076686329828164e-05,
+      "loss": 0.5797,
+      "step": 9288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407604513977943e-05,
+      "loss": 0.5879,
+      "step": 9289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40754039497307e-05,
+      "loss": 0.5249,
+      "step": 9290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407476275968197e-05,
+      "loss": 0.5868,
+      "step": 9291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4074121569633245e-05,
+      "loss": 0.5162,
+      "step": 9292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407348037958451e-05,
+      "loss": 0.5285,
+      "step": 9293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407283918953578e-05,
+      "loss": 0.6131,
+      "step": 9294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407219799948705e-05,
+      "loss": 0.6162,
+      "step": 9295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407155680943832e-05,
+      "loss": 0.6831,
+      "step": 9296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407091561938959e-05,
+      "loss": 0.6165,
+      "step": 9297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4070274429340855e-05,
+      "loss": 0.5539,
+      "step": 9298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4069633239292126e-05,
+      "loss": 0.6098,
+      "step": 9299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40689920492434e-05,
+      "loss": 0.6522,
+      "step": 9300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406835085919467e-05,
+      "loss": 0.5255,
+      "step": 9301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4067709669145935e-05,
+      "loss": 0.6155,
+      "step": 9302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4067068479097206e-05,
+      "loss": 0.5068,
+      "step": 9303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406642728904848e-05,
+      "loss": 0.62,
+      "step": 9304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406578609899974e-05,
+      "loss": 0.5457,
+      "step": 9305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4065144908951015e-05,
+      "loss": 0.5366,
+      "step": 9306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4064503718902286e-05,
+      "loss": 0.5811,
+      "step": 9307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406386252885356e-05,
+      "loss": 0.5048,
+      "step": 9308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406322133880482e-05,
+      "loss": 0.613,
+      "step": 9309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4062580148756095e-05,
+      "loss": 0.5253,
+      "step": 9310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406193895870736e-05,
+      "loss": 0.5669,
+      "step": 9311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406129776865863e-05,
+      "loss": 0.5713,
+      "step": 9312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40606565786099e-05,
+      "loss": 0.5417,
+      "step": 9313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406001538856117e-05,
+      "loss": 0.5575,
+      "step": 9314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405937419851244e-05,
+      "loss": 0.5065,
+      "step": 9315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405873300846371e-05,
+      "loss": 0.5171,
+      "step": 9316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405809181841498e-05,
+      "loss": 0.6333,
+      "step": 9317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405745062836625e-05,
+      "loss": 0.6819,
+      "step": 9318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405680943831752e-05,
+      "loss": 0.4751,
+      "step": 9319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4056168248268785e-05,
+      "loss": 0.5264,
+      "step": 9320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4055527058220057e-05,
+      "loss": 0.5423,
+      "step": 9321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405488586817133e-05,
+      "loss": 0.5307,
+      "step": 9322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40542446781226e-05,
+      "loss": 0.5527,
+      "step": 9323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405360348807387e-05,
+      "loss": 0.5023,
+      "step": 9324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405296229802514e-05,
+      "loss": 0.6149,
+      "step": 9325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405232110797641e-05,
+      "loss": 0.4944,
+      "step": 9326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405167991792767e-05,
+      "loss": 0.7054,
+      "step": 9327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4051038727878945e-05,
+      "loss": 0.5857,
+      "step": 9328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405039753783021e-05,
+      "loss": 0.6116,
+      "step": 9329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404975634778148e-05,
+      "loss": 0.5057,
+      "step": 9330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4049115157732753e-05,
+      "loss": 0.418,
+      "step": 9331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4048473967684025e-05,
+      "loss": 0.4906,
+      "step": 9332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40478327776353e-05,
+      "loss": 0.6364,
+      "step": 9333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404719158758656e-05,
+      "loss": 0.517,
+      "step": 9334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4046550397537834e-05,
+      "loss": 0.5238,
+      "step": 9335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40459092074891e-05,
+      "loss": 0.6191,
+      "step": 9336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404526801744037e-05,
+      "loss": 0.5255,
+      "step": 9337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4044626827391635e-05,
+      "loss": 0.5625,
+      "step": 9338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4043985637342914e-05,
+      "loss": 0.603,
+      "step": 9339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4043344447294185e-05,
+      "loss": 0.6767,
+      "step": 9340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404270325724545e-05,
+      "loss": 0.5752,
+      "step": 9341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404206206719672e-05,
+      "loss": 0.5898,
+      "step": 9342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404142087714799e-05,
+      "loss": 0.4674,
+      "step": 9343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404077968709926e-05,
+      "loss": 0.5849,
+      "step": 9344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4040138497050524e-05,
+      "loss": 0.5315,
+      "step": 9345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4039497307001795e-05,
+      "loss": 0.5977,
+      "step": 9346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403885611695307e-05,
+      "loss": 0.5368,
+      "step": 9347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403821492690434e-05,
+      "loss": 0.5228,
+      "step": 9348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403757373685561e-05,
+      "loss": 0.5726,
+      "step": 9349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4036932546806875e-05,
+      "loss": 0.5876,
+      "step": 9350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403629135675815e-05,
+      "loss": 0.4469,
+      "step": 9351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403565016670941e-05,
+      "loss": 0.5683,
+      "step": 9352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4035008976660684e-05,
+      "loss": 0.5117,
+      "step": 9353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403436778661195e-05,
+      "loss": 0.6131,
+      "step": 9354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403372659656323e-05,
+      "loss": 0.5453,
+      "step": 9355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403308540651449e-05,
+      "loss": 0.5063,
+      "step": 9356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4032444216465764e-05,
+      "loss": 0.5377,
+      "step": 9357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4031803026417036e-05,
+      "loss": 0.7024,
+      "step": 9358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40311618363683e-05,
+      "loss": 0.5592,
+      "step": 9359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403052064631957e-05,
+      "loss": 0.5762,
+      "step": 9360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402987945627084e-05,
+      "loss": 0.6076,
+      "step": 9361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402923826622211e-05,
+      "loss": 0.5793,
+      "step": 9362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402859707617338e-05,
+      "loss": 0.5247,
+      "step": 9363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402795588612465e-05,
+      "loss": 0.5346,
+      "step": 9364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402731469607592e-05,
+      "loss": 0.5375,
+      "step": 9365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402667350602719e-05,
+      "loss": 0.5949,
+      "step": 9366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402603231597846e-05,
+      "loss": 0.5478,
+      "step": 9367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4025391125929726e-05,
+      "loss": 0.6779,
+      "step": 9368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4024749935881e-05,
+      "loss": 0.5343,
+      "step": 9369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402410874583226e-05,
+      "loss": 0.6185,
+      "step": 9370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402346755578354e-05,
+      "loss": 0.5288,
+      "step": 9371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4022826365734806e-05,
+      "loss": 0.4614,
+      "step": 9372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402218517568608e-05,
+      "loss": 0.5696,
+      "step": 9373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402154398563734e-05,
+      "loss": 0.6105,
+      "step": 9374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4020902795588614e-05,
+      "loss": 0.6731,
+      "step": 9375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4020261605539886e-05,
+      "loss": 0.4831,
+      "step": 9376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401962041549115e-05,
+      "loss": 0.5793,
+      "step": 9377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401897922544242e-05,
+      "loss": 0.5056,
+      "step": 9378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4018338035393694e-05,
+      "loss": 0.5847,
+      "step": 9379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4017696845344966e-05,
+      "loss": 0.4507,
+      "step": 9380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401705565529623e-05,
+      "loss": 0.5189,
+      "step": 9381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40164144652475e-05,
+      "loss": 0.5944,
+      "step": 9382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401577327519877e-05,
+      "loss": 0.6385,
+      "step": 9383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401513208515004e-05,
+      "loss": 0.4986,
+      "step": 9384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401449089510131e-05,
+      "loss": 0.591,
+      "step": 9385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4013849705052576e-05,
+      "loss": 0.5928,
+      "step": 9386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401320851500385e-05,
+      "loss": 0.4893,
+      "step": 9387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401256732495512e-05,
+      "loss": 0.6728,
+      "step": 9388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401192613490639e-05,
+      "loss": 0.5537,
+      "step": 9389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4011284944857656e-05,
+      "loss": 0.552,
+      "step": 9390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401064375480893e-05,
+      "loss": 0.5846,
+      "step": 9391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401000256476019e-05,
+      "loss": 0.4602,
+      "step": 9392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4009361374711464e-05,
+      "loss": 0.6659,
+      "step": 9393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4008720184662736e-05,
+      "loss": 0.6435,
+      "step": 9394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400807899461401e-05,
+      "loss": 0.6234,
+      "step": 9395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400743780456528e-05,
+      "loss": 0.6965,
+      "step": 9396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4006796614516544e-05,
+      "loss": 0.508,
+      "step": 9397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4006155424467816e-05,
+      "loss": 0.4922,
+      "step": 9398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400551423441908e-05,
+      "loss": 0.5737,
+      "step": 9399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400487304437035e-05,
+      "loss": 0.5543,
+      "step": 9400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400423185432162e-05,
+      "loss": 0.5388,
+      "step": 9401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400359066427289e-05,
+      "loss": 0.6197,
+      "step": 9402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400294947422416e-05,
+      "loss": 0.6082,
+      "step": 9403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400230828417543e-05,
+      "loss": 0.4548,
+      "step": 9404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4001667094126705e-05,
+      "loss": 0.622,
+      "step": 9405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400102590407797e-05,
+      "loss": 0.5456,
+      "step": 9406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400038471402924e-05,
+      "loss": 0.5597,
+      "step": 9407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3999743523980506e-05,
+      "loss": 0.5949,
+      "step": 9408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399910233393178e-05,
+      "loss": 0.5443,
+      "step": 9409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399846114388305e-05,
+      "loss": 0.568,
+      "step": 9410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399781995383432e-05,
+      "loss": 0.5472,
+      "step": 9411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399717876378559e-05,
+      "loss": 0.6077,
+      "step": 9412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399653757373686e-05,
+      "loss": 0.6171,
+      "step": 9413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399589638368813e-05,
+      "loss": 0.5865,
+      "step": 9414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3995255193639395e-05,
+      "loss": 0.5566,
+      "step": 9415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3994614003590666e-05,
+      "loss": 0.6651,
+      "step": 9416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399397281354193e-05,
+      "loss": 0.5921,
+      "step": 9417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.39933316234932e-05,
+      "loss": 0.6212,
+      "step": 9418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3992690433444475e-05,
+      "loss": 0.5679,
+      "step": 9419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3992049243395746e-05,
+      "loss": 0.529,
+      "step": 9420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399140805334702e-05,
+      "loss": 0.6165,
+      "step": 9421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399076686329828e-05,
+      "loss": 0.5766,
+      "step": 9422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3990125673249555e-05,
+      "loss": 0.6037,
+      "step": 9423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398948448320082e-05,
+      "loss": 0.5947,
+      "step": 9424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398884329315209e-05,
+      "loss": 0.5919,
+      "step": 9425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3988202103103356e-05,
+      "loss": 0.6319,
+      "step": 9426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3987560913054635e-05,
+      "loss": 0.5629,
+      "step": 9427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398691972300591e-05,
+      "loss": 0.6236,
+      "step": 9428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398627853295717e-05,
+      "loss": 0.5578,
+      "step": 9429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398563734290844e-05,
+      "loss": 0.6084,
+      "step": 9430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398499615285971e-05,
+      "loss": 0.4476,
+      "step": 9431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398435496281098e-05,
+      "loss": 0.6575,
+      "step": 9432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3983713772762245e-05,
+      "loss": 0.4811,
+      "step": 9433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3983072582713517e-05,
+      "loss": 0.5643,
+      "step": 9434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398243139266479e-05,
+      "loss": 0.5833,
+      "step": 9435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398179020261606e-05,
+      "loss": 0.5701,
+      "step": 9436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.398114901256733e-05,
+      "loss": 0.5758,
+      "step": 9437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.39805078225186e-05,
+      "loss": 0.6513,
+      "step": 9438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397986663246987e-05,
+      "loss": 0.5702,
+      "step": 9439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397922544242113e-05,
+      "loss": 0.5781,
+      "step": 9440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3978584252372405e-05,
+      "loss": 0.5287,
+      "step": 9441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397794306232367e-05,
+      "loss": 0.4945,
+      "step": 9442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397730187227495e-05,
+      "loss": 0.608,
+      "step": 9443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3976660682226213e-05,
+      "loss": 0.6789,
+      "step": 9444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3976019492177485e-05,
+      "loss": 0.4617,
+      "step": 9445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397537830212876e-05,
+      "loss": 0.5639,
+      "step": 9446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397473711208002e-05,
+      "loss": 0.52,
+      "step": 9447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3974095922031294e-05,
+      "loss": 0.548,
+      "step": 9448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397345473198256e-05,
+      "loss": 0.5793,
+      "step": 9449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397281354193383e-05,
+      "loss": 0.6878,
+      "step": 9450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.39721723518851e-05,
+      "loss": 0.4981,
+      "step": 9451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3971531161836374e-05,
+      "loss": 0.6311,
+      "step": 9452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397088997178764e-05,
+      "loss": 0.5654,
+      "step": 9453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397024878173891e-05,
+      "loss": 0.6053,
+      "step": 9454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396960759169018e-05,
+      "loss": 0.5366,
+      "step": 9455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396896640164145e-05,
+      "loss": 0.6149,
+      "step": 9456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396832521159272e-05,
+      "loss": 0.5851,
+      "step": 9457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3967684021543984e-05,
+      "loss": 0.5722,
+      "step": 9458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396704283149526e-05,
+      "loss": 0.5935,
+      "step": 9459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396640164144653e-05,
+      "loss": 0.5347,
+      "step": 9460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.39657604513978e-05,
+      "loss": 0.5762,
+      "step": 9461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3965119261349064e-05,
+      "loss": 0.4949,
+      "step": 9462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3964478071300335e-05,
+      "loss": 0.5862,
+      "step": 9463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396383688125161e-05,
+      "loss": 0.558,
+      "step": 9464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396319569120287e-05,
+      "loss": 0.5964,
+      "step": 9465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3962554501154144e-05,
+      "loss": 0.5496,
+      "step": 9466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3961913311105415e-05,
+      "loss": 0.5065,
+      "step": 9467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396127212105669e-05,
+      "loss": 0.5299,
+      "step": 9468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396063093100795e-05,
+      "loss": 0.6382,
+      "step": 9469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3959989740959224e-05,
+      "loss": 0.5174,
+      "step": 9470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395934855091049e-05,
+      "loss": 0.6052,
+      "step": 9471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395870736086176e-05,
+      "loss": 0.5658,
+      "step": 9472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395806617081303e-05,
+      "loss": 0.553,
+      "step": 9473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.39574249807643e-05,
+      "loss": 0.6136,
+      "step": 9474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395678379071557e-05,
+      "loss": 0.6178,
+      "step": 9475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395614260066684e-05,
+      "loss": 0.6735,
+      "step": 9476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395550141061811e-05,
+      "loss": 0.502,
+      "step": 9477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395486022056938e-05,
+      "loss": 0.5001,
+      "step": 9478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395421903052065e-05,
+      "loss": 0.6693,
+      "step": 9479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3953577840471914e-05,
+      "loss": 0.5865,
+      "step": 9480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3952936650423186e-05,
+      "loss": 0.4968,
+      "step": 9481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395229546037446e-05,
+      "loss": 0.5903,
+      "step": 9482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.395165427032573e-05,
+      "loss": 0.6499,
+      "step": 9483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3951013080277e-05,
+      "loss": 0.6203,
+      "step": 9484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3950371890228266e-05,
+      "loss": 0.5485,
+      "step": 9485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.394973070017954e-05,
+      "loss": 0.6827,
+      "step": 9486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.39490895101308e-05,
+      "loss": 0.5493,
+      "step": 9487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3948448320082074e-05,
+      "loss": 0.5038,
+      "step": 9488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.394780713003334e-05,
+      "loss": 0.497,
+      "step": 9489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.394716593998461e-05,
+      "loss": 0.5672,
+      "step": 9490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.394652474993588e-05,
+      "loss": 0.5414,
+      "step": 9491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3945883559887154e-05,
+      "loss": 0.6298,
+      "step": 9492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3945242369838426e-05,
+      "loss": 0.5477,
+      "step": 9493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394460117978969e-05,
+      "loss": 0.5739,
+      "step": 9494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394395998974096e-05,
+      "loss": 0.5521,
+      "step": 9495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394331879969223e-05,
+      "loss": 0.5899,
+      "step": 9496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.39426776096435e-05,
+      "loss": 0.7203,
+      "step": 9497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394203641959477e-05,
+      "loss": 0.575,
+      "step": 9498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394139522954604e-05,
+      "loss": 0.6602,
+      "step": 9499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3940754039497314e-05,
+      "loss": 0.4822,
+      "step": 9500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394011284944858e-05,
+      "loss": 0.5366,
+      "step": 9501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393947165939985e-05,
+      "loss": 0.5116,
+      "step": 9502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3938830469351116e-05,
+      "loss": 0.509,
+      "step": 9503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393818927930239e-05,
+      "loss": 0.5768,
+      "step": 9504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393754808925365e-05,
+      "loss": 0.5454,
+      "step": 9505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3936906899204924e-05,
+      "loss": 0.6316,
+      "step": 9506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3936265709156196e-05,
+      "loss": 0.5434,
+      "step": 9507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393562451910747e-05,
+      "loss": 0.5579,
+      "step": 9508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393498332905874e-05,
+      "loss": 0.5199,
+      "step": 9509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3934342139010004e-05,
+      "loss": 0.6068,
+      "step": 9510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3933700948961276e-05,
+      "loss": 0.502,
+      "step": 9511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393305975891254e-05,
+      "loss": 0.6169,
+      "step": 9512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393241856886381e-05,
+      "loss": 0.6402,
+      "step": 9513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393177737881508e-05,
+      "loss": 0.5549,
+      "step": 9514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3931136188766356e-05,
+      "loss": 0.5513,
+      "step": 9515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393049499871762e-05,
+      "loss": 0.6416,
+      "step": 9516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392985380866889e-05,
+      "loss": 0.5659,
+      "step": 9517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3929212618620165e-05,
+      "loss": 0.6124,
+      "step": 9518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392857142857143e-05,
+      "loss": 0.5883,
+      "step": 9519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.39279302385227e-05,
+      "loss": 0.5319,
+      "step": 9520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3927289048473966e-05,
+      "loss": 0.562,
+      "step": 9521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392664785842524e-05,
+      "loss": 0.5222,
+      "step": 9522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392600666837651e-05,
+      "loss": 0.5152,
+      "step": 9523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392536547832778e-05,
+      "loss": 0.525,
+      "step": 9524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3924724288279046e-05,
+      "loss": 0.5466,
+      "step": 9525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392408309823032e-05,
+      "loss": 0.6033,
+      "step": 9526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392344190818159e-05,
+      "loss": 0.5665,
+      "step": 9527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3922800718132855e-05,
+      "loss": 0.5787,
+      "step": 9528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3922159528084126e-05,
+      "loss": 0.6386,
+      "step": 9529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392151833803539e-05,
+      "loss": 0.4473,
+      "step": 9530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392087714798667e-05,
+      "loss": 0.4643,
+      "step": 9531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3920235957937935e-05,
+      "loss": 0.5677,
+      "step": 9532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3919594767889206e-05,
+      "loss": 0.5942,
+      "step": 9533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391895357784047e-05,
+      "loss": 0.5989,
+      "step": 9534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391831238779174e-05,
+      "loss": 0.5575,
+      "step": 9535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3917671197743015e-05,
+      "loss": 0.54,
+      "step": 9536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391703000769428e-05,
+      "loss": 0.5883,
+      "step": 9537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391638881764555e-05,
+      "loss": 0.5484,
+      "step": 9538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391574762759682e-05,
+      "loss": 0.4656,
+      "step": 9539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3915106437548095e-05,
+      "loss": 0.5898,
+      "step": 9540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391446524749936e-05,
+      "loss": 0.5296,
+      "step": 9541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391382405745063e-05,
+      "loss": 0.4358,
+      "step": 9542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3913182867401897e-05,
+      "loss": 0.6538,
+      "step": 9543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391254167735317e-05,
+      "loss": 0.4963,
+      "step": 9544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391190048730444e-05,
+      "loss": 0.5574,
+      "step": 9545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3911259297255705e-05,
+      "loss": 0.5227,
+      "step": 9546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391061810720698e-05,
+      "loss": 0.5575,
+      "step": 9547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390997691715825e-05,
+      "loss": 0.6103,
+      "step": 9548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390933572710952e-05,
+      "loss": 0.5216,
+      "step": 9549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3908694537060785e-05,
+      "loss": 0.5667,
+      "step": 9550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390805334701206e-05,
+      "loss": 0.4168,
+      "step": 9551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390741215696332e-05,
+      "loss": 0.5973,
+      "step": 9552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3906770966914593e-05,
+      "loss": 0.5961,
+      "step": 9553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3906129776865865e-05,
+      "loss": 0.5796,
+      "step": 9554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390548858681714e-05,
+      "loss": 0.5255,
+      "step": 9555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390484739676841e-05,
+      "loss": 0.6409,
+      "step": 9556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3904206206719673e-05,
+      "loss": 0.6189,
+      "step": 9557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3903565016670945e-05,
+      "loss": 0.5723,
+      "step": 9558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390292382662221e-05,
+      "loss": 0.4899,
+      "step": 9559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390228263657348e-05,
+      "loss": 0.5708,
+      "step": 9560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390164144652475e-05,
+      "loss": 0.5407,
+      "step": 9561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390100025647602e-05,
+      "loss": 0.5669,
+      "step": 9562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390035906642729e-05,
+      "loss": 0.4627,
+      "step": 9563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389971787637856e-05,
+      "loss": 0.5477,
+      "step": 9564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3899076686329834e-05,
+      "loss": 0.6117,
+      "step": 9565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38984354962811e-05,
+      "loss": 0.5721,
+      "step": 9566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389779430623237e-05,
+      "loss": 0.4952,
+      "step": 9567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3897153116183635e-05,
+      "loss": 0.5608,
+      "step": 9568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389651192613491e-05,
+      "loss": 0.5282,
+      "step": 9569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389587073608618e-05,
+      "loss": 0.6349,
+      "step": 9570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389522954603745e-05,
+      "loss": 0.505,
+      "step": 9571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389458835598872e-05,
+      "loss": 0.5481,
+      "step": 9572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389394716593999e-05,
+      "loss": 0.5361,
+      "step": 9573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389330597589126e-05,
+      "loss": 0.6887,
+      "step": 9574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3892664785842524e-05,
+      "loss": 0.6089,
+      "step": 9575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3892023595793795e-05,
+      "loss": 0.5113,
+      "step": 9576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389138240574506e-05,
+      "loss": 0.5813,
+      "step": 9577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389074121569633e-05,
+      "loss": 0.6779,
+      "step": 9578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3890100025647604e-05,
+      "loss": 0.5443,
+      "step": 9579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3889458835598876e-05,
+      "loss": 0.5105,
+      "step": 9580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388881764555015e-05,
+      "loss": 0.5524,
+      "step": 9581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388817645550141e-05,
+      "loss": 0.4871,
+      "step": 9582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3887535265452684e-05,
+      "loss": 0.5757,
+      "step": 9583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388689407540395e-05,
+      "loss": 0.6287,
+      "step": 9584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388625288535522e-05,
+      "loss": 0.5703,
+      "step": 9585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3885611695306485e-05,
+      "loss": 0.5822,
+      "step": 9586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3884970505257764e-05,
+      "loss": 0.6639,
+      "step": 9587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388432931520903e-05,
+      "loss": 0.5129,
+      "step": 9588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38836881251603e-05,
+      "loss": 0.5573,
+      "step": 9589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388304693511157e-05,
+      "loss": 0.5846,
+      "step": 9590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388240574506284e-05,
+      "loss": 0.7084,
+      "step": 9591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388176455501411e-05,
+      "loss": 0.5206,
+      "step": 9592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3881123364965374e-05,
+      "loss": 0.5858,
+      "step": 9593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3880482174916646e-05,
+      "loss": 0.5644,
+      "step": 9594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387984098486792e-05,
+      "loss": 0.4757,
+      "step": 9595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387919979481919e-05,
+      "loss": 0.596,
+      "step": 9596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3878558604770454e-05,
+      "loss": 0.5237,
+      "step": 9597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3877917414721726e-05,
+      "loss": 0.5381,
+      "step": 9598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3877276224673e-05,
+      "loss": 0.6628,
+      "step": 9599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387663503462426e-05,
+      "loss": 0.5813,
+      "step": 9600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3875993844575534e-05,
+      "loss": 0.468,
+      "step": 9601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38753526545268e-05,
+      "loss": 0.625,
+      "step": 9602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387471146447808e-05,
+      "loss": 0.5073,
+      "step": 9603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387407027442934e-05,
+      "loss": 0.5163,
+      "step": 9604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3873429084380614e-05,
+      "loss": 0.6419,
+      "step": 9605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387278789433188e-05,
+      "loss": 0.5155,
+      "step": 9606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387214670428315e-05,
+      "loss": 0.66,
+      "step": 9607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387150551423442e-05,
+      "loss": 0.5532,
+      "step": 9608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387086432418569e-05,
+      "loss": 0.6075,
+      "step": 9609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387022313413696e-05,
+      "loss": 0.5425,
+      "step": 9610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386958194408823e-05,
+      "loss": 0.5396,
+      "step": 9611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38689407540395e-05,
+      "loss": 0.6039,
+      "step": 9612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386829956399077e-05,
+      "loss": 0.4884,
+      "step": 9613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386765837394204e-05,
+      "loss": 0.5086,
+      "step": 9614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386701718389331e-05,
+      "loss": 0.4358,
+      "step": 9615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3866375993844576e-05,
+      "loss": 0.5569,
+      "step": 9616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386573480379585e-05,
+      "loss": 0.6097,
+      "step": 9617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386509361374711e-05,
+      "loss": 0.6122,
+      "step": 9618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386445242369839e-05,
+      "loss": 0.4538,
+      "step": 9619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3863811233649656e-05,
+      "loss": 0.5734,
+      "step": 9620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386317004360093e-05,
+      "loss": 0.5637,
+      "step": 9621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386252885355219e-05,
+      "loss": 0.5748,
+      "step": 9622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3861887663503464e-05,
+      "loss": 0.5225,
+      "step": 9623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3861246473454736e-05,
+      "loss": 0.5442,
+      "step": 9624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3860605283406e-05,
+      "loss": 0.551,
+      "step": 9625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385996409335727e-05,
+      "loss": 0.6664,
+      "step": 9626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3859322903308545e-05,
+      "loss": 0.5812,
+      "step": 9627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3858681713259816e-05,
+      "loss": 0.574,
+      "step": 9628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385804052321108e-05,
+      "loss": 0.6197,
+      "step": 9629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385739933316235e-05,
+      "loss": 0.576,
+      "step": 9630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385675814311362e-05,
+      "loss": 0.6144,
+      "step": 9631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385611695306489e-05,
+      "loss": 0.5403,
+      "step": 9632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385547576301616e-05,
+      "loss": 0.6049,
+      "step": 9633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3854834572967426e-05,
+      "loss": 0.6125,
+      "step": 9634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38541933829187e-05,
+      "loss": 0.5737,
+      "step": 9635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385355219286997e-05,
+      "loss": 0.5348,
+      "step": 9636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385291100282124e-05,
+      "loss": 0.572,
+      "step": 9637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3852269812772506e-05,
+      "loss": 0.5296,
+      "step": 9638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385162862272378e-05,
+      "loss": 0.5863,
+      "step": 9639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385098743267504e-05,
+      "loss": 0.5869,
+      "step": 9640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3850346242626315e-05,
+      "loss": 0.5466,
+      "step": 9641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3849705052577586e-05,
+      "loss": 0.4956,
+      "step": 9642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384906386252886e-05,
+      "loss": 0.6111,
+      "step": 9643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384842267248013e-05,
+      "loss": 0.5032,
+      "step": 9644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3847781482431395e-05,
+      "loss": 0.6063,
+      "step": 9645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3847140292382667e-05,
+      "loss": 0.4822,
+      "step": 9646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384649910233393e-05,
+      "loss": 0.6523,
+      "step": 9647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38458579122852e-05,
+      "loss": 0.5259,
+      "step": 9648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384521672223647e-05,
+      "loss": 0.5407,
+      "step": 9649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384457553218774e-05,
+      "loss": 0.5313,
+      "step": 9650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384393434213901e-05,
+      "loss": 0.613,
+      "step": 9651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384329315209028e-05,
+      "loss": 0.5252,
+      "step": 9652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3842651962041555e-05,
+      "loss": 0.6507,
+      "step": 9653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384201077199282e-05,
+      "loss": 0.6914,
+      "step": 9654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384136958194409e-05,
+      "loss": 0.5759,
+      "step": 9655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3840728391895357e-05,
+      "loss": 0.5838,
+      "step": 9656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.384008720184663e-05,
+      "loss": 0.5415,
+      "step": 9657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38394460117979e-05,
+      "loss": 0.6187,
+      "step": 9658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.383880482174917e-05,
+      "loss": 0.4915,
+      "step": 9659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3838163631700443e-05,
+      "loss": 0.623,
+      "step": 9660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.383752244165171e-05,
+      "loss": 0.6085,
+      "step": 9661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.383688125160298e-05,
+      "loss": 0.6174,
+      "step": 9662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3836240061554245e-05,
+      "loss": 0.5771,
+      "step": 9663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.383559887150552e-05,
+      "loss": 0.552,
+      "step": 9664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.383495768145678e-05,
+      "loss": 0.5642,
+      "step": 9665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3834316491408053e-05,
+      "loss": 0.5699,
+      "step": 9666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3833675301359325e-05,
+      "loss": 0.646,
+      "step": 9667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38330341113106e-05,
+      "loss": 0.6895,
+      "step": 9668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.383239292126187e-05,
+      "loss": 0.4776,
+      "step": 9669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3831751731213134e-05,
+      "loss": 0.5204,
+      "step": 9670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3831110541164405e-05,
+      "loss": 0.6052,
+      "step": 9671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.383046935111567e-05,
+      "loss": 0.5233,
+      "step": 9672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382982816106694e-05,
+      "loss": 0.5938,
+      "step": 9673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382918697101821e-05,
+      "loss": 0.5419,
+      "step": 9674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3828545780969485e-05,
+      "loss": 0.5301,
+      "step": 9675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382790459092075e-05,
+      "loss": 0.6571,
+      "step": 9676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382726340087202e-05,
+      "loss": 0.5774,
+      "step": 9677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3826622210823294e-05,
+      "loss": 0.5406,
+      "step": 9678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382598102077456e-05,
+      "loss": 0.5437,
+      "step": 9679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382533983072583e-05,
+      "loss": 0.6224,
+      "step": 9680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3824698640677095e-05,
+      "loss": 0.5408,
+      "step": 9681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382405745062837e-05,
+      "loss": 0.5249,
+      "step": 9682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382341626057964e-05,
+      "loss": 0.5262,
+      "step": 9683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382277507053091e-05,
+      "loss": 0.5413,
+      "step": 9684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3822133880482175e-05,
+      "loss": 0.6076,
+      "step": 9685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382149269043345e-05,
+      "loss": 0.6594,
+      "step": 9686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382085150038472e-05,
+      "loss": 0.5409,
+      "step": 9687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3820210310335984e-05,
+      "loss": 0.626,
+      "step": 9688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3819569120287255e-05,
+      "loss": 0.5779,
+      "step": 9689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381892793023852e-05,
+      "loss": 0.6202,
+      "step": 9690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38182867401898e-05,
+      "loss": 0.5518,
+      "step": 9691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3817645550141064e-05,
+      "loss": 0.5631,
+      "step": 9692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3817004360092336e-05,
+      "loss": 0.5094,
+      "step": 9693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38163631700436e-05,
+      "loss": 0.7,
+      "step": 9694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381572197999487e-05,
+      "loss": 0.5744,
+      "step": 9695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3815080789946144e-05,
+      "loss": 0.6047,
+      "step": 9696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381443959989741e-05,
+      "loss": 0.4863,
+      "step": 9697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381379840984868e-05,
+      "loss": 0.5898,
+      "step": 9698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381315721979995e-05,
+      "loss": 0.5559,
+      "step": 9699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3812516029751224e-05,
+      "loss": 0.533,
+      "step": 9700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381187483970249e-05,
+      "loss": 0.6015,
+      "step": 9701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381123364965376e-05,
+      "loss": 0.635,
+      "step": 9702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3810592459605026e-05,
+      "loss": 0.5683,
+      "step": 9703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38099512695563e-05,
+      "loss": 0.5396,
+      "step": 9704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380931007950757e-05,
+      "loss": 0.5002,
+      "step": 9705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3808668889458834e-05,
+      "loss": 0.4738,
+      "step": 9706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3808027699410106e-05,
+      "loss": 0.5655,
+      "step": 9707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380738650936138e-05,
+      "loss": 0.5361,
+      "step": 9708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380674531931265e-05,
+      "loss": 0.6336,
+      "step": 9709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3806104129263914e-05,
+      "loss": 0.53,
+      "step": 9710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3805462939215186e-05,
+      "loss": 0.6147,
+      "step": 9711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380482174916645e-05,
+      "loss": 0.5792,
+      "step": 9712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380418055911772e-05,
+      "loss": 0.5322,
+      "step": 9713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3803539369068994e-05,
+      "loss": 0.5806,
+      "step": 9714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3802898179020266e-05,
+      "loss": 0.4737,
+      "step": 9715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380225698897154e-05,
+      "loss": 0.6595,
+      "step": 9716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38016157989228e-05,
+      "loss": 0.6054,
+      "step": 9717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3800974608874074e-05,
+      "loss": 0.6057,
+      "step": 9718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.380033341882534e-05,
+      "loss": 0.5857,
+      "step": 9719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379969222877661e-05,
+      "loss": 0.4941,
+      "step": 9720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3799051038727876e-05,
+      "loss": 0.4962,
+      "step": 9721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379840984867915e-05,
+      "loss": 0.4627,
+      "step": 9722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379776865863042e-05,
+      "loss": 0.6207,
+      "step": 9723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379712746858169e-05,
+      "loss": 0.5814,
+      "step": 9724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379648627853296e-05,
+      "loss": 0.5417,
+      "step": 9725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379584508848423e-05,
+      "loss": 0.5017,
+      "step": 9726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.37952038984355e-05,
+      "loss": 0.5128,
+      "step": 9727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3794562708386764e-05,
+      "loss": 0.5236,
+      "step": 9728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3793921518338036e-05,
+      "loss": 0.665,
+      "step": 9729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379328032828931e-05,
+      "loss": 0.5147,
+      "step": 9730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379263913824058e-05,
+      "loss": 0.5848,
+      "step": 9731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379199794819185e-05,
+      "loss": 0.5964,
+      "step": 9732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3791356758143116e-05,
+      "loss": 0.5501,
+      "step": 9733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379071556809439e-05,
+      "loss": 0.5951,
+      "step": 9734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.379007437804565e-05,
+      "loss": 0.5618,
+      "step": 9735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3789433187996925e-05,
+      "loss": 0.5945,
+      "step": 9736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378879199794819e-05,
+      "loss": 0.5415,
+      "step": 9737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378815080789946e-05,
+      "loss": 0.6327,
+      "step": 9738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378750961785073e-05,
+      "loss": 0.5689,
+      "step": 9739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3786868427802005e-05,
+      "loss": 0.6256,
+      "step": 9740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3786227237753276e-05,
+      "loss": 0.5701,
+      "step": 9741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378558604770454e-05,
+      "loss": 0.5568,
+      "step": 9742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378494485765581e-05,
+      "loss": 0.5561,
+      "step": 9743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378430366760708e-05,
+      "loss": 0.6252,
+      "step": 9744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378366247755835e-05,
+      "loss": 0.5428,
+      "step": 9745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378302128750962e-05,
+      "loss": 0.5335,
+      "step": 9746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378238009746089e-05,
+      "loss": 0.615,
+      "step": 9747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378173890741216e-05,
+      "loss": 0.6174,
+      "step": 9748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378109771736343e-05,
+      "loss": 0.5468,
+      "step": 9749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.37804565273147e-05,
+      "loss": 0.6111,
+      "step": 9750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3779815337265966e-05,
+      "loss": 0.5068,
+      "step": 9751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.377917414721724e-05,
+      "loss": 0.5104,
+      "step": 9752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.37785329571685e-05,
+      "loss": 0.5458,
+      "step": 9753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3777891767119775e-05,
+      "loss": 0.504,
+      "step": 9754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3777250577071046e-05,
+      "loss": 0.5602,
+      "step": 9755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377660938702232e-05,
+      "loss": 0.5389,
+      "step": 9756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377596819697358e-05,
+      "loss": 0.5406,
+      "step": 9757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3775327006924855e-05,
+      "loss": 0.5531,
+      "step": 9758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3774685816876127e-05,
+      "loss": 0.5894,
+      "step": 9759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377404462682739e-05,
+      "loss": 0.5646,
+      "step": 9760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377340343677866e-05,
+      "loss": 0.6152,
+      "step": 9761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377276224672993e-05,
+      "loss": 0.5128,
+      "step": 9762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377212105668121e-05,
+      "loss": 0.5346,
+      "step": 9763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377147986663247e-05,
+      "loss": 0.5487,
+      "step": 9764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377083867658374e-05,
+      "loss": 0.5218,
+      "step": 9765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377019748653501e-05,
+      "loss": 0.5096,
+      "step": 9766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376955629648628e-05,
+      "loss": 0.5304,
+      "step": 9767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376891510643755e-05,
+      "loss": 0.5219,
+      "step": 9768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376827391638882e-05,
+      "loss": 0.5267,
+      "step": 9769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376763272634009e-05,
+      "loss": 0.5896,
+      "step": 9770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376699153629136e-05,
+      "loss": 0.5656,
+      "step": 9771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376635034624263e-05,
+      "loss": 0.5585,
+      "step": 9772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37657091561939e-05,
+      "loss": 0.6023,
+      "step": 9773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376506796614517e-05,
+      "loss": 0.5344,
+      "step": 9774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376442677609643e-05,
+      "loss": 0.4962,
+      "step": 9775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3763785586047705e-05,
+      "loss": 0.5631,
+      "step": 9776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376314439599898e-05,
+      "loss": 0.5663,
+      "step": 9777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376250320595024e-05,
+      "loss": 0.6351,
+      "step": 9778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376186201590152e-05,
+      "loss": 0.4507,
+      "step": 9779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3761220825852785e-05,
+      "loss": 0.5411,
+      "step": 9780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376057963580406e-05,
+      "loss": 0.6491,
+      "step": 9781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375993844575532e-05,
+      "loss": 0.4307,
+      "step": 9782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3759297255706594e-05,
+      "loss": 0.4473,
+      "step": 9783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375865606565786e-05,
+      "loss": 0.6395,
+      "step": 9784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375801487560913e-05,
+      "loss": 0.5466,
+      "step": 9785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37573736855604e-05,
+      "loss": 0.5728,
+      "step": 9786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3756732495511674e-05,
+      "loss": 0.7599,
+      "step": 9787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3756091305462945e-05,
+      "loss": 0.618,
+      "step": 9788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375545011541421e-05,
+      "loss": 0.4574,
+      "step": 9789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375480892536548e-05,
+      "loss": 0.4724,
+      "step": 9790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375416773531675e-05,
+      "loss": 0.5329,
+      "step": 9791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375352654526802e-05,
+      "loss": 0.5289,
+      "step": 9792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3752885355219284e-05,
+      "loss": 0.4686,
+      "step": 9793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3752244165170555e-05,
+      "loss": 0.5773,
+      "step": 9794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375160297512183e-05,
+      "loss": 0.5445,
+      "step": 9795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37509617850731e-05,
+      "loss": 0.7723,
+      "step": 9796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.375032059502437e-05,
+      "loss": 0.5716,
+      "step": 9797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3749679404975635e-05,
+      "loss": 0.6059,
+      "step": 9798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374903821492691e-05,
+      "loss": 0.6274,
+      "step": 9799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374839702487817e-05,
+      "loss": 0.6422,
+      "step": 9800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3747755834829444e-05,
+      "loss": 0.4537,
+      "step": 9801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3747114644780716e-05,
+      "loss": 0.5552,
+      "step": 9802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374647345473199e-05,
+      "loss": 0.6009,
+      "step": 9803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374583226468326e-05,
+      "loss": 0.4983,
+      "step": 9804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3745191074634524e-05,
+      "loss": 0.5531,
+      "step": 9805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3744549884585796e-05,
+      "loss": 0.5108,
+      "step": 9806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374390869453706e-05,
+      "loss": 0.5218,
+      "step": 9807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374326750448833e-05,
+      "loss": 0.5845,
+      "step": 9808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37426263144396e-05,
+      "loss": 0.5262,
+      "step": 9809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374198512439087e-05,
+      "loss": 0.4786,
+      "step": 9810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374134393434214e-05,
+      "loss": 0.4962,
+      "step": 9811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374070274429341e-05,
+      "loss": 0.4725,
+      "step": 9812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3740061554244684e-05,
+      "loss": 0.5341,
+      "step": 9813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373942036419595e-05,
+      "loss": 0.4781,
+      "step": 9814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373877917414722e-05,
+      "loss": 0.5631,
+      "step": 9815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3738137984098486e-05,
+      "loss": 0.6004,
+      "step": 9816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373749679404976e-05,
+      "loss": 0.5542,
+      "step": 9817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373685560400103e-05,
+      "loss": 0.5775,
+      "step": 9818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37362144139523e-05,
+      "loss": 0.6144,
+      "step": 9819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373557322390357e-05,
+      "loss": 0.4468,
+      "step": 9820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373493203385484e-05,
+      "loss": 0.4681,
+      "step": 9821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373429084380611e-05,
+      "loss": 0.5133,
+      "step": 9822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3733649653757374e-05,
+      "loss": 0.4899,
+      "step": 9823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3733008463708646e-05,
+      "loss": 0.6388,
+      "step": 9824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373236727365991e-05,
+      "loss": 0.4989,
+      "step": 9825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373172608361118e-05,
+      "loss": 0.5597,
+      "step": 9826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3731084893562454e-05,
+      "loss": 0.5365,
+      "step": 9827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3730443703513726e-05,
+      "loss": 0.6292,
+      "step": 9828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3729802513465e-05,
+      "loss": 0.5982,
+      "step": 9829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372916132341626e-05,
+      "loss": 0.5566,
+      "step": 9830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3728520133367534e-05,
+      "loss": 0.5331,
+      "step": 9831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37278789433188e-05,
+      "loss": 0.5579,
+      "step": 9832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372723775327007e-05,
+      "loss": 0.6135,
+      "step": 9833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3726596563221336e-05,
+      "loss": 0.6179,
+      "step": 9834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3725955373172614e-05,
+      "loss": 0.6876,
+      "step": 9835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372531418312388e-05,
+      "loss": 0.5172,
+      "step": 9836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372467299307515e-05,
+      "loss": 0.5265,
+      "step": 9837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372403180302642e-05,
+      "loss": 0.5494,
+      "step": 9838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372339061297769e-05,
+      "loss": 0.6344,
+      "step": 9839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372274942292896e-05,
+      "loss": 0.5731,
+      "step": 9840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3722108232880224e-05,
+      "loss": 0.5769,
+      "step": 9841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3721467042831496e-05,
+      "loss": 0.6547,
+      "step": 9842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372082585278277e-05,
+      "loss": 0.6115,
+      "step": 9843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372018466273404e-05,
+      "loss": 0.5051,
+      "step": 9844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3719543472685304e-05,
+      "loss": 0.4776,
+      "step": 9845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3718902282636576e-05,
+      "loss": 0.6089,
+      "step": 9846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371826109258785e-05,
+      "loss": 0.6087,
+      "step": 9847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371761990253911e-05,
+      "loss": 0.6164,
+      "step": 9848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3716978712490385e-05,
+      "loss": 0.5848,
+      "step": 9849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371633752244165e-05,
+      "loss": 0.5605,
+      "step": 9850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371569633239293e-05,
+      "loss": 0.5281,
+      "step": 9851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371505514234419e-05,
+      "loss": 0.5051,
+      "step": 9852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3714413952295465e-05,
+      "loss": 0.5378,
+      "step": 9853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371377276224673e-05,
+      "loss": 0.5242,
+      "step": 9854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3713131572198e-05,
+      "loss": 0.5376,
+      "step": 9855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371249038214927e-05,
+      "loss": 0.4943,
+      "step": 9856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371184919210054e-05,
+      "loss": 0.6165,
+      "step": 9857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371120800205181e-05,
+      "loss": 0.3977,
+      "step": 9858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371056681200308e-05,
+      "loss": 0.5666,
+      "step": 9859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370992562195435e-05,
+      "loss": 0.6058,
+      "step": 9860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370928443190562e-05,
+      "loss": 0.5345,
+      "step": 9861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370864324185689e-05,
+      "loss": 0.6421,
+      "step": 9862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3708002051808155e-05,
+      "loss": 0.7414,
+      "step": 9863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3707360861759426e-05,
+      "loss": 0.4815,
+      "step": 9864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37067196717107e-05,
+      "loss": 0.6294,
+      "step": 9865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370607848166196e-05,
+      "loss": 0.465,
+      "step": 9866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370543729161324e-05,
+      "loss": 0.5357,
+      "step": 9867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3704796101564506e-05,
+      "loss": 0.5397,
+      "step": 9868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370415491151578e-05,
+      "loss": 0.4629,
+      "step": 9869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370351372146704e-05,
+      "loss": 0.4805,
+      "step": 9870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3702872531418315e-05,
+      "loss": 0.5009,
+      "step": 9871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370223134136958e-05,
+      "loss": 0.5066,
+      "step": 9872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370159015132085e-05,
+      "loss": 0.58,
+      "step": 9873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370094896127212e-05,
+      "loss": 0.518,
+      "step": 9874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3700307771223395e-05,
+      "loss": 0.6339,
+      "step": 9875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369966658117467e-05,
+      "loss": 0.5091,
+      "step": 9876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369902539112593e-05,
+      "loss": 0.5032,
+      "step": 9877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36983842010772e-05,
+      "loss": 0.4899,
+      "step": 9878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369774301102847e-05,
+      "loss": 0.4643,
+      "step": 9879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369710182097974e-05,
+      "loss": 0.5495,
+      "step": 9880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3696460630931005e-05,
+      "loss": 0.6095,
+      "step": 9881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369581944088228e-05,
+      "loss": 0.4782,
+      "step": 9882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369517825083355e-05,
+      "loss": 0.5562,
+      "step": 9883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369453706078482e-05,
+      "loss": 0.5976,
+      "step": 9884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369389587073609e-05,
+      "loss": 0.661,
+      "step": 9885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369325468068736e-05,
+      "loss": 0.6638,
+      "step": 9886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369261349063863e-05,
+      "loss": 0.5315,
+      "step": 9887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3691972300589893e-05,
+      "loss": 0.4968,
+      "step": 9888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3691331110541165e-05,
+      "loss": 0.5719,
+      "step": 9889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369068992049244e-05,
+      "loss": 0.5498,
+      "step": 9890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369004873044371e-05,
+      "loss": 0.6032,
+      "step": 9891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368940754039498e-05,
+      "loss": 0.4819,
+      "step": 9892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3688766350346245e-05,
+      "loss": 0.5665,
+      "step": 9893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368812516029752e-05,
+      "loss": 0.6636,
+      "step": 9894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368748397024878e-05,
+      "loss": 0.5686,
+      "step": 9895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3686842780200054e-05,
+      "loss": 0.6346,
+      "step": 9896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368620159015132e-05,
+      "loss": 0.5396,
+      "step": 9897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368556040010259e-05,
+      "loss": 0.5584,
+      "step": 9898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368491921005386e-05,
+      "loss": 0.5553,
+      "step": 9899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3684278020005134e-05,
+      "loss": 0.517,
+      "step": 9900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3683636829956405e-05,
+      "loss": 0.6631,
+      "step": 9901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368299563990767e-05,
+      "loss": 0.5673,
+      "step": 9902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368235444985894e-05,
+      "loss": 0.62,
+      "step": 9903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368171325981021e-05,
+      "loss": 0.489,
+      "step": 9904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368107206976148e-05,
+      "loss": 0.5907,
+      "step": 9905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.368043087971275e-05,
+      "loss": 0.5309,
+      "step": 9906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367978968966402e-05,
+      "loss": 0.5045,
+      "step": 9907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367914849961529e-05,
+      "loss": 0.5585,
+      "step": 9908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367850730956656e-05,
+      "loss": 0.5391,
+      "step": 9909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367786611951783e-05,
+      "loss": 0.5642,
+      "step": 9910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3677224929469095e-05,
+      "loss": 0.5177,
+      "step": 9911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367658373942037e-05,
+      "loss": 0.5124,
+      "step": 9912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367594254937163e-05,
+      "loss": 0.5777,
+      "step": 9913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3675301359322904e-05,
+      "loss": 0.5605,
+      "step": 9914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3674660169274176e-05,
+      "loss": 0.5809,
+      "step": 9915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367401897922545e-05,
+      "loss": 0.6302,
+      "step": 9916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367337778917671e-05,
+      "loss": 0.5946,
+      "step": 9917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3672736599127984e-05,
+      "loss": 0.5752,
+      "step": 9918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3672095409079256e-05,
+      "loss": 0.5892,
+      "step": 9919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367145421903052e-05,
+      "loss": 0.7134,
+      "step": 9920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367081302898179e-05,
+      "loss": 0.5403,
+      "step": 9921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.367017183893306e-05,
+      "loss": 0.5168,
+      "step": 9922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3669530648884336e-05,
+      "loss": 0.551,
+      "step": 9923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36688894588356e-05,
+      "loss": 0.6352,
+      "step": 9924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366824826878687e-05,
+      "loss": 0.5932,
+      "step": 9925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366760707873814e-05,
+      "loss": 0.5253,
+      "step": 9926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366696588868941e-05,
+      "loss": 0.4984,
+      "step": 9927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366632469864068e-05,
+      "loss": 0.6404,
+      "step": 9928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3665683508591946e-05,
+      "loss": 0.5102,
+      "step": 9929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366504231854322e-05,
+      "loss": 0.5611,
+      "step": 9930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366440112849449e-05,
+      "loss": 0.5691,
+      "step": 9931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366375993844576e-05,
+      "loss": 0.6302,
+      "step": 9932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3663118748397026e-05,
+      "loss": 0.525,
+      "step": 9933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36624775583483e-05,
+      "loss": 0.4826,
+      "step": 9934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366183636829956e-05,
+      "loss": 0.5423,
+      "step": 9935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3661195178250834e-05,
+      "loss": 0.6523,
+      "step": 9936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3660553988202106e-05,
+      "loss": 0.5367,
+      "step": 9937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365991279815337e-05,
+      "loss": 0.4719,
+      "step": 9938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365927160810465e-05,
+      "loss": 0.4829,
+      "step": 9939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3658630418055914e-05,
+      "loss": 0.6272,
+      "step": 9940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3657989228007186e-05,
+      "loss": 0.4895,
+      "step": 9941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365734803795845e-05,
+      "loss": 0.5715,
+      "step": 9942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365670684790972e-05,
+      "loss": 0.5765,
+      "step": 9943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365606565786099e-05,
+      "loss": 0.5512,
+      "step": 9944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365542446781226e-05,
+      "loss": 0.4473,
+      "step": 9945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365478327776353e-05,
+      "loss": 0.5996,
+      "step": 9946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36541420877148e-05,
+      "loss": 0.5026,
+      "step": 9947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3653500897666074e-05,
+      "loss": 0.6041,
+      "step": 9948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365285970761734e-05,
+      "loss": 0.6055,
+      "step": 9949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365221851756861e-05,
+      "loss": 0.7124,
+      "step": 9950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3651577327519876e-05,
+      "loss": 0.5714,
+      "step": 9951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365093613747115e-05,
+      "loss": 0.6254,
+      "step": 9952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365029494742241e-05,
+      "loss": 0.6542,
+      "step": 9953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3649653757373684e-05,
+      "loss": 0.6067,
+      "step": 9954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3649012567324956e-05,
+      "loss": 0.615,
+      "step": 9955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.364837137727623e-05,
+      "loss": 0.574,
+      "step": 9956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36477301872275e-05,
+      "loss": 0.6413,
+      "step": 9957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3647088997178764e-05,
+      "loss": 0.6198,
+      "step": 9958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3646447807130036e-05,
+      "loss": 0.5712,
+      "step": 9959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36458066170813e-05,
+      "loss": 0.5934,
+      "step": 9960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.364516542703257e-05,
+      "loss": 0.5507,
+      "step": 9961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3644524236983845e-05,
+      "loss": 0.4915,
+      "step": 9962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3643883046935116e-05,
+      "loss": 0.5768,
+      "step": 9963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.364324185688639e-05,
+      "loss": 0.6045,
+      "step": 9964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.364260066683765e-05,
+      "loss": 0.5909,
+      "step": 9965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3641959476788925e-05,
+      "loss": 0.5339,
+      "step": 9966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.364131828674019e-05,
+      "loss": 0.5079,
+      "step": 9967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.364067709669146e-05,
+      "loss": 0.5928,
+      "step": 9968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3640035906642726e-05,
+      "loss": 0.6499,
+      "step": 9969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3639394716594e-05,
+      "loss": 0.5639,
+      "step": 9970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363875352654527e-05,
+      "loss": 0.5559,
+      "step": 9971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363811233649654e-05,
+      "loss": 0.5669,
+      "step": 9972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363747114644781e-05,
+      "loss": 0.5232,
+      "step": 9973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363682995639908e-05,
+      "loss": 0.5918,
+      "step": 9974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363618876635035e-05,
+      "loss": 0.4965,
+      "step": 9975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3635547576301615e-05,
+      "loss": 0.5984,
+      "step": 9976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3634906386252886e-05,
+      "loss": 0.5773,
+      "step": 9977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363426519620416e-05,
+      "loss": 0.594,
+      "step": 9978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363362400615543e-05,
+      "loss": 0.5794,
+      "step": 9979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3632982816106695e-05,
+      "loss": 0.6208,
+      "step": 9980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3632341626057967e-05,
+      "loss": 0.6266,
+      "step": 9981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363170043600924e-05,
+      "loss": 0.5104,
+      "step": 9982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36310592459605e-05,
+      "loss": 0.5467,
+      "step": 9983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3630418055911775e-05,
+      "loss": 0.6182,
+      "step": 9984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362977686586304e-05,
+      "loss": 0.5824,
+      "step": 9985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362913567581431e-05,
+      "loss": 0.5886,
+      "step": 9986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362849448576558e-05,
+      "loss": 0.5536,
+      "step": 9987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3627853295716855e-05,
+      "loss": 0.5639,
+      "step": 9988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362721210566813e-05,
+      "loss": 0.5733,
+      "step": 9989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362657091561939e-05,
+      "loss": 0.5807,
+      "step": 9990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362592972557066e-05,
+      "loss": 0.5875,
+      "step": 9991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362528853552193e-05,
+      "loss": 0.491,
+      "step": 9992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36246473454732e-05,
+      "loss": 0.4846,
+      "step": 9993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3624006155424465e-05,
+      "loss": 0.5881,
+      "step": 9994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3623364965375743e-05,
+      "loss": 0.598,
+      "step": 9995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362272377532701e-05,
+      "loss": 0.5306,
+      "step": 9996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362208258527828e-05,
+      "loss": 0.5221,
+      "step": 9997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362144139522955e-05,
+      "loss": 0.5392,
+      "step": 9998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362080020518082e-05,
+      "loss": 0.5599,
+      "step": 9999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362015901513209e-05,
+      "loss": 0.595,
+      "step": 10000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3619517825083353e-05,
+      "loss": 0.5503,
+      "step": 10001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3618876635034625e-05,
+      "loss": 0.553,
+      "step": 10002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36182354449859e-05,
+      "loss": 0.4927,
+      "step": 10003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.361759425493717e-05,
+      "loss": 0.5907,
+      "step": 10004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3616953064888434e-05,
+      "loss": 0.5177,
+      "step": 10005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3616311874839705e-05,
+      "loss": 0.4543,
+      "step": 10006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.361567068479098e-05,
+      "loss": 0.4947,
+      "step": 10007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.361502949474224e-05,
+      "loss": 0.5541,
+      "step": 10008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3614388304693514e-05,
+      "loss": 0.575,
+      "step": 10009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.361374711464478e-05,
+      "loss": 0.5934,
+      "step": 10010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.361310592459606e-05,
+      "loss": 0.5468,
+      "step": 10011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.361246473454732e-05,
+      "loss": 0.5982,
+      "step": 10012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3611823544498594e-05,
+      "loss": 0.4948,
+      "step": 10013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.361118235444986e-05,
+      "loss": 0.5034,
+      "step": 10014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.361054116440113e-05,
+      "loss": 0.4579,
+      "step": 10015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.36098999743524e-05,
+      "loss": 0.5136,
+      "step": 10016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360925878430367e-05,
+      "loss": 0.5961,
+      "step": 10017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360861759425494e-05,
+      "loss": 0.5473,
+      "step": 10018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360797640420621e-05,
+      "loss": 0.6665,
+      "step": 10019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360733521415748e-05,
+      "loss": 0.612,
+      "step": 10020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360669402410875e-05,
+      "loss": 0.5972,
+      "step": 10021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360605283406002e-05,
+      "loss": 0.5437,
+      "step": 10022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3605411644011284e-05,
+      "loss": 0.6813,
+      "step": 10023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3604770453962555e-05,
+      "loss": 0.6311,
+      "step": 10024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360412926391383e-05,
+      "loss": 0.6191,
+      "step": 10025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360348807386509e-05,
+      "loss": 0.4873,
+      "step": 10026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360284688381637e-05,
+      "loss": 0.5583,
+      "step": 10027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3602205693767636e-05,
+      "loss": 0.6197,
+      "step": 10028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360156450371891e-05,
+      "loss": 0.551,
+      "step": 10029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360092331367017e-05,
+      "loss": 0.6926,
+      "step": 10030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3600282123621444e-05,
+      "loss": 0.6023,
+      "step": 10031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359964093357271e-05,
+      "loss": 0.5925,
+      "step": 10032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359899974352398e-05,
+      "loss": 0.6257,
+      "step": 10033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359835855347525e-05,
+      "loss": 0.5617,
+      "step": 10034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3597717363426524e-05,
+      "loss": 0.5823,
+      "step": 10035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3597076173377796e-05,
+      "loss": 0.5883,
+      "step": 10036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359643498332906e-05,
+      "loss": 0.6145,
+      "step": 10037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359579379328033e-05,
+      "loss": 0.5804,
+      "step": 10038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35951526032316e-05,
+      "loss": 0.6096,
+      "step": 10039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359451141318287e-05,
+      "loss": 0.5522,
+      "step": 10040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3593870223134134e-05,
+      "loss": 0.5628,
+      "step": 10041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3593229033085406e-05,
+      "loss": 0.5559,
+      "step": 10042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359258784303668e-05,
+      "loss": 0.5456,
+      "step": 10043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359194665298795e-05,
+      "loss": 0.6936,
+      "step": 10044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359130546293922e-05,
+      "loss": 0.5694,
+      "step": 10045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3590664272890486e-05,
+      "loss": 0.5406,
+      "step": 10046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359002308284176e-05,
+      "loss": 0.5335,
+      "step": 10047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358938189279302e-05,
+      "loss": 0.6092,
+      "step": 10048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3588740702744294e-05,
+      "loss": 0.5872,
+      "step": 10049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3588099512695566e-05,
+      "loss": 0.5882,
+      "step": 10050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358745832264684e-05,
+      "loss": 0.6187,
+      "step": 10051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358681713259811e-05,
+      "loss": 0.4796,
+      "step": 10052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3586175942549374e-05,
+      "loss": 0.5007,
+      "step": 10053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3585534752500646e-05,
+      "loss": 0.6542,
+      "step": 10054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358489356245191e-05,
+      "loss": 0.5249,
+      "step": 10055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358425237240318e-05,
+      "loss": 0.5676,
+      "step": 10056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358361118235445e-05,
+      "loss": 0.4875,
+      "step": 10057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358296999230572e-05,
+      "loss": 0.5577,
+      "step": 10058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358232880225699e-05,
+      "loss": 0.5969,
+      "step": 10059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358168761220826e-05,
+      "loss": 0.5327,
+      "step": 10060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3581046422159534e-05,
+      "loss": 0.5232,
+      "step": 10061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35804052321108e-05,
+      "loss": 0.4772,
+      "step": 10062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357976404206207e-05,
+      "loss": 0.629,
+      "step": 10063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3579122852013336e-05,
+      "loss": 0.6112,
+      "step": 10064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357848166196461e-05,
+      "loss": 0.5262,
+      "step": 10065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357784047191588e-05,
+      "loss": 0.6984,
+      "step": 10066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357719928186715e-05,
+      "loss": 0.6167,
+      "step": 10067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3576558091818416e-05,
+      "loss": 0.4805,
+      "step": 10068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357591690176969e-05,
+      "loss": 0.6047,
+      "step": 10069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357527571172096e-05,
+      "loss": 0.5667,
+      "step": 10070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3574634521672225e-05,
+      "loss": 0.4967,
+      "step": 10071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3573993331623496e-05,
+      "loss": 0.5115,
+      "step": 10072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357335214157476e-05,
+      "loss": 0.5243,
+      "step": 10073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357271095152603e-05,
+      "loss": 0.5791,
+      "step": 10074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3572069761477305e-05,
+      "loss": 0.5658,
+      "step": 10075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3571428571428576e-05,
+      "loss": 0.5207,
+      "step": 10076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357078738137984e-05,
+      "loss": 0.714,
+      "step": 10077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357014619133111e-05,
+      "loss": 0.6105,
+      "step": 10078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3569505001282385e-05,
+      "loss": 0.5822,
+      "step": 10079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356886381123365e-05,
+      "loss": 0.5518,
+      "step": 10080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356822262118492e-05,
+      "loss": 0.5486,
+      "step": 10081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3567581431136186e-05,
+      "loss": 0.5987,
+      "step": 10082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3566940241087465e-05,
+      "loss": 0.5249,
+      "step": 10083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356629905103873e-05,
+      "loss": 0.5996,
+      "step": 10084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356565786099e-05,
+      "loss": 0.6106,
+      "step": 10085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3565016670941266e-05,
+      "loss": 0.5323,
+      "step": 10086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356437548089254e-05,
+      "loss": 0.4846,
+      "step": 10087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356373429084381e-05,
+      "loss": 0.6395,
+      "step": 10088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3563093100795075e-05,
+      "loss": 0.547,
+      "step": 10089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3562451910746346e-05,
+      "loss": 0.5616,
+      "step": 10090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356181072069762e-05,
+      "loss": 0.4647,
+      "step": 10091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356116953064889e-05,
+      "loss": 0.56,
+      "step": 10092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3560528340600155e-05,
+      "loss": 0.6717,
+      "step": 10093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3559887150551427e-05,
+      "loss": 0.5408,
+      "step": 10094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355924596050269e-05,
+      "loss": 0.6168,
+      "step": 10095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355860477045396e-05,
+      "loss": 0.5657,
+      "step": 10096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3557963580405235e-05,
+      "loss": 0.7164,
+      "step": 10097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35573223903565e-05,
+      "loss": 0.5314,
+      "step": 10098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355668120030778e-05,
+      "loss": 0.474,
+      "step": 10099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355604001025904e-05,
+      "loss": 0.4791,
+      "step": 10100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3555398820210315e-05,
+      "loss": 0.5636,
+      "step": 10101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355475763016158e-05,
+      "loss": 0.5727,
+      "step": 10102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355411644011285e-05,
+      "loss": 0.642,
+      "step": 10103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355347525006412e-05,
+      "loss": 0.545,
+      "step": 10104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355283406001539e-05,
+      "loss": 0.488,
+      "step": 10105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355219286996666e-05,
+      "loss": 0.5627,
+      "step": 10106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355155167991793e-05,
+      "loss": 0.5668,
+      "step": 10107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3550910489869203e-05,
+      "loss": 0.5278,
+      "step": 10108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355026929982047e-05,
+      "loss": 0.6097,
+      "step": 10109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354962810977174e-05,
+      "loss": 0.6293,
+      "step": 10110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3548986919723005e-05,
+      "loss": 0.6425,
+      "step": 10111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354834572967428e-05,
+      "loss": 0.4955,
+      "step": 10112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354770453962554e-05,
+      "loss": 0.619,
+      "step": 10113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3547063349576813e-05,
+      "loss": 0.5793,
+      "step": 10114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3546422159528085e-05,
+      "loss": 0.5887,
+      "step": 10115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354578096947936e-05,
+      "loss": 0.6903,
+      "step": 10116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354513977943063e-05,
+      "loss": 0.5472,
+      "step": 10117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3544498589381894e-05,
+      "loss": 0.5473,
+      "step": 10118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3543857399333165e-05,
+      "loss": 0.4781,
+      "step": 10119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354321620928443e-05,
+      "loss": 0.6349,
+      "step": 10120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35425750192357e-05,
+      "loss": 0.5057,
+      "step": 10121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3541933829186974e-05,
+      "loss": 0.5237,
+      "step": 10122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3541292639138245e-05,
+      "loss": 0.5825,
+      "step": 10123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354065144908952e-05,
+      "loss": 0.6909,
+      "step": 10124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354001025904078e-05,
+      "loss": 0.4715,
+      "step": 10125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3539369068992054e-05,
+      "loss": 0.5941,
+      "step": 10126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353872787894332e-05,
+      "loss": 0.6281,
+      "step": 10127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353808668889459e-05,
+      "loss": 0.5967,
+      "step": 10128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3537445498845855e-05,
+      "loss": 0.4914,
+      "step": 10129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353680430879713e-05,
+      "loss": 0.5548,
+      "step": 10130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35361631187484e-05,
+      "loss": 0.5453,
+      "step": 10131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353552192869967e-05,
+      "loss": 0.6145,
+      "step": 10132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353488073865094e-05,
+      "loss": 0.6212,
+      "step": 10133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353423954860221e-05,
+      "loss": 0.5083,
+      "step": 10134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353359835855348e-05,
+      "loss": 0.5074,
+      "step": 10135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3532957168504744e-05,
+      "loss": 0.5634,
+      "step": 10136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3532315978456016e-05,
+      "loss": 0.5156,
+      "step": 10137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353167478840729e-05,
+      "loss": 0.5557,
+      "step": 10138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353103359835856e-05,
+      "loss": 0.5007,
+      "step": 10139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3530392408309824e-05,
+      "loss": 0.4673,
+      "step": 10140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3529751218261096e-05,
+      "loss": 0.4671,
+      "step": 10141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352911002821237e-05,
+      "loss": 0.5526,
+      "step": 10142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352846883816363e-05,
+      "loss": 0.5158,
+      "step": 10143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3527827648114904e-05,
+      "loss": 0.5074,
+      "step": 10144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352718645806617e-05,
+      "loss": 0.5107,
+      "step": 10145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352654526801744e-05,
+      "loss": 0.5604,
+      "step": 10146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352590407796871e-05,
+      "loss": 0.5327,
+      "step": 10147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3525262887919984e-05,
+      "loss": 0.4741,
+      "step": 10148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352462169787125e-05,
+      "loss": 0.6109,
+      "step": 10149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352398050782252e-05,
+      "loss": 0.5522,
+      "step": 10150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352333931777379e-05,
+      "loss": 0.5601,
+      "step": 10151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352269812772506e-05,
+      "loss": 0.6388,
+      "step": 10152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352205693767633e-05,
+      "loss": 0.571,
+      "step": 10153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35214157476276e-05,
+      "loss": 0.5753,
+      "step": 10154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352077455757887e-05,
+      "loss": 0.5289,
+      "step": 10155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352013336753014e-05,
+      "loss": 0.5507,
+      "step": 10156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351949217748141e-05,
+      "loss": 0.475,
+      "step": 10157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3518850987432674e-05,
+      "loss": 0.5782,
+      "step": 10158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3518209797383946e-05,
+      "loss": 0.5657,
+      "step": 10159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351756860733522e-05,
+      "loss": 0.5212,
+      "step": 10160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351692741728648e-05,
+      "loss": 0.5278,
+      "step": 10161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3516286227237754e-05,
+      "loss": 0.5832,
+      "step": 10162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3515645037189026e-05,
+      "loss": 0.5286,
+      "step": 10163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35150038471403e-05,
+      "loss": 0.5349,
+      "step": 10164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351436265709156e-05,
+      "loss": 0.5533,
+      "step": 10165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3513721467042834e-05,
+      "loss": 0.6156,
+      "step": 10166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35130802769941e-05,
+      "loss": 0.5825,
+      "step": 10167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351243908694537e-05,
+      "loss": 0.5927,
+      "step": 10168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351179789689664e-05,
+      "loss": 0.5857,
+      "step": 10169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.351115670684791e-05,
+      "loss": 0.5197,
+      "step": 10170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3510515516799186e-05,
+      "loss": 0.5889,
+      "step": 10171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350987432675045e-05,
+      "loss": 0.6776,
+      "step": 10172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350923313670172e-05,
+      "loss": 0.5768,
+      "step": 10173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350859194665299e-05,
+      "loss": 0.5819,
+      "step": 10174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350795075660426e-05,
+      "loss": 0.6189,
+      "step": 10175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3507309566555524e-05,
+      "loss": 0.5417,
+      "step": 10176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3506668376506796e-05,
+      "loss": 0.5617,
+      "step": 10177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350602718645807e-05,
+      "loss": 0.5596,
+      "step": 10178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350538599640934e-05,
+      "loss": 0.5121,
+      "step": 10179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350474480636061e-05,
+      "loss": 0.6033,
+      "step": 10180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3504103616311876e-05,
+      "loss": 0.6129,
+      "step": 10181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350346242626315e-05,
+      "loss": 0.5303,
+      "step": 10182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350282123621441e-05,
+      "loss": 0.6165,
+      "step": 10183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3502180046165685e-05,
+      "loss": 0.7298,
+      "step": 10184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3501538856116956e-05,
+      "loss": 0.5379,
+      "step": 10185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.350089766606822e-05,
+      "loss": 0.5507,
+      "step": 10186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35002564760195e-05,
+      "loss": 0.5293,
+      "step": 10187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3499615285970765e-05,
+      "loss": 0.5822,
+      "step": 10188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3498974095922036e-05,
+      "loss": 0.6623,
+      "step": 10189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.34983329058733e-05,
+      "loss": 0.5689,
+      "step": 10190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349769171582457e-05,
+      "loss": 0.6399,
+      "step": 10191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349705052577584e-05,
+      "loss": 0.6041,
+      "step": 10192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349640933572711e-05,
+      "loss": 0.5366,
+      "step": 10193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349576814567838e-05,
+      "loss": 0.5405,
+      "step": 10194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349512695562965e-05,
+      "loss": 0.6163,
+      "step": 10195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3494485765580925e-05,
+      "loss": 0.6626,
+      "step": 10196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349384457553219e-05,
+      "loss": 0.5265,
+      "step": 10197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349320338548346e-05,
+      "loss": 0.514,
+      "step": 10198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3492562195434726e-05,
+      "loss": 0.5085,
+      "step": 10199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3491921005386e-05,
+      "loss": 0.5413,
+      "step": 10200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349127981533726e-05,
+      "loss": 0.5422,
+      "step": 10201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3490638625288535e-05,
+      "loss": 0.5224,
+      "step": 10202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3489997435239807e-05,
+      "loss": 0.6299,
+      "step": 10203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348935624519108e-05,
+      "loss": 0.5933,
+      "step": 10204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348871505514235e-05,
+      "loss": 0.4935,
+      "step": 10205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3488073865093615e-05,
+      "loss": 0.5321,
+      "step": 10206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3487432675044887e-05,
+      "loss": 0.5247,
+      "step": 10207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348679148499615e-05,
+      "loss": 0.704,
+      "step": 10208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348615029494742e-05,
+      "loss": 0.5528,
+      "step": 10209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3485509104898695e-05,
+      "loss": 0.5048,
+      "step": 10210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348486791484997e-05,
+      "loss": 0.6288,
+      "step": 10211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348422672480124e-05,
+      "loss": 0.5054,
+      "step": 10212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.34835855347525e-05,
+      "loss": 0.5775,
+      "step": 10213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3482944344703775e-05,
+      "loss": 0.5177,
+      "step": 10214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348230315465504e-05,
+      "loss": 0.5254,
+      "step": 10215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348166196460631e-05,
+      "loss": 0.4985,
+      "step": 10216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348102077455758e-05,
+      "loss": 0.5218,
+      "step": 10217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348037958450885e-05,
+      "loss": 0.5792,
+      "step": 10218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347973839446012e-05,
+      "loss": 0.5437,
+      "step": 10219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347909720441139e-05,
+      "loss": 0.4755,
+      "step": 10220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3478456014362664e-05,
+      "loss": 0.6426,
+      "step": 10221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347781482431393e-05,
+      "loss": 0.5283,
+      "step": 10222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.34771736342652e-05,
+      "loss": 0.6087,
+      "step": 10223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3476532444216465e-05,
+      "loss": 0.5288,
+      "step": 10224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347589125416774e-05,
+      "loss": 0.5563,
+      "step": 10225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347525006411901e-05,
+      "loss": 0.4884,
+      "step": 10226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347460887407028e-05,
+      "loss": 0.6693,
+      "step": 10227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3473967684021545e-05,
+      "loss": 0.5464,
+      "step": 10228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347332649397282e-05,
+      "loss": 0.5245,
+      "step": 10229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347268530392409e-05,
+      "loss": 0.5692,
+      "step": 10230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3472044113875354e-05,
+      "loss": 0.5342,
+      "step": 10231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3471402923826625e-05,
+      "loss": 0.5913,
+      "step": 10232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347076173377789e-05,
+      "loss": 0.4886,
+      "step": 10233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347012054372916e-05,
+      "loss": 0.5406,
+      "step": 10234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3469479353680434e-05,
+      "loss": 0.5325,
+      "step": 10235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3468838163631705e-05,
+      "loss": 0.6788,
+      "step": 10236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346819697358297e-05,
+      "loss": 0.5487,
+      "step": 10237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346755578353424e-05,
+      "loss": 0.5292,
+      "step": 10238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3466914593485514e-05,
+      "loss": 0.661,
+      "step": 10239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346627340343678e-05,
+      "loss": 0.6276,
+      "step": 10240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346563221338805e-05,
+      "loss": 0.6402,
+      "step": 10241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3464991023339315e-05,
+      "loss": 0.6873,
+      "step": 10242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3464349833290594e-05,
+      "loss": 0.5639,
+      "step": 10243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346370864324186e-05,
+      "loss": 0.5433,
+      "step": 10244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346306745319313e-05,
+      "loss": 0.5044,
+      "step": 10245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3462426263144395e-05,
+      "loss": 0.5151,
+      "step": 10246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346178507309567e-05,
+      "loss": 0.586,
+      "step": 10247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.346114388304694e-05,
+      "loss": 0.6336,
+      "step": 10248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3460502692998204e-05,
+      "loss": 0.5436,
+      "step": 10249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3459861502949476e-05,
+      "loss": 0.5236,
+      "step": 10250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345922031290075e-05,
+      "loss": 0.5922,
+      "step": 10251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345857912285202e-05,
+      "loss": 0.5752,
+      "step": 10252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3457937932803284e-05,
+      "loss": 0.6306,
+      "step": 10253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3457296742754556e-05,
+      "loss": 0.4414,
+      "step": 10254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345665555270582e-05,
+      "loss": 0.5246,
+      "step": 10255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345601436265709e-05,
+      "loss": 0.5915,
+      "step": 10256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3455373172608364e-05,
+      "loss": 0.5436,
+      "step": 10257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345473198255963e-05,
+      "loss": 0.6562,
+      "step": 10258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345409079251091e-05,
+      "loss": 0.6972,
+      "step": 10259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345344960246217e-05,
+      "loss": 0.558,
+      "step": 10260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3452808412413444e-05,
+      "loss": 0.4778,
+      "step": 10261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345216722236471e-05,
+      "loss": 0.6044,
+      "step": 10262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345152603231598e-05,
+      "loss": 0.5998,
+      "step": 10263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3450884842267246e-05,
+      "loss": 0.5027,
+      "step": 10264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345024365221852e-05,
+      "loss": 0.6212,
+      "step": 10265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.344960246216979e-05,
+      "loss": 0.5845,
+      "step": 10266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.344896127212106e-05,
+      "loss": 0.5846,
+      "step": 10267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.344832008207233e-05,
+      "loss": 0.5162,
+      "step": 10268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.34476788920236e-05,
+      "loss": 0.5182,
+      "step": 10269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.344703770197487e-05,
+      "loss": 0.5729,
+      "step": 10270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3446396511926134e-05,
+      "loss": 0.6318,
+      "step": 10271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3445755321877406e-05,
+      "loss": 0.637,
+      "step": 10272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.344511413182867e-05,
+      "loss": 0.5848,
+      "step": 10273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344447294177994e-05,
+      "loss": 0.5955,
+      "step": 10274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344383175173122e-05,
+      "loss": 0.5268,
+      "step": 10275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3443190561682486e-05,
+      "loss": 0.5509,
+      "step": 10276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344254937163376e-05,
+      "loss": 0.5859,
+      "step": 10277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344190818158502e-05,
+      "loss": 0.6051,
+      "step": 10278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3441266991536294e-05,
+      "loss": 0.5962,
+      "step": 10279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344062580148756e-05,
+      "loss": 0.5147,
+      "step": 10280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343998461143883e-05,
+      "loss": 0.5518,
+      "step": 10281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34393434213901e-05,
+      "loss": 0.6561,
+      "step": 10282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3438702231341374e-05,
+      "loss": 0.573,
+      "step": 10283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3438061041292646e-05,
+      "loss": 0.5577,
+      "step": 10284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343741985124391e-05,
+      "loss": 0.5199,
+      "step": 10285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343677866119518e-05,
+      "loss": 0.4917,
+      "step": 10286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343613747114645e-05,
+      "loss": 0.6768,
+      "step": 10287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343549628109772e-05,
+      "loss": 0.5096,
+      "step": 10288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3434855091048984e-05,
+      "loss": 0.496,
+      "step": 10289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3434213901000256e-05,
+      "loss": 0.5168,
+      "step": 10290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343357271095153e-05,
+      "loss": 0.6787,
+      "step": 10291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34329315209028e-05,
+      "loss": 0.6202,
+      "step": 10292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343229033085407e-05,
+      "loss": 0.5736,
+      "step": 10293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3431649140805336e-05,
+      "loss": 0.5821,
+      "step": 10294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343100795075661e-05,
+      "loss": 0.4534,
+      "step": 10295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343036676070787e-05,
+      "loss": 0.6192,
+      "step": 10296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3429725570659145e-05,
+      "loss": 0.5561,
+      "step": 10297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3429084380610416e-05,
+      "loss": 0.627,
+      "step": 10298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342844319056169e-05,
+      "loss": 0.6432,
+      "step": 10299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342780200051295e-05,
+      "loss": 0.4945,
+      "step": 10300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3427160810464225e-05,
+      "loss": 0.5987,
+      "step": 10301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3426519620415496e-05,
+      "loss": 0.6375,
+      "step": 10302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342587843036676e-05,
+      "loss": 0.6277,
+      "step": 10303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342523724031803e-05,
+      "loss": 0.5564,
+      "step": 10304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34245960502693e-05,
+      "loss": 0.5371,
+      "step": 10305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342395486022057e-05,
+      "loss": 0.6502,
+      "step": 10306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342331367017184e-05,
+      "loss": 0.5181,
+      "step": 10307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342267248012311e-05,
+      "loss": 0.6306,
+      "step": 10308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342203129007438e-05,
+      "loss": 0.6441,
+      "step": 10309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342139010002565e-05,
+      "loss": 0.5465,
+      "step": 10310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342074890997692e-05,
+      "loss": 0.5618,
+      "step": 10311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3420107719928186e-05,
+      "loss": 0.5921,
+      "step": 10312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341946652987946e-05,
+      "loss": 0.5548,
+      "step": 10313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341882533983073e-05,
+      "loss": 0.5602,
+      "step": 10314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3418184149782e-05,
+      "loss": 0.5857,
+      "step": 10315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3417542959733267e-05,
+      "loss": 0.538,
+      "step": 10316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341690176968454e-05,
+      "loss": 0.5908,
+      "step": 10317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34162605796358e-05,
+      "loss": 0.584,
+      "step": 10318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3415619389587075e-05,
+      "loss": 0.6275,
+      "step": 10319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341497819953835e-05,
+      "loss": 0.5527,
+      "step": 10320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341433700948961e-05,
+      "loss": 0.5576,
+      "step": 10321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341369581944088e-05,
+      "loss": 0.5492,
+      "step": 10322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3413054629392155e-05,
+      "loss": 0.5506,
+      "step": 10323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341241343934343e-05,
+      "loss": 0.6107,
+      "step": 10324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341177224929469e-05,
+      "loss": 0.6101,
+      "step": 10325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341113105924596e-05,
+      "loss": 0.6107,
+      "step": 10326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341048986919723e-05,
+      "loss": 0.5902,
+      "step": 10327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34098486791485e-05,
+      "loss": 0.7695,
+      "step": 10328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340920748909977e-05,
+      "loss": 0.5286,
+      "step": 10329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340856629905104e-05,
+      "loss": 0.514,
+      "step": 10330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3407925109002315e-05,
+      "loss": 0.6081,
+      "step": 10331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340728391895358e-05,
+      "loss": 0.5639,
+      "step": 10332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340664272890485e-05,
+      "loss": 0.5353,
+      "step": 10333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340600153885612e-05,
+      "loss": 0.5593,
+      "step": 10334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340536034880739e-05,
+      "loss": 0.5689,
+      "step": 10335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3404719158758653e-05,
+      "loss": 0.6367,
+      "step": 10336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3404077968709925e-05,
+      "loss": 0.5552,
+      "step": 10337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34034367786612e-05,
+      "loss": 0.5547,
+      "step": 10338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340279558861247e-05,
+      "loss": 0.5864,
+      "step": 10339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340215439856374e-05,
+      "loss": 0.6654,
+      "step": 10340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3401513208515005e-05,
+      "loss": 0.598,
+      "step": 10341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340087201846628e-05,
+      "loss": 0.5392,
+      "step": 10342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340023082841754e-05,
+      "loss": 0.5358,
+      "step": 10343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3399589638368814e-05,
+      "loss": 0.5311,
+      "step": 10344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339894844832008e-05,
+      "loss": 0.6803,
+      "step": 10345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339830725827135e-05,
+      "loss": 0.6673,
+      "step": 10346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339766606822263e-05,
+      "loss": 0.5792,
+      "step": 10347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3397024878173894e-05,
+      "loss": 0.6677,
+      "step": 10348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3396383688125165e-05,
+      "loss": 0.5015,
+      "step": 10349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339574249807643e-05,
+      "loss": 0.5425,
+      "step": 10350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33951013080277e-05,
+      "loss": 0.6295,
+      "step": 10351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339446011797897e-05,
+      "loss": 0.6489,
+      "step": 10352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339381892793024e-05,
+      "loss": 0.6226,
+      "step": 10353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339317773788151e-05,
+      "loss": 0.604,
+      "step": 10354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339253654783278e-05,
+      "loss": 0.4802,
+      "step": 10355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3391895357784054e-05,
+      "loss": 0.656,
+      "step": 10356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339125416773532e-05,
+      "loss": 0.5636,
+      "step": 10357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339061297768659e-05,
+      "loss": 0.5482,
+      "step": 10358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3389971787637855e-05,
+      "loss": 0.5563,
+      "step": 10359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338933059758913e-05,
+      "loss": 0.5445,
+      "step": 10360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338868940754039e-05,
+      "loss": 0.579,
+      "step": 10361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3388048217491664e-05,
+      "loss": 0.5559,
+      "step": 10362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3387407027442936e-05,
+      "loss": 0.683,
+      "step": 10363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338676583739421e-05,
+      "loss": 0.499,
+      "step": 10364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338612464734548e-05,
+      "loss": 0.567,
+      "step": 10365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3385483457296744e-05,
+      "loss": 0.5242,
+      "step": 10366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3384842267248016e-05,
+      "loss": 0.576,
+      "step": 10367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338420107719928e-05,
+      "loss": 0.5902,
+      "step": 10368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338355988715055e-05,
+      "loss": 0.6493,
+      "step": 10369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3382918697101824e-05,
+      "loss": 0.5179,
+      "step": 10370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3382277507053096e-05,
+      "loss": 0.6596,
+      "step": 10371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338163631700437e-05,
+      "loss": 0.5208,
+      "step": 10372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338099512695563e-05,
+      "loss": 0.5538,
+      "step": 10373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3380353936906904e-05,
+      "loss": 0.6228,
+      "step": 10374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337971274685817e-05,
+      "loss": 0.5153,
+      "step": 10375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337907155680944e-05,
+      "loss": 0.6286,
+      "step": 10376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3378430366760706e-05,
+      "loss": 0.6348,
+      "step": 10377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337778917671198e-05,
+      "loss": 0.5032,
+      "step": 10378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337714798666325e-05,
+      "loss": 0.5548,
+      "step": 10379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337650679661452e-05,
+      "loss": 0.4541,
+      "step": 10380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337586560656579e-05,
+      "loss": 0.5632,
+      "step": 10381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337522441651706e-05,
+      "loss": 0.5326,
+      "step": 10382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337458322646833e-05,
+      "loss": 0.5637,
+      "step": 10383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3373942036419594e-05,
+      "loss": 0.5621,
+      "step": 10384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3373300846370866e-05,
+      "loss": 0.4965,
+      "step": 10385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337265965632214e-05,
+      "loss": 0.5517,
+      "step": 10386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337201846627341e-05,
+      "loss": 0.5298,
+      "step": 10387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3371377276224674e-05,
+      "loss": 0.4818,
+      "step": 10388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3370736086175946e-05,
+      "loss": 0.5238,
+      "step": 10389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337009489612722e-05,
+      "loss": 0.5174,
+      "step": 10390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336945370607848e-05,
+      "loss": 0.558,
+      "step": 10391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3368812516029754e-05,
+      "loss": 0.722,
+      "step": 10392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336817132598102e-05,
+      "loss": 0.5301,
+      "step": 10393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336753013593229e-05,
+      "loss": 0.5421,
+      "step": 10394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336688894588356e-05,
+      "loss": 0.4854,
+      "step": 10395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3366247755834834e-05,
+      "loss": 0.6562,
+      "step": 10396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33656065657861e-05,
+      "loss": 0.5835,
+      "step": 10397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336496537573737e-05,
+      "loss": 0.6251,
+      "step": 10398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336432418568864e-05,
+      "loss": 0.4841,
+      "step": 10399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336368299563991e-05,
+      "loss": 0.5249,
+      "step": 10400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336304180559118e-05,
+      "loss": 0.5812,
+      "step": 10401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3362400615542444e-05,
+      "loss": 0.5432,
+      "step": 10402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336175942549372e-05,
+      "loss": 0.4581,
+      "step": 10403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336111823544499e-05,
+      "loss": 0.5919,
+      "step": 10404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336047704539626e-05,
+      "loss": 0.6182,
+      "step": 10405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3359835855347525e-05,
+      "loss": 0.5192,
+      "step": 10406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3359194665298796e-05,
+      "loss": 0.5798,
+      "step": 10407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335855347525007e-05,
+      "loss": 0.5309,
+      "step": 10408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335791228520133e-05,
+      "loss": 0.6301,
+      "step": 10409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3357271095152605e-05,
+      "loss": 0.5896,
+      "step": 10410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3356629905103876e-05,
+      "loss": 0.6159,
+      "step": 10411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335598871505515e-05,
+      "loss": 0.5349,
+      "step": 10412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335534752500641e-05,
+      "loss": 0.4984,
+      "step": 10413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3354706334957685e-05,
+      "loss": 0.4742,
+      "step": 10414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335406514490895e-05,
+      "loss": 0.5663,
+      "step": 10415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335342395486022e-05,
+      "loss": 0.5923,
+      "step": 10416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335278276481149e-05,
+      "loss": 0.6354,
+      "step": 10417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335214157476276e-05,
+      "loss": 0.5914,
+      "step": 10418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3351500384714037e-05,
+      "loss": 0.574,
+      "step": 10419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33508591946653e-05,
+      "loss": 0.5309,
+      "step": 10420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335021800461657e-05,
+      "loss": 0.5565,
+      "step": 10421
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334957681456784e-05,
+      "loss": 0.6937,
+      "step": 10422
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334893562451911e-05,
+      "loss": 0.6174,
+      "step": 10423
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3348294434470375e-05,
+      "loss": 0.5713,
+      "step": 10424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3347653244421646e-05,
+      "loss": 0.5886,
+      "step": 10425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334701205437292e-05,
+      "loss": 0.6153,
+      "step": 10426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334637086432419e-05,
+      "loss": 0.5072,
+      "step": 10427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334572967427546e-05,
+      "loss": 0.4938,
+      "step": 10428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3345088484226727e-05,
+      "loss": 0.5888,
+      "step": 10429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3344447294178e-05,
+      "loss": 0.4895,
+      "step": 10430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334380610412926e-05,
+      "loss": 0.5673,
+      "step": 10431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3343164914080535e-05,
+      "loss": 0.516,
+      "step": 10432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33425237240318e-05,
+      "loss": 0.4615,
+      "step": 10433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334188253398307e-05,
+      "loss": 0.5841,
+      "step": 10434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334124134393435e-05,
+      "loss": 0.5523,
+      "step": 10435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3340600153885615e-05,
+      "loss": 0.5755,
+      "step": 10436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333995896383689e-05,
+      "loss": 0.5269,
+      "step": 10437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333931777378815e-05,
+      "loss": 0.5521,
+      "step": 10438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3338676583739423e-05,
+      "loss": 0.5114,
+      "step": 10439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333803539369069e-05,
+      "loss": 0.459,
+      "step": 10440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333739420364196e-05,
+      "loss": 0.626,
+      "step": 10441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333675301359323e-05,
+      "loss": 0.6119,
+      "step": 10442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3336111823544504e-05,
+      "loss": 0.57,
+      "step": 10443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3335470633495775e-05,
+      "loss": 0.6022,
+      "step": 10444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333482944344704e-05,
+      "loss": 0.4927,
+      "step": 10445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333418825339831e-05,
+      "loss": 0.634,
+      "step": 10446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333354706334958e-05,
+      "loss": 0.5756,
+      "step": 10447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333290587330085e-05,
+      "loss": 0.6134,
+      "step": 10448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3332264683252113e-05,
+      "loss": 0.6082,
+      "step": 10449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3331623493203385e-05,
+      "loss": 0.538,
+      "step": 10450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333098230315466e-05,
+      "loss": 0.5896,
+      "step": 10451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.333034111310593e-05,
+      "loss": 0.4359,
+      "step": 10452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33296999230572e-05,
+      "loss": 0.5013,
+      "step": 10453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3329058733008465e-05,
+      "loss": 0.5999,
+      "step": 10454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332841754295974e-05,
+      "loss": 0.5093,
+      "step": 10455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3327776352911e-05,
+      "loss": 0.5128,
+      "step": 10456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3327135162862274e-05,
+      "loss": 0.6372,
+      "step": 10457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3326493972813545e-05,
+      "loss": 0.5556,
+      "step": 10458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332585278276482e-05,
+      "loss": 0.5746,
+      "step": 10459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332521159271608e-05,
+      "loss": 0.632,
+      "step": 10460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3324570402667354e-05,
+      "loss": 0.4639,
+      "step": 10461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3323929212618625e-05,
+      "loss": 0.5415,
+      "step": 10462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332328802256989e-05,
+      "loss": 0.5699,
+      "step": 10463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332264683252116e-05,
+      "loss": 0.494,
+      "step": 10464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332200564247243e-05,
+      "loss": 0.4798,
+      "step": 10465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33213644524237e-05,
+      "loss": 0.5071,
+      "step": 10466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332072326237497e-05,
+      "loss": 0.6004,
+      "step": 10467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.332008207232624e-05,
+      "loss": 0.6362,
+      "step": 10468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331944088227751e-05,
+      "loss": 0.5825,
+      "step": 10469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331879969222878e-05,
+      "loss": 0.5839,
+      "step": 10470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331815850218005e-05,
+      "loss": 0.5479,
+      "step": 10471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3317517312131316e-05,
+      "loss": 0.5734,
+      "step": 10472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331687612208259e-05,
+      "loss": 0.6236,
+      "step": 10473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331623493203386e-05,
+      "loss": 0.5635,
+      "step": 10474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331559374198513e-05,
+      "loss": 0.5622,
+      "step": 10475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3314952551936396e-05,
+      "loss": 0.5443,
+      "step": 10476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331431136188767e-05,
+      "loss": 0.5671,
+      "step": 10477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331367017183893e-05,
+      "loss": 0.5522,
+      "step": 10478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3313028981790204e-05,
+      "loss": 0.607,
+      "step": 10479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3312387791741476e-05,
+      "loss": 0.545,
+      "step": 10480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331174660169274e-05,
+      "loss": 0.6437,
+      "step": 10481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331110541164401e-05,
+      "loss": 0.5374,
+      "step": 10482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3310464221595284e-05,
+      "loss": 0.5748,
+      "step": 10483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3309823031546556e-05,
+      "loss": 0.6666,
+      "step": 10484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330918184149782e-05,
+      "loss": 0.6265,
+      "step": 10485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330854065144909e-05,
+      "loss": 0.575,
+      "step": 10486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330789946140036e-05,
+      "loss": 0.5783,
+      "step": 10487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330725827135163e-05,
+      "loss": 0.6074,
+      "step": 10488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33066170813029e-05,
+      "loss": 0.579,
+      "step": 10489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3305975891254166e-05,
+      "loss": 0.5904,
+      "step": 10490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3305334701205444e-05,
+      "loss": 0.4889,
+      "step": 10491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330469351115671e-05,
+      "loss": 0.67,
+      "step": 10492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330405232110798e-05,
+      "loss": 0.5282,
+      "step": 10493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3303411131059246e-05,
+      "loss": 0.5395,
+      "step": 10494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330276994101052e-05,
+      "loss": 0.6308,
+      "step": 10495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330212875096178e-05,
+      "loss": 0.4893,
+      "step": 10496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3301487560913054e-05,
+      "loss": 0.554,
+      "step": 10497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3300846370864326e-05,
+      "loss": 0.5055,
+      "step": 10498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33002051808156e-05,
+      "loss": 0.5326,
+      "step": 10499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329956399076687e-05,
+      "loss": 0.5948,
+      "step": 10500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3298922800718134e-05,
+      "loss": 0.5544,
+      "step": 10501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3298281610669406e-05,
+      "loss": 0.5386,
+      "step": 10502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329764042062067e-05,
+      "loss": 0.5081,
+      "step": 10503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329699923057194e-05,
+      "loss": 0.569,
+      "step": 10504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329635804052321e-05,
+      "loss": 0.5992,
+      "step": 10505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329571685047448e-05,
+      "loss": 0.5428,
+      "step": 10506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329507566042576e-05,
+      "loss": 0.5792,
+      "step": 10507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329443447037702e-05,
+      "loss": 0.5572,
+      "step": 10508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3293793280328294e-05,
+      "loss": 0.5675,
+      "step": 10509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329315209027956e-05,
+      "loss": 0.5471,
+      "step": 10510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329251090023083e-05,
+      "loss": 0.5919,
+      "step": 10511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3291869710182096e-05,
+      "loss": 0.5497,
+      "step": 10512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329122852013337e-05,
+      "loss": 0.5285,
+      "step": 10513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.329058733008464e-05,
+      "loss": 0.5629,
+      "step": 10514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328994614003591e-05,
+      "loss": 0.5425,
+      "step": 10515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328930494998718e-05,
+      "loss": 0.6192,
+      "step": 10516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328866375993845e-05,
+      "loss": 0.6157,
+      "step": 10517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328802256988972e-05,
+      "loss": 0.5117,
+      "step": 10518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3287381379840985e-05,
+      "loss": 0.5179,
+      "step": 10519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3286740189792256e-05,
+      "loss": 0.4572,
+      "step": 10520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328609899974352e-05,
+      "loss": 0.4202,
+      "step": 10521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328545780969479e-05,
+      "loss": 0.6019,
+      "step": 10522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3284816619646065e-05,
+      "loss": 0.6175,
+      "step": 10523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3284175429597336e-05,
+      "loss": 0.4934,
+      "step": 10524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328353423954861e-05,
+      "loss": 0.5089,
+      "step": 10525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328289304949987e-05,
+      "loss": 0.4699,
+      "step": 10526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3282251859451145e-05,
+      "loss": 0.4837,
+      "step": 10527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328161066940241e-05,
+      "loss": 0.5892,
+      "step": 10528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328096947935368e-05,
+      "loss": 0.573,
+      "step": 10529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.328032828930495e-05,
+      "loss": 0.5592,
+      "step": 10530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3279687099256225e-05,
+      "loss": 0.6189,
+      "step": 10531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.327904590920749e-05,
+      "loss": 0.5323,
+      "step": 10532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.327840471915876e-05,
+      "loss": 0.465,
+      "step": 10533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.327776352911003e-05,
+      "loss": 0.487,
+      "step": 10534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32771223390613e-05,
+      "loss": 0.543,
+      "step": 10535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327648114901257e-05,
+      "loss": 0.636,
+      "step": 10536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3275839958963835e-05,
+      "loss": 0.4408,
+      "step": 10537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3275198768915107e-05,
+      "loss": 0.5124,
+      "step": 10538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327455757886638e-05,
+      "loss": 0.645,
+      "step": 10539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327391638881765e-05,
+      "loss": 0.5104,
+      "step": 10540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3273275198768915e-05,
+      "loss": 0.5175,
+      "step": 10541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3272634008720187e-05,
+      "loss": 0.5298,
+      "step": 10542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327199281867146e-05,
+      "loss": 0.5426,
+      "step": 10543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327135162862272e-05,
+      "loss": 0.5951,
+      "step": 10544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3270710438573995e-05,
+      "loss": 0.5582,
+      "step": 10545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327006924852527e-05,
+      "loss": 0.5305,
+      "step": 10546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326942805847654e-05,
+      "loss": 0.6086,
+      "step": 10547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32687868684278e-05,
+      "loss": 0.6285,
+      "step": 10548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3268145678379075e-05,
+      "loss": 0.5243,
+      "step": 10549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326750448833034e-05,
+      "loss": 0.4714,
+      "step": 10550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326686329828161e-05,
+      "loss": 0.5631,
+      "step": 10551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3266222108232883e-05,
+      "loss": 0.5136,
+      "step": 10552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326558091818415e-05,
+      "loss": 0.5383,
+      "step": 10553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326493972813542e-05,
+      "loss": 0.5671,
+      "step": 10554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326429853808669e-05,
+      "loss": 0.5071,
+      "step": 10555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3263657348037964e-05,
+      "loss": 0.5311,
+      "step": 10556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326301615798923e-05,
+      "loss": 0.5461,
+      "step": 10557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32623749679405e-05,
+      "loss": 0.5715,
+      "step": 10558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326173377789177e-05,
+      "loss": 0.5531,
+      "step": 10559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326109258784304e-05,
+      "loss": 0.5349,
+      "step": 10560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326045139779431e-05,
+      "loss": 0.6403,
+      "step": 10561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325981020774558e-05,
+      "loss": 0.6287,
+      "step": 10562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325916901769685e-05,
+      "loss": 0.6118,
+      "step": 10563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325852782764812e-05,
+      "loss": 0.506,
+      "step": 10564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325788663759939e-05,
+      "loss": 0.5973,
+      "step": 10565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3257245447550654e-05,
+      "loss": 0.5592,
+      "step": 10566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3256604257501925e-05,
+      "loss": 0.6387,
+      "step": 10567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32559630674532e-05,
+      "loss": 0.5231,
+      "step": 10568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325532187740446e-05,
+      "loss": 0.5729,
+      "step": 10569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3254680687355734e-05,
+      "loss": 0.6019,
+      "step": 10570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3254039497307005e-05,
+      "loss": 0.6298,
+      "step": 10571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325339830725828e-05,
+      "loss": 0.5212,
+      "step": 10572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325275711720954e-05,
+      "loss": 0.4963,
+      "step": 10573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3252115927160814e-05,
+      "loss": 0.5448,
+      "step": 10574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325147473711208e-05,
+      "loss": 0.5351,
+      "step": 10575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325083354706335e-05,
+      "loss": 0.5616,
+      "step": 10576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325019235701462e-05,
+      "loss": 0.6294,
+      "step": 10577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324955116696589e-05,
+      "loss": 0.5972,
+      "step": 10578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3248909976917166e-05,
+      "loss": 0.5468,
+      "step": 10579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324826878686843e-05,
+      "loss": 0.5382,
+      "step": 10580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32476275968197e-05,
+      "loss": 0.4965,
+      "step": 10581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324698640677097e-05,
+      "loss": 0.6207,
+      "step": 10582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324634521672224e-05,
+      "loss": 0.5824,
+      "step": 10583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3245704026673504e-05,
+      "loss": 0.5456,
+      "step": 10584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3245062836624776e-05,
+      "loss": 0.5344,
+      "step": 10585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324442164657605e-05,
+      "loss": 0.5985,
+      "step": 10586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324378045652732e-05,
+      "loss": 0.5676,
+      "step": 10587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324313926647859e-05,
+      "loss": 0.557,
+      "step": 10588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3242498076429856e-05,
+      "loss": 0.603,
+      "step": 10589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324185688638113e-05,
+      "loss": 0.5438,
+      "step": 10590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324121569633239e-05,
+      "loss": 0.5963,
+      "step": 10591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3240574506283664e-05,
+      "loss": 0.5673,
+      "step": 10592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323993331623493e-05,
+      "loss": 0.5945,
+      "step": 10593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32392921261862e-05,
+      "loss": 0.5322,
+      "step": 10594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323865093613748e-05,
+      "loss": 0.6162,
+      "step": 10595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3238009746088744e-05,
+      "loss": 0.6469,
+      "step": 10596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3237368556040016e-05,
+      "loss": 0.5726,
+      "step": 10597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323672736599128e-05,
+      "loss": 0.4812,
+      "step": 10598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323608617594255e-05,
+      "loss": 0.5429,
+      "step": 10599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323544498589382e-05,
+      "loss": 0.5773,
+      "step": 10600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323480379584509e-05,
+      "loss": 0.6242,
+      "step": 10601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323416260579636e-05,
+      "loss": 0.5887,
+      "step": 10602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323352141574763e-05,
+      "loss": 0.5651,
+      "step": 10603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3232880225698904e-05,
+      "loss": 0.5369,
+      "step": 10604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323223903565017e-05,
+      "loss": 0.5521,
+      "step": 10605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323159784560144e-05,
+      "loss": 0.5404,
+      "step": 10606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3230956655552706e-05,
+      "loss": 0.5614,
+      "step": 10607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323031546550398e-05,
+      "loss": 0.5627,
+      "step": 10608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322967427545524e-05,
+      "loss": 0.6952,
+      "step": 10609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3229033085406514e-05,
+      "loss": 0.5354,
+      "step": 10610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3228391895357786e-05,
+      "loss": 0.468,
+      "step": 10611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322775070530906e-05,
+      "loss": 0.6299,
+      "step": 10612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322710951526033e-05,
+      "loss": 0.5773,
+      "step": 10613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3226468325211594e-05,
+      "loss": 0.6349,
+      "step": 10614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3225827135162866e-05,
+      "loss": 0.5098,
+      "step": 10615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322518594511413e-05,
+      "loss": 0.51,
+      "step": 10616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32245447550654e-05,
+      "loss": 0.5029,
+      "step": 10617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3223903565016674e-05,
+      "loss": 0.5604,
+      "step": 10618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3223262374967946e-05,
+      "loss": 0.6018,
+      "step": 10619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322262118491921e-05,
+      "loss": 0.5437,
+      "step": 10620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322197999487048e-05,
+      "loss": 0.4634,
+      "step": 10621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3221338804821755e-05,
+      "loss": 0.6021,
+      "step": 10622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322069761477302e-05,
+      "loss": 0.5789,
+      "step": 10623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322005642472429e-05,
+      "loss": 0.5348,
+      "step": 10624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3219415234675556e-05,
+      "loss": 0.4879,
+      "step": 10625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321877404462683e-05,
+      "loss": 0.5904,
+      "step": 10626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32181328545781e-05,
+      "loss": 0.6495,
+      "step": 10627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321749166452937e-05,
+      "loss": 0.6355,
+      "step": 10628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3216850474480636e-05,
+      "loss": 0.555,
+      "step": 10629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321620928443191e-05,
+      "loss": 0.5594,
+      "step": 10630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321556809438318e-05,
+      "loss": 0.5524,
+      "step": 10631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3214926904334445e-05,
+      "loss": 0.586,
+      "step": 10632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3214285714285716e-05,
+      "loss": 0.5796,
+      "step": 10633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321364452423699e-05,
+      "loss": 0.564,
+      "step": 10634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321300333418826e-05,
+      "loss": 0.4877,
+      "step": 10635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3212362144139525e-05,
+      "loss": 0.5309,
+      "step": 10636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3211720954090796e-05,
+      "loss": 0.5692,
+      "step": 10637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321107976404206e-05,
+      "loss": 0.5416,
+      "step": 10638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321043857399333e-05,
+      "loss": 0.4386,
+      "step": 10639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3209797383944605e-05,
+      "loss": 0.6329,
+      "step": 10640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320915619389587e-05,
+      "loss": 0.5668,
+      "step": 10641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320851500384714e-05,
+      "loss": 0.5828,
+      "step": 10642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320787381379841e-05,
+      "loss": 0.5407,
+      "step": 10643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3207232623749685e-05,
+      "loss": 0.6059,
+      "step": 10644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320659143370095e-05,
+      "loss": 0.4572,
+      "step": 10645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320595024365222e-05,
+      "loss": 0.4968,
+      "step": 10646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3205309053603486e-05,
+      "loss": 0.5006,
+      "step": 10647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320466786355476e-05,
+      "loss": 0.627,
+      "step": 10648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320402667350603e-05,
+      "loss": 0.5166,
+      "step": 10649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3203385483457295e-05,
+      "loss": 0.5497,
+      "step": 10650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320274429340857e-05,
+      "loss": 0.6141,
+      "step": 10651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320210310335984e-05,
+      "loss": 0.5275,
+      "step": 10652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320146191331111e-05,
+      "loss": 0.5333,
+      "step": 10653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3200820723262375e-05,
+      "loss": 0.5552,
+      "step": 10654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320017953321365e-05,
+      "loss": 0.5327,
+      "step": 10655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319953834316491e-05,
+      "loss": 0.6137,
+      "step": 10656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319889715311618e-05,
+      "loss": 0.4906,
+      "step": 10657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3198255963067455e-05,
+      "loss": 0.501,
+      "step": 10658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319761477301873e-05,
+      "loss": 0.6508,
+      "step": 10659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319697358297e-05,
+      "loss": 0.5431,
+      "step": 10660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319633239292126e-05,
+      "loss": 0.6786,
+      "step": 10661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3195691202872535e-05,
+      "loss": 0.5946,
+      "step": 10662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.31950500128238e-05,
+      "loss": 0.5442,
+      "step": 10663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319440882277507e-05,
+      "loss": 0.5613,
+      "step": 10664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319376763272634e-05,
+      "loss": 0.7859,
+      "step": 10665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319312644267761e-05,
+      "loss": 0.5275,
+      "step": 10666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319248525262889e-05,
+      "loss": 0.5705,
+      "step": 10667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319184406258015e-05,
+      "loss": 0.5419,
+      "step": 10668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3191202872531424e-05,
+      "loss": 0.4872,
+      "step": 10669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319056168248269e-05,
+      "loss": 0.5203,
+      "step": 10670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318992049243396e-05,
+      "loss": 0.4988,
+      "step": 10671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3189279302385225e-05,
+      "loss": 0.5341,
+      "step": 10672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.31886381123365e-05,
+      "loss": 0.569,
+      "step": 10673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318799692228777e-05,
+      "loss": 0.5567,
+      "step": 10674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318735573223904e-05,
+      "loss": 0.4583,
+      "step": 10675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318671454219031e-05,
+      "loss": 0.6393,
+      "step": 10676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318607335214158e-05,
+      "loss": 0.5632,
+      "step": 10677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318543216209285e-05,
+      "loss": 0.5851,
+      "step": 10678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3184790972044114e-05,
+      "loss": 0.6287,
+      "step": 10679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3184149781995385e-05,
+      "loss": 0.5043,
+      "step": 10680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318350859194665e-05,
+      "loss": 0.5928,
+      "step": 10681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318286740189792e-05,
+      "loss": 0.5698,
+      "step": 10682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3182226211849194e-05,
+      "loss": 0.6731,
+      "step": 10683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3181585021800465e-05,
+      "loss": 0.6382,
+      "step": 10684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.318094383175174e-05,
+      "loss": 0.6463,
+      "step": 10685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3180302641703e-05,
+      "loss": 0.634,
+      "step": 10686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3179661451654274e-05,
+      "loss": 0.641,
+      "step": 10687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317902026160554e-05,
+      "loss": 0.6659,
+      "step": 10688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317837907155681e-05,
+      "loss": 0.6661,
+      "step": 10689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317773788150808e-05,
+      "loss": 0.5192,
+      "step": 10690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3177096691459354e-05,
+      "loss": 0.5228,
+      "step": 10691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317645550141062e-05,
+      "loss": 0.5814,
+      "step": 10692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317581431136189e-05,
+      "loss": 0.6213,
+      "step": 10693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317517312131316e-05,
+      "loss": 0.5874,
+      "step": 10694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317453193126443e-05,
+      "loss": 0.6111,
+      "step": 10695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.31738907412157e-05,
+      "loss": 0.5389,
+      "step": 10696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3173249551166964e-05,
+      "loss": 0.564,
+      "step": 10697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3172608361118236e-05,
+      "loss": 0.5662,
+      "step": 10698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317196717106951e-05,
+      "loss": 0.5676,
+      "step": 10699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317132598102078e-05,
+      "loss": 0.5143,
+      "step": 10700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3170684790972044e-05,
+      "loss": 0.5045,
+      "step": 10701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3170043600923316e-05,
+      "loss": 0.5154,
+      "step": 10702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316940241087459e-05,
+      "loss": 0.6139,
+      "step": 10703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316876122082585e-05,
+      "loss": 0.6055,
+      "step": 10704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3168120030777124e-05,
+      "loss": 0.6913,
+      "step": 10705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3167478840728396e-05,
+      "loss": 0.5529,
+      "step": 10706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316683765067967e-05,
+      "loss": 0.5308,
+      "step": 10707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316619646063093e-05,
+      "loss": 0.5512,
+      "step": 10708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3165555270582204e-05,
+      "loss": 0.4446,
+      "step": 10709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316491408053347e-05,
+      "loss": 0.5572,
+      "step": 10710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316427289048474e-05,
+      "loss": 0.5577,
+      "step": 10711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316363170043601e-05,
+      "loss": 0.5864,
+      "step": 10712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316299051038728e-05,
+      "loss": 0.5895,
+      "step": 10713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316234932033855e-05,
+      "loss": 0.6783,
+      "step": 10714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316170813028982e-05,
+      "loss": 0.4197,
+      "step": 10715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316106694024109e-05,
+      "loss": 0.678,
+      "step": 10716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316042575019236e-05,
+      "loss": 0.4791,
+      "step": 10717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315978456014363e-05,
+      "loss": 0.5929,
+      "step": 10718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3159143370094894e-05,
+      "loss": 0.4489,
+      "step": 10719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3158502180046166e-05,
+      "loss": 0.5855,
+      "step": 10720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315786098999744e-05,
+      "loss": 0.5083,
+      "step": 10721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315721979994871e-05,
+      "loss": 0.5366,
+      "step": 10722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315657860989998e-05,
+      "loss": 0.5837,
+      "step": 10723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3155937419851246e-05,
+      "loss": 0.538,
+      "step": 10724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315529622980252e-05,
+      "loss": 0.5899,
+      "step": 10725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315465503975378e-05,
+      "loss": 0.6297,
+      "step": 10726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3154013849705054e-05,
+      "loss": 0.5676,
+      "step": 10727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315337265965632e-05,
+      "loss": 0.5502,
+      "step": 10728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315273146960759e-05,
+      "loss": 0.5327,
+      "step": 10729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315209027955886e-05,
+      "loss": 0.5431,
+      "step": 10730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3151449089510134e-05,
+      "loss": 0.6054,
+      "step": 10731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3150807899461406e-05,
+      "loss": 0.5551,
+      "step": 10732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315016670941267e-05,
+      "loss": 0.5374,
+      "step": 10733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314952551936394e-05,
+      "loss": 0.5438,
+      "step": 10734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314888432931521e-05,
+      "loss": 0.5235,
+      "step": 10735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314824313926648e-05,
+      "loss": 0.5163,
+      "step": 10736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3147601949217744e-05,
+      "loss": 0.538,
+      "step": 10737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3146960759169016e-05,
+      "loss": 0.6064,
+      "step": 10738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3146319569120295e-05,
+      "loss": 0.5658,
+      "step": 10739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314567837907156e-05,
+      "loss": 0.688,
+      "step": 10740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314503718902283e-05,
+      "loss": 0.6604,
+      "step": 10741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3144395998974096e-05,
+      "loss": 0.6046,
+      "step": 10742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314375480892537e-05,
+      "loss": 0.5659,
+      "step": 10743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314311361887663e-05,
+      "loss": 0.5359,
+      "step": 10744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3142472428827905e-05,
+      "loss": 0.5319,
+      "step": 10745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3141831238779176e-05,
+      "loss": 0.5426,
+      "step": 10746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314119004873045e-05,
+      "loss": 0.5482,
+      "step": 10747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.314054885868172e-05,
+      "loss": 0.4899,
+      "step": 10748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3139907668632985e-05,
+      "loss": 0.5766,
+      "step": 10749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3139266478584256e-05,
+      "loss": 0.6562,
+      "step": 10750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313862528853552e-05,
+      "loss": 0.5157,
+      "step": 10751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313798409848679e-05,
+      "loss": 0.66,
+      "step": 10752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313734290843806e-05,
+      "loss": 0.5796,
+      "step": 10753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313670171838933e-05,
+      "loss": 0.5783,
+      "step": 10754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313606052834061e-05,
+      "loss": 0.6771,
+      "step": 10755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313541933829187e-05,
+      "loss": 0.5214,
+      "step": 10756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3134778148243145e-05,
+      "loss": 0.4907,
+      "step": 10757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313413695819441e-05,
+      "loss": 0.6055,
+      "step": 10758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313349576814568e-05,
+      "loss": 0.5684,
+      "step": 10759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3132854578096946e-05,
+      "loss": 0.4504,
+      "step": 10760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313221338804822e-05,
+      "loss": 0.5958,
+      "step": 10761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313157219799949e-05,
+      "loss": 0.5781,
+      "step": 10762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313093100795076e-05,
+      "loss": 0.6041,
+      "step": 10763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313028981790203e-05,
+      "loss": 0.6015,
+      "step": 10764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.31296486278533e-05,
+      "loss": 0.6041,
+      "step": 10765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312900743780457e-05,
+      "loss": 0.5744,
+      "step": 10766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3128366247755835e-05,
+      "loss": 0.6132,
+      "step": 10767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312772505770711e-05,
+      "loss": 0.5374,
+      "step": 10768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312708386765837e-05,
+      "loss": 0.5042,
+      "step": 10769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312644267760964e-05,
+      "loss": 0.5222,
+      "step": 10770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3125801487560915e-05,
+      "loss": 0.6937,
+      "step": 10771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312516029751219e-05,
+      "loss": 0.6106,
+      "step": 10772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312451910746346e-05,
+      "loss": 0.5995,
+      "step": 10773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3123877917414723e-05,
+      "loss": 0.5728,
+      "step": 10774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3123236727365995e-05,
+      "loss": 0.5534,
+      "step": 10775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312259553731726e-05,
+      "loss": 0.5444,
+      "step": 10776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312195434726853e-05,
+      "loss": 0.5323,
+      "step": 10777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3121313157219804e-05,
+      "loss": 0.5264,
+      "step": 10778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3120671967171075e-05,
+      "loss": 0.5253,
+      "step": 10779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312003077712234e-05,
+      "loss": 0.6138,
+      "step": 10780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311938958707361e-05,
+      "loss": 0.5514,
+      "step": 10781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3118748397024884e-05,
+      "loss": 0.494,
+      "step": 10782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311810720697615e-05,
+      "loss": 0.5244,
+      "step": 10783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311746601692742e-05,
+      "loss": 0.582,
+      "step": 10784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3116824826878685e-05,
+      "loss": 0.6602,
+      "step": 10785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311618363682996e-05,
+      "loss": 0.5654,
+      "step": 10786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311554244678123e-05,
+      "loss": 0.6319,
+      "step": 10787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.31149012567325e-05,
+      "loss": 0.5218,
+      "step": 10788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3114260066683765e-05,
+      "loss": 0.6221,
+      "step": 10789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311361887663504e-05,
+      "loss": 0.568,
+      "step": 10790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311297768658631e-05,
+      "loss": 0.6765,
+      "step": 10791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3112336496537574e-05,
+      "loss": 0.5534,
+      "step": 10792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3111695306488845e-05,
+      "loss": 0.586,
+      "step": 10793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.311105411644012e-05,
+      "loss": 0.5613,
+      "step": 10794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.311041292639139e-05,
+      "loss": 0.5148,
+      "step": 10795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3109771736342654e-05,
+      "loss": 0.5765,
+      "step": 10796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3109130546293925e-05,
+      "loss": 0.5288,
+      "step": 10797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310848935624519e-05,
+      "loss": 0.5879,
+      "step": 10798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310784816619646e-05,
+      "loss": 0.4958,
+      "step": 10799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3107206976147734e-05,
+      "loss": 0.4794,
+      "step": 10800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3106565786099e-05,
+      "loss": 0.4931,
+      "step": 10801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310592459605027e-05,
+      "loss": 0.5421,
+      "step": 10802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310528340600154e-05,
+      "loss": 0.5578,
+      "step": 10803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3104642215952814e-05,
+      "loss": 0.5261,
+      "step": 10804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310400102590408e-05,
+      "loss": 0.5571,
+      "step": 10805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310335983585535e-05,
+      "loss": 0.5894,
+      "step": 10806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3102718645806616e-05,
+      "loss": 0.6482,
+      "step": 10807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310207745575789e-05,
+      "loss": 0.5499,
+      "step": 10808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310143626570916e-05,
+      "loss": 0.4607,
+      "step": 10809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310079507566043e-05,
+      "loss": 0.5385,
+      "step": 10810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.31001538856117e-05,
+      "loss": 0.5874,
+      "step": 10811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309951269556297e-05,
+      "loss": 0.5197,
+      "step": 10812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309887150551424e-05,
+      "loss": 0.6077,
+      "step": 10813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3098230315465504e-05,
+      "loss": 0.5712,
+      "step": 10814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3097589125416776e-05,
+      "loss": 0.5524,
+      "step": 10815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309694793536804e-05,
+      "loss": 0.5976,
+      "step": 10816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309630674531931e-05,
+      "loss": 0.6141,
+      "step": 10817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3095665555270584e-05,
+      "loss": 0.5275,
+      "step": 10818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3095024365221856e-05,
+      "loss": 0.6217,
+      "step": 10819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309438317517313e-05,
+      "loss": 0.5492,
+      "step": 10820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309374198512439e-05,
+      "loss": 0.6037,
+      "step": 10821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3093100795075664e-05,
+      "loss": 0.5357,
+      "step": 10822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309245960502693e-05,
+      "loss": 0.4759,
+      "step": 10823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30918184149782e-05,
+      "loss": 0.5691,
+      "step": 10824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3091177224929466e-05,
+      "loss": 0.5554,
+      "step": 10825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309053603488074e-05,
+      "loss": 0.5886,
+      "step": 10826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3089894844832016e-05,
+      "loss": 0.5984,
+      "step": 10827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308925365478328e-05,
+      "loss": 0.4907,
+      "step": 10828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308861246473455e-05,
+      "loss": 0.5249,
+      "step": 10829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308797127468582e-05,
+      "loss": 0.6663,
+      "step": 10830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308733008463709e-05,
+      "loss": 0.5363,
+      "step": 10831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3086688894588354e-05,
+      "loss": 0.4861,
+      "step": 10832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3086047704539626e-05,
+      "loss": 0.6095,
+      "step": 10833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30854065144909e-05,
+      "loss": 0.5304,
+      "step": 10834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308476532444217e-05,
+      "loss": 0.576,
+      "step": 10835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308412413439344e-05,
+      "loss": 0.5088,
+      "step": 10836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3083482944344706e-05,
+      "loss": 0.621,
+      "step": 10837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308284175429598e-05,
+      "loss": 0.4422,
+      "step": 10838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308220056424724e-05,
+      "loss": 0.5612,
+      "step": 10839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3081559374198514e-05,
+      "loss": 0.5582,
+      "step": 10840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308091818414978e-05,
+      "loss": 0.6001,
+      "step": 10841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308027699410105e-05,
+      "loss": 0.4859,
+      "step": 10842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307963580405232e-05,
+      "loss": 0.5482,
+      "step": 10843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3078994614003595e-05,
+      "loss": 0.5584,
+      "step": 10844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3078353423954866e-05,
+      "loss": 0.5555,
+      "step": 10845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307771223390613e-05,
+      "loss": 0.5539,
+      "step": 10846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30770710438574e-05,
+      "loss": 0.5547,
+      "step": 10847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307642985380867e-05,
+      "loss": 0.5814,
+      "step": 10848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307578866375994e-05,
+      "loss": 0.5033,
+      "step": 10849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307514747371121e-05,
+      "loss": 0.601,
+      "step": 10850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307450628366248e-05,
+      "loss": 0.6035,
+      "step": 10851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307386509361375e-05,
+      "loss": 0.5173,
+      "step": 10852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307322390356502e-05,
+      "loss": 0.5747,
+      "step": 10853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307258271351629e-05,
+      "loss": 0.6409,
+      "step": 10854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3071941523467556e-05,
+      "loss": 0.6004,
+      "step": 10855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307130033341883e-05,
+      "loss": 0.4381,
+      "step": 10856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307065914337009e-05,
+      "loss": 0.6282,
+      "step": 10857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3070017953321365e-05,
+      "loss": 0.537,
+      "step": 10858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3069376763272636e-05,
+      "loss": 0.6275,
+      "step": 10859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306873557322391e-05,
+      "loss": 0.6944,
+      "step": 10860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306809438317517e-05,
+      "loss": 0.6612,
+      "step": 10861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3067453193126445e-05,
+      "loss": 0.5943,
+      "step": 10862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3066812003077716e-05,
+      "loss": 0.5628,
+      "step": 10863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306617081302898e-05,
+      "loss": 0.5223,
+      "step": 10864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306552962298025e-05,
+      "loss": 0.5444,
+      "step": 10865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3064888432931525e-05,
+      "loss": 0.5843,
+      "step": 10866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3064247242882797e-05,
+      "loss": 0.6346,
+      "step": 10867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306360605283406e-05,
+      "loss": 0.5229,
+      "step": 10868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306296486278533e-05,
+      "loss": 0.4958,
+      "step": 10869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30623236727366e-05,
+      "loss": 0.4933,
+      "step": 10870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306168248268787e-05,
+      "loss": 0.4624,
+      "step": 10871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306104129263914e-05,
+      "loss": 0.5585,
+      "step": 10872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3060400102590407e-05,
+      "loss": 0.6255,
+      "step": 10873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305975891254168e-05,
+      "loss": 0.5577,
+      "step": 10874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305911772249295e-05,
+      "loss": 0.4542,
+      "step": 10875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305847653244422e-05,
+      "loss": 0.5693,
+      "step": 10876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305783534239549e-05,
+      "loss": 0.5732,
+      "step": 10877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305719415234676e-05,
+      "loss": 0.505,
+      "step": 10878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305655296229802e-05,
+      "loss": 0.4996,
+      "step": 10879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3055911772249295e-05,
+      "loss": 0.5364,
+      "step": 10880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305527058220057e-05,
+      "loss": 0.6477,
+      "step": 10881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305462939215184e-05,
+      "loss": 0.5106,
+      "step": 10882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305398820210311e-05,
+      "loss": 0.6276,
+      "step": 10883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3053347012054375e-05,
+      "loss": 0.5167,
+      "step": 10884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305270582200565e-05,
+      "loss": 0.5444,
+      "step": 10885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305206463195691e-05,
+      "loss": 0.5531,
+      "step": 10886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3051423441908183e-05,
+      "loss": 0.6037,
+      "step": 10887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305078225185945e-05,
+      "loss": 0.5392,
+      "step": 10888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305014106181072e-05,
+      "loss": 0.5169,
+      "step": 10889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304949987176199e-05,
+      "loss": 0.5129,
+      "step": 10890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3048858681713264e-05,
+      "loss": 0.5354,
+      "step": 10891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3048217491664535e-05,
+      "loss": 0.5744,
+      "step": 10892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30475763016158e-05,
+      "loss": 0.631,
+      "step": 10893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304693511156707e-05,
+      "loss": 0.6455,
+      "step": 10894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304629392151834e-05,
+      "loss": 0.6297,
+      "step": 10895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304565273146961e-05,
+      "loss": 0.4859,
+      "step": 10896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3045011541420874e-05,
+      "loss": 0.5539,
+      "step": 10897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3044370351372145e-05,
+      "loss": 0.5926,
+      "step": 10898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3043729161323424e-05,
+      "loss": 0.5468,
+      "step": 10899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304308797127469e-05,
+      "loss": 0.5328,
+      "step": 10900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304244678122596e-05,
+      "loss": 0.6191,
+      "step": 10901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3041805591177225e-05,
+      "loss": 0.6095,
+      "step": 10902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30411644011285e-05,
+      "loss": 0.648,
+      "step": 10903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304052321107976e-05,
+      "loss": 0.5054,
+      "step": 10904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3039882021031034e-05,
+      "loss": 0.5029,
+      "step": 10905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3039240830982305e-05,
+      "loss": 0.6157,
+      "step": 10906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303859964093358e-05,
+      "loss": 0.5902,
+      "step": 10907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303795845088485e-05,
+      "loss": 0.573,
+      "step": 10908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3037317260836114e-05,
+      "loss": 0.5734,
+      "step": 10909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3036676070787386e-05,
+      "loss": 0.5734,
+      "step": 10910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303603488073865e-05,
+      "loss": 0.559,
+      "step": 10911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303539369068992e-05,
+      "loss": 0.4641,
+      "step": 10912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303475250064119e-05,
+      "loss": 0.469,
+      "step": 10913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303411131059246e-05,
+      "loss": 0.6319,
+      "step": 10914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303347012054373e-05,
+      "loss": 0.5635,
+      "step": 10915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3032828930495e-05,
+      "loss": 0.6754,
+      "step": 10916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3032187740446274e-05,
+      "loss": 0.5495,
+      "step": 10917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303154655039754e-05,
+      "loss": 0.5653,
+      "step": 10918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303090536034881e-05,
+      "loss": 0.4938,
+      "step": 10919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3030264170300076e-05,
+      "loss": 0.5913,
+      "step": 10920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302962298025135e-05,
+      "loss": 0.5275,
+      "step": 10921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302898179020262e-05,
+      "loss": 0.5254,
+      "step": 10922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302834060015389e-05,
+      "loss": 0.5314,
+      "step": 10923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3027699410105156e-05,
+      "loss": 0.4998,
+      "step": 10924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302705822005643e-05,
+      "loss": 0.5098,
+      "step": 10925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30264170300077e-05,
+      "loss": 0.6392,
+      "step": 10926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3025775839958964e-05,
+      "loss": 0.532,
+      "step": 10927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3025134649910236e-05,
+      "loss": 0.5849,
+      "step": 10928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30244934598615e-05,
+      "loss": 0.5505,
+      "step": 10929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302385226981277e-05,
+      "loss": 0.6384,
+      "step": 10930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3023211079764044e-05,
+      "loss": 0.5864,
+      "step": 10931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3022569889715316e-05,
+      "loss": 0.5433,
+      "step": 10932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302192869966658e-05,
+      "loss": 0.596,
+      "step": 10933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302128750961785e-05,
+      "loss": 0.6687,
+      "step": 10934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3020646319569124e-05,
+      "loss": 0.4802,
+      "step": 10935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302000512952039e-05,
+      "loss": 0.5995,
+      "step": 10936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301936393947166e-05,
+      "loss": 0.5152,
+      "step": 10937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301872274942293e-05,
+      "loss": 0.5101,
+      "step": 10938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3018081559374204e-05,
+      "loss": 0.5139,
+      "step": 10939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301744036932547e-05,
+      "loss": 0.5452,
+      "step": 10940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301679917927674e-05,
+      "loss": 0.5695,
+      "step": 10941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301615798922801e-05,
+      "loss": 0.5496,
+      "step": 10942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301551679917928e-05,
+      "loss": 0.5691,
+      "step": 10943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301487560913055e-05,
+      "loss": 0.7062,
+      "step": 10944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3014234419081814e-05,
+      "loss": 0.4894,
+      "step": 10945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3013593229033086e-05,
+      "loss": 0.5588,
+      "step": 10946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301295203898436e-05,
+      "loss": 0.6613,
+      "step": 10947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301231084893563e-05,
+      "loss": 0.5491,
+      "step": 10948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3011669658886894e-05,
+      "loss": 0.566,
+      "step": 10949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3011028468838166e-05,
+      "loss": 0.5573,
+      "step": 10950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.301038727878944e-05,
+      "loss": 0.6059,
+      "step": 10951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30097460887407e-05,
+      "loss": 0.5569,
+      "step": 10952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3009104898691974e-05,
+      "loss": 0.5638,
+      "step": 10953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3008463708643246e-05,
+      "loss": 0.5903,
+      "step": 10954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300782251859452e-05,
+      "loss": 0.5404,
+      "step": 10955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300718132854578e-05,
+      "loss": 0.5389,
+      "step": 10956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3006540138497055e-05,
+      "loss": 0.5905,
+      "step": 10957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300589894844832e-05,
+      "loss": 0.5474,
+      "step": 10958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300525775839959e-05,
+      "loss": 0.4842,
+      "step": 10959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300461656835086e-05,
+      "loss": 0.5831,
+      "step": 10960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300397537830213e-05,
+      "loss": 0.5914,
+      "step": 10961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30033341882534e-05,
+      "loss": 0.5244,
+      "step": 10962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300269299820467e-05,
+      "loss": 0.4438,
+      "step": 10963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300205180815594e-05,
+      "loss": 0.5266,
+      "step": 10964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300141061810721e-05,
+      "loss": 0.5588,
+      "step": 10965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.300076942805848e-05,
+      "loss": 0.5849,
+      "step": 10966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3000128238009745e-05,
+      "loss": 0.6447,
+      "step": 10967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2999487047961016e-05,
+      "loss": 0.5947,
+      "step": 10968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299884585791229e-05,
+      "loss": 0.5452,
+      "step": 10969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299820466786356e-05,
+      "loss": 0.5548,
+      "step": 10970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299756347781483e-05,
+      "loss": 0.5662,
+      "step": 10971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2996922287766096e-05,
+      "loss": 0.3963,
+      "step": 10972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299628109771737e-05,
+      "loss": 0.578,
+      "step": 10973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299563990766863e-05,
+      "loss": 0.6119,
+      "step": 10974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2994998717619905e-05,
+      "loss": 0.5141,
+      "step": 10975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299435752757117e-05,
+      "loss": 0.6186,
+      "step": 10976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299371633752244e-05,
+      "loss": 0.5467,
+      "step": 10977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299307514747371e-05,
+      "loss": 0.5843,
+      "step": 10978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2992433957424985e-05,
+      "loss": 0.4997,
+      "step": 10979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2991792767376257e-05,
+      "loss": 0.5369,
+      "step": 10980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299115157732752e-05,
+      "loss": 0.4702,
+      "step": 10981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.299051038727879e-05,
+      "loss": 0.547,
+      "step": 10982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298986919723006e-05,
+      "loss": 0.4823,
+      "step": 10983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298922800718133e-05,
+      "loss": 0.5211,
+      "step": 10984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2988586817132595e-05,
+      "loss": 0.7369,
+      "step": 10985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2987945627083867e-05,
+      "loss": 0.6876,
+      "step": 10986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2987304437035145e-05,
+      "loss": 0.5051,
+      "step": 10987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298666324698641e-05,
+      "loss": 0.4301,
+      "step": 10988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298602205693768e-05,
+      "loss": 0.5199,
+      "step": 10989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298538086688895e-05,
+      "loss": 0.6557,
+      "step": 10990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298473967684022e-05,
+      "loss": 0.5624,
+      "step": 10991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298409848679148e-05,
+      "loss": 0.6043,
+      "step": 10992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2983457296742755e-05,
+      "loss": 0.5968,
+      "step": 10993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298281610669403e-05,
+      "loss": 0.4944,
+      "step": 10994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.29821749166453e-05,
+      "loss": 0.5577,
+      "step": 10995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298153372659657e-05,
+      "loss": 0.5131,
+      "step": 10996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2980892536547835e-05,
+      "loss": 0.5157,
+      "step": 10997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.298025134649911e-05,
+      "loss": 0.6975,
+      "step": 10998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297961015645037e-05,
+      "loss": 0.5118,
+      "step": 10999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2978968966401643e-05,
+      "loss": 0.559,
+      "step": 11000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297832777635291e-05,
+      "loss": 0.6917,
+      "step": 11001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297768658630418e-05,
+      "loss": 0.5821,
+      "step": 11002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297704539625545e-05,
+      "loss": 0.5963,
+      "step": 11003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2976404206206724e-05,
+      "loss": 0.5341,
+      "step": 11004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2975763016157995e-05,
+      "loss": 0.5514,
+      "step": 11005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297512182610926e-05,
+      "loss": 0.57,
+      "step": 11006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297448063606053e-05,
+      "loss": 0.5928,
+      "step": 11007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.29738394460118e-05,
+      "loss": 0.5383,
+      "step": 11008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297319825596307e-05,
+      "loss": 0.5097,
+      "step": 11009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297255706591434e-05,
+      "loss": 0.5665,
+      "step": 11010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297191587586561e-05,
+      "loss": 0.6282,
+      "step": 11011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297127468581688e-05,
+      "loss": 0.5791,
+      "step": 11012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297063349576815e-05,
+      "loss": 0.5074,
+      "step": 11013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296999230571942e-05,
+      "loss": 0.5802,
+      "step": 11014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2969351115670685e-05,
+      "loss": 0.5946,
+      "step": 11015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296870992562196e-05,
+      "loss": 0.5764,
+      "step": 11016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296806873557322e-05,
+      "loss": 0.5723,
+      "step": 11017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2967427545524494e-05,
+      "loss": 0.6295,
+      "step": 11018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2966786355475765e-05,
+      "loss": 0.557,
+      "step": 11019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296614516542704e-05,
+      "loss": 0.4723,
+      "step": 11020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.29655039753783e-05,
+      "loss": 0.5742,
+      "step": 11021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2964862785329574e-05,
+      "loss": 0.6388,
+      "step": 11022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2964221595280846e-05,
+      "loss": 0.566,
+      "step": 11023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296358040523211e-05,
+      "loss": 0.5858,
+      "step": 11024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296293921518338e-05,
+      "loss": 0.4875,
+      "step": 11025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2962298025134654e-05,
+      "loss": 0.5322,
+      "step": 11026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2961656835085926e-05,
+      "loss": 0.6143,
+      "step": 11027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296101564503719e-05,
+      "loss": 0.506,
+      "step": 11028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.296037445498846e-05,
+      "loss": 0.6391,
+      "step": 11029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295973326493973e-05,
+      "loss": 0.4583,
+      "step": 11030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2959092074891e-05,
+      "loss": 0.6095,
+      "step": 11031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295845088484227e-05,
+      "loss": 0.4393,
+      "step": 11032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2957809694793536e-05,
+      "loss": 0.6214,
+      "step": 11033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295716850474481e-05,
+      "loss": 0.5607,
+      "step": 11034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295652731469608e-05,
+      "loss": 0.6088,
+      "step": 11035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295588612464735e-05,
+      "loss": 0.548,
+      "step": 11036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2955244934598616e-05,
+      "loss": 0.6845,
+      "step": 11037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295460374454989e-05,
+      "loss": 0.5638,
+      "step": 11038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295396255450115e-05,
+      "loss": 0.6512,
+      "step": 11039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2953321364452424e-05,
+      "loss": 0.58,
+      "step": 11040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2952680174403696e-05,
+      "loss": 0.5676,
+      "step": 11041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295203898435497e-05,
+      "loss": 0.5793,
+      "step": 11042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295139779430624e-05,
+      "loss": 0.5716,
+      "step": 11043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2950756604257504e-05,
+      "loss": 0.5894,
+      "step": 11044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2950115414208776e-05,
+      "loss": 0.6467,
+      "step": 11045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.294947422416004e-05,
+      "loss": 0.5495,
+      "step": 11046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.294883303411131e-05,
+      "loss": 0.5982,
+      "step": 11047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.294819184406258e-05,
+      "loss": 0.5022,
+      "step": 11048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.294755065401385e-05,
+      "loss": 0.5505,
+      "step": 11049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.294690946396512e-05,
+      "loss": 0.5139,
+      "step": 11050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.294626827391639e-05,
+      "loss": 0.514,
+      "step": 11051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2945627083867664e-05,
+      "loss": 0.5644,
+      "step": 11052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.294498589381893e-05,
+      "loss": 0.5593,
+      "step": 11053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.29443447037702e-05,
+      "loss": 0.7057,
+      "step": 11054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2943703513721466e-05,
+      "loss": 0.513,
+      "step": 11055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.294306232367274e-05,
+      "loss": 0.4898,
+      "step": 11056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2942421133624e-05,
+      "loss": 0.4919,
+      "step": 11057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2941779943575274e-05,
+      "loss": 0.5887,
+      "step": 11058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.294113875352655e-05,
+      "loss": 0.5578,
+      "step": 11059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.294049756347782e-05,
+      "loss": 0.4903,
+      "step": 11060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293985637342909e-05,
+      "loss": 0.5798,
+      "step": 11061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2939215183380354e-05,
+      "loss": 0.5505,
+      "step": 11062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2938573993331626e-05,
+      "loss": 0.5847,
+      "step": 11063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293793280328289e-05,
+      "loss": 0.5069,
+      "step": 11064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293729161323416e-05,
+      "loss": 0.5128,
+      "step": 11065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2936650423185434e-05,
+      "loss": 0.5794,
+      "step": 11066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2936009233136706e-05,
+      "loss": 0.5605,
+      "step": 11067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293536804308798e-05,
+      "loss": 0.493,
+      "step": 11068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293472685303924e-05,
+      "loss": 0.5111,
+      "step": 11069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2934085662990515e-05,
+      "loss": 0.6134,
+      "step": 11070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293344447294178e-05,
+      "loss": 0.591,
+      "step": 11071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293280328289305e-05,
+      "loss": 0.5064,
+      "step": 11072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2932162092844316e-05,
+      "loss": 0.6081,
+      "step": 11073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293152090279559e-05,
+      "loss": 0.5016,
+      "step": 11074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293087971274686e-05,
+      "loss": 0.6032,
+      "step": 11075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293023852269813e-05,
+      "loss": 0.7387,
+      "step": 11076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29295973326494e-05,
+      "loss": 0.5303,
+      "step": 11077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292895614260067e-05,
+      "loss": 0.472,
+      "step": 11078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292831495255194e-05,
+      "loss": 0.6947,
+      "step": 11079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2927673762503205e-05,
+      "loss": 0.6303,
+      "step": 11080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2927032572454476e-05,
+      "loss": 0.6243,
+      "step": 11081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292639138240575e-05,
+      "loss": 0.5504,
+      "step": 11082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292575019235702e-05,
+      "loss": 0.5685,
+      "step": 11083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2925109002308285e-05,
+      "loss": 0.5541,
+      "step": 11084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2924467812259556e-05,
+      "loss": 0.5258,
+      "step": 11085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292382662221083e-05,
+      "loss": 0.5249,
+      "step": 11086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292318543216209e-05,
+      "loss": 0.5669,
+      "step": 11087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2922544242113365e-05,
+      "loss": 0.5418,
+      "step": 11088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292190305206463e-05,
+      "loss": 0.6463,
+      "step": 11089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29212618620159e-05,
+      "loss": 0.647,
+      "step": 11090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292062067196717e-05,
+      "loss": 0.4851,
+      "step": 11091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2919979481918445e-05,
+      "loss": 0.5483,
+      "step": 11092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291933829186971e-05,
+      "loss": 0.6409,
+      "step": 11093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291869710182098e-05,
+      "loss": 0.6248,
+      "step": 11094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291805591177225e-05,
+      "loss": 0.5995,
+      "step": 11095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291741472172352e-05,
+      "loss": 0.6039,
+      "step": 11096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291677353167479e-05,
+      "loss": 0.5248,
+      "step": 11097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291613234162606e-05,
+      "loss": 0.5657,
+      "step": 11098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291549115157733e-05,
+      "loss": 0.6847,
+      "step": 11099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29148499615286e-05,
+      "loss": 0.4837,
+      "step": 11100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291420877147987e-05,
+      "loss": 0.5889,
+      "step": 11101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2913567581431135e-05,
+      "loss": 0.5673,
+      "step": 11102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291292639138241e-05,
+      "loss": 0.5314,
+      "step": 11103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291228520133368e-05,
+      "loss": 0.5969,
+      "step": 11104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291164401128494e-05,
+      "loss": 0.5127,
+      "step": 11105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2911002821236215e-05,
+      "loss": 0.5292,
+      "step": 11106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291036163118749e-05,
+      "loss": 0.4638,
+      "step": 11107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290972044113876e-05,
+      "loss": 0.5354,
+      "step": 11108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2909079251090023e-05,
+      "loss": 0.4726,
+      "step": 11109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2908438061041295e-05,
+      "loss": 0.5344,
+      "step": 11110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290779687099256e-05,
+      "loss": 0.6269,
+      "step": 11111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290715568094383e-05,
+      "loss": 0.5551,
+      "step": 11112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2906514490895104e-05,
+      "loss": 0.4903,
+      "step": 11113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2905873300846375e-05,
+      "loss": 0.5867,
+      "step": 11114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290523211079765e-05,
+      "loss": 0.5019,
+      "step": 11115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290459092074891e-05,
+      "loss": 0.6204,
+      "step": 11116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2903949730700184e-05,
+      "loss": 0.5753,
+      "step": 11117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290330854065145e-05,
+      "loss": 0.617,
+      "step": 11118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290266735060272e-05,
+      "loss": 0.556,
+      "step": 11119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2902026160553985e-05,
+      "loss": 0.6244,
+      "step": 11120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290138497050526e-05,
+      "loss": 0.6312,
+      "step": 11121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290074378045653e-05,
+      "loss": 0.5162,
+      "step": 11122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29001025904078e-05,
+      "loss": 0.5251,
+      "step": 11123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289946140035907e-05,
+      "loss": 0.6254,
+      "step": 11124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289882021031034e-05,
+      "loss": 0.5756,
+      "step": 11125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289817902026161e-05,
+      "loss": 0.5488,
+      "step": 11126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2897537830212874e-05,
+      "loss": 0.4786,
+      "step": 11127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2896896640164145e-05,
+      "loss": 0.4566,
+      "step": 11128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289625545011542e-05,
+      "loss": 0.6067,
+      "step": 11129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289561426006669e-05,
+      "loss": 0.5364,
+      "step": 11130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289497307001796e-05,
+      "loss": 0.505,
+      "step": 11131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2894331879969225e-05,
+      "loss": 0.5004,
+      "step": 11132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28936906899205e-05,
+      "loss": 0.5936,
+      "step": 11133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289304949987176e-05,
+      "loss": 0.5622,
+      "step": 11134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2892408309823034e-05,
+      "loss": 0.4841,
+      "step": 11135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28917671197743e-05,
+      "loss": 0.5355,
+      "step": 11136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289112592972557e-05,
+      "loss": 0.6184,
+      "step": 11137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289048473967684e-05,
+      "loss": 0.5304,
+      "step": 11138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2889843549628114e-05,
+      "loss": 0.4825,
+      "step": 11139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2889202359579386e-05,
+      "loss": 0.5995,
+      "step": 11140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288856116953065e-05,
+      "loss": 0.6034,
+      "step": 11141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288791997948192e-05,
+      "loss": 0.5195,
+      "step": 11142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288727878943319e-05,
+      "loss": 0.6459,
+      "step": 11143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288663759938446e-05,
+      "loss": 0.5798,
+      "step": 11144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2885996409335724e-05,
+      "loss": 0.5879,
+      "step": 11145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2885355219286996e-05,
+      "loss": 0.4671,
+      "step": 11146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2884714029238274e-05,
+      "loss": 0.5167,
+      "step": 11147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288407283918954e-05,
+      "loss": 0.5844,
+      "step": 11148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288343164914081e-05,
+      "loss": 0.594,
+      "step": 11149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2882790459092076e-05,
+      "loss": 0.5277,
+      "step": 11150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288214926904335e-05,
+      "loss": 0.466,
+      "step": 11151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288150807899461e-05,
+      "loss": 0.5534,
+      "step": 11152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2880866888945884e-05,
+      "loss": 0.5855,
+      "step": 11153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2880225698897156e-05,
+      "loss": 0.497,
+      "step": 11154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287958450884843e-05,
+      "loss": 0.5034,
+      "step": 11155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28789433187997e-05,
+      "loss": 0.6233,
+      "step": 11156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2878302128750964e-05,
+      "loss": 0.5586,
+      "step": 11157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2877660938702236e-05,
+      "loss": 0.6395,
+      "step": 11158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28770197486535e-05,
+      "loss": 0.6516,
+      "step": 11159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287637855860477e-05,
+      "loss": 0.53,
+      "step": 11160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287573736855604e-05,
+      "loss": 0.5122,
+      "step": 11161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287509617850731e-05,
+      "loss": 0.5685,
+      "step": 11162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287445498845858e-05,
+      "loss": 0.4702,
+      "step": 11163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287381379840985e-05,
+      "loss": 0.5564,
+      "step": 11164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2873172608361124e-05,
+      "loss": 0.5224,
+      "step": 11165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287253141831239e-05,
+      "loss": 0.4554,
+      "step": 11166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.287189022826366e-05,
+      "loss": 0.594,
+      "step": 11167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2871249038214926e-05,
+      "loss": 0.5927,
+      "step": 11168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28706078481662e-05,
+      "loss": 0.5357,
+      "step": 11169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286996665811747e-05,
+      "loss": 0.4918,
+      "step": 11170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286932546806874e-05,
+      "loss": 0.4914,
+      "step": 11171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2868684278020006e-05,
+      "loss": 0.6033,
+      "step": 11172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286804308797128e-05,
+      "loss": 0.5316,
+      "step": 11173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286740189792255e-05,
+      "loss": 0.581,
+      "step": 11174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2866760707873814e-05,
+      "loss": 0.4881,
+      "step": 11175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2866119517825086e-05,
+      "loss": 0.6143,
+      "step": 11176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286547832777635e-05,
+      "loss": 0.6842,
+      "step": 11177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286483713772762e-05,
+      "loss": 0.6155,
+      "step": 11178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2864195947678895e-05,
+      "loss": 0.4912,
+      "step": 11179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2863554757630166e-05,
+      "loss": 0.7158,
+      "step": 11180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286291356758143e-05,
+      "loss": 0.6348,
+      "step": 11181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28622723775327e-05,
+      "loss": 0.4188,
+      "step": 11182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2861631187483975e-05,
+      "loss": 0.5362,
+      "step": 11183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286098999743524e-05,
+      "loss": 0.5731,
+      "step": 11184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286034880738651e-05,
+      "loss": 0.5777,
+      "step": 11185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285970761733778e-05,
+      "loss": 0.5484,
+      "step": 11186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2859066427289055e-05,
+      "loss": 0.5769,
+      "step": 11187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285842523724032e-05,
+      "loss": 0.6036,
+      "step": 11188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285778404719159e-05,
+      "loss": 0.5224,
+      "step": 11189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.5294,
+      "step": 11190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285650166709413e-05,
+      "loss": 0.5747,
+      "step": 11191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28558604770454e-05,
+      "loss": 0.5925,
+      "step": 11192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2855219286996665e-05,
+      "loss": 0.5203,
+      "step": 11193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2854578096947936e-05,
+      "loss": 0.5146,
+      "step": 11194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285393690689921e-05,
+      "loss": 0.5188,
+      "step": 11195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285329571685048e-05,
+      "loss": 0.585,
+      "step": 11196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2852654526801745e-05,
+      "loss": 0.5075,
+      "step": 11197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2852013336753016e-05,
+      "loss": 0.548,
+      "step": 11198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285137214670428e-05,
+      "loss": 0.5475,
+      "step": 11199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285073095665555e-05,
+      "loss": 0.4887,
+      "step": 11200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2850089766606825e-05,
+      "loss": 0.4676,
+      "step": 11201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2849448576558097e-05,
+      "loss": 0.5351,
+      "step": 11202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284880738650937e-05,
+      "loss": 0.5457,
+      "step": 11203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284816619646063e-05,
+      "loss": 0.529,
+      "step": 11204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2847525006411905e-05,
+      "loss": 0.5363,
+      "step": 11205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284688381636317e-05,
+      "loss": 0.5455,
+      "step": 11206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284624262631444e-05,
+      "loss": 0.5937,
+      "step": 11207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2845601436265707e-05,
+      "loss": 0.7407,
+      "step": 11208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284496024621698e-05,
+      "loss": 0.6444,
+      "step": 11209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284431905616825e-05,
+      "loss": 0.5066,
+      "step": 11210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284367786611952e-05,
+      "loss": 0.5948,
+      "step": 11211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2843036676070793e-05,
+      "loss": 0.562,
+      "step": 11212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284239548602206e-05,
+      "loss": 0.6305,
+      "step": 11213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284175429597333e-05,
+      "loss": 0.5374,
+      "step": 11214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2841113105924595e-05,
+      "loss": 0.6195,
+      "step": 11215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284047191587587e-05,
+      "loss": 0.5279,
+      "step": 11216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283983072582713e-05,
+      "loss": 0.5282,
+      "step": 11217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283918953577841e-05,
+      "loss": 0.5627,
+      "step": 11218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283854834572968e-05,
+      "loss": 0.6113,
+      "step": 11219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283790715568095e-05,
+      "loss": 0.604,
+      "step": 11220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283726596563222e-05,
+      "loss": 0.6501,
+      "step": 11221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2836624775583483e-05,
+      "loss": 0.5595,
+      "step": 11222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2835983585534755e-05,
+      "loss": 0.6428,
+      "step": 11223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283534239548602e-05,
+      "loss": 0.5217,
+      "step": 11224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283470120543729e-05,
+      "loss": 0.5595,
+      "step": 11225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2834060015388564e-05,
+      "loss": 0.5832,
+      "step": 11226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2833418825339835e-05,
+      "loss": 0.4727,
+      "step": 11227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283277763529111e-05,
+      "loss": 0.569,
+      "step": 11228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283213644524237e-05,
+      "loss": 0.6637,
+      "step": 11229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2831495255193644e-05,
+      "loss": 0.5427,
+      "step": 11230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283085406514491e-05,
+      "loss": 0.5467,
+      "step": 11231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.283021287509618e-05,
+      "loss": 0.575,
+      "step": 11232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2829571685047445e-05,
+      "loss": 0.5855,
+      "step": 11233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282893049499872e-05,
+      "loss": 0.4777,
+      "step": 11234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282828930494999e-05,
+      "loss": 0.5991,
+      "step": 11235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282764811490126e-05,
+      "loss": 0.574,
+      "step": 11236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282700692485253e-05,
+      "loss": 0.6401,
+      "step": 11237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28263657348038e-05,
+      "loss": 0.5819,
+      "step": 11238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282572454475507e-05,
+      "loss": 0.5437,
+      "step": 11239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2825083354706334e-05,
+      "loss": 0.4904,
+      "step": 11240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2824442164657605e-05,
+      "loss": 0.6335,
+      "step": 11241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282380097460888e-05,
+      "loss": 0.5966,
+      "step": 11242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282315978456015e-05,
+      "loss": 0.6254,
+      "step": 11243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2822518594511414e-05,
+      "loss": 0.6087,
+      "step": 11244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2821877404462686e-05,
+      "loss": 0.5931,
+      "step": 11245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282123621441396e-05,
+      "loss": 0.5585,
+      "step": 11246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282059502436522e-05,
+      "loss": 0.6203,
+      "step": 11247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2819953834316494e-05,
+      "loss": 0.4781,
+      "step": 11248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281931264426776e-05,
+      "loss": 0.455,
+      "step": 11249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281867145421903e-05,
+      "loss": 0.6225,
+      "step": 11250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28180302641703e-05,
+      "loss": 0.5466,
+      "step": 11251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2817389074121574e-05,
+      "loss": 0.5672,
+      "step": 11252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281674788407284e-05,
+      "loss": 0.4969,
+      "step": 11253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281610669402411e-05,
+      "loss": 0.6054,
+      "step": 11254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281546550397538e-05,
+      "loss": 0.5643,
+      "step": 11255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281482431392665e-05,
+      "loss": 0.6434,
+      "step": 11256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281418312387792e-05,
+      "loss": 0.4948,
+      "step": 11257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281354193382919e-05,
+      "loss": 0.5762,
+      "step": 11258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281290074378046e-05,
+      "loss": 0.5849,
+      "step": 11259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281225955373173e-05,
+      "loss": 0.5353,
+      "step": 11260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2811618363683e-05,
+      "loss": 0.5824,
+      "step": 11261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2810977173634264e-05,
+      "loss": 0.5569,
+      "step": 11262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2810335983585536e-05,
+      "loss": 0.6416,
+      "step": 11263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280969479353681e-05,
+      "loss": 0.6957,
+      "step": 11264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280905360348807e-05,
+      "loss": 0.4966,
+      "step": 11265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2808412413439344e-05,
+      "loss": 0.4774,
+      "step": 11266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2807771223390616e-05,
+      "loss": 0.5691,
+      "step": 11267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280713003334189e-05,
+      "loss": 0.6111,
+      "step": 11268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280648884329315e-05,
+      "loss": 0.5371,
+      "step": 11269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2805847653244424e-05,
+      "loss": 0.4788,
+      "step": 11270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280520646319569e-05,
+      "loss": 0.5586,
+      "step": 11271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280456527314696e-05,
+      "loss": 0.6079,
+      "step": 11272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280392408309823e-05,
+      "loss": 0.5132,
+      "step": 11273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2803282893049504e-05,
+      "loss": 0.5018,
+      "step": 11274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2802641703000776e-05,
+      "loss": 0.5953,
+      "step": 11275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280200051295204e-05,
+      "loss": 0.5301,
+      "step": 11276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280135932290331e-05,
+      "loss": 0.5693,
+      "step": 11277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280071813285458e-05,
+      "loss": 0.5405,
+      "step": 11278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.280007694280585e-05,
+      "loss": 0.5386,
+      "step": 11279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2799435752757114e-05,
+      "loss": 0.569,
+      "step": 11280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2798794562708386e-05,
+      "loss": 0.6519,
+      "step": 11281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279815337265966e-05,
+      "loss": 0.5845,
+      "step": 11282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279751218261093e-05,
+      "loss": 0.5121,
+      "step": 11283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.27968709925622e-05,
+      "loss": 0.5542,
+      "step": 11284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2796229802513466e-05,
+      "loss": 0.5794,
+      "step": 11285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279558861246474e-05,
+      "loss": 0.6672,
+      "step": 11286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2794947422416e-05,
+      "loss": 0.5052,
+      "step": 11287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2794306232367274e-05,
+      "loss": 0.5979,
+      "step": 11288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279366504231854e-05,
+      "loss": 0.4896,
+      "step": 11289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279302385226982e-05,
+      "loss": 0.5891,
+      "step": 11290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279238266222109e-05,
+      "loss": 0.5888,
+      "step": 11291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2791741472172355e-05,
+      "loss": 0.5956,
+      "step": 11292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2791100282123626e-05,
+      "loss": 0.5527,
+      "step": 11293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.279045909207489e-05,
+      "loss": 0.5329,
+      "step": 11294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278981790202616e-05,
+      "loss": 0.564,
+      "step": 11295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278917671197743e-05,
+      "loss": 0.5376,
+      "step": 11296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.27885355219287e-05,
+      "loss": 0.5176,
+      "step": 11297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278789433187997e-05,
+      "loss": 0.6279,
+      "step": 11298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278725314183124e-05,
+      "loss": 0.6354,
+      "step": 11299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2786611951782515e-05,
+      "loss": 0.5443,
+      "step": 11300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278597076173378e-05,
+      "loss": 0.4738,
+      "step": 11301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278532957168505e-05,
+      "loss": 0.5253,
+      "step": 11302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2784688381636316e-05,
+      "loss": 0.5681,
+      "step": 11303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278404719158759e-05,
+      "loss": 0.5697,
+      "step": 11304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278340600153885e-05,
+      "loss": 0.5295,
+      "step": 11305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2782764811490125e-05,
+      "loss": 0.582,
+      "step": 11306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2782123621441396e-05,
+      "loss": 0.4803,
+      "step": 11307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278148243139267e-05,
+      "loss": 0.5312,
+      "step": 11308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278084124134394e-05,
+      "loss": 0.5483,
+      "step": 11309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2780200051295205e-05,
+      "loss": 0.5792,
+      "step": 11310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2779558861246477e-05,
+      "loss": 0.5223,
+      "step": 11311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.277891767119774e-05,
+      "loss": 0.5228,
+      "step": 11312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.277827648114901e-05,
+      "loss": 0.6864,
+      "step": 11313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2777635291100285e-05,
+      "loss": 0.5994,
+      "step": 11314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2776994101051557e-05,
+      "loss": 0.5511,
+      "step": 11315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277635291100283e-05,
+      "loss": 0.6041,
+      "step": 11316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277571172095409e-05,
+      "loss": 0.5759,
+      "step": 11317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2775070530905365e-05,
+      "loss": 0.5936,
+      "step": 11318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277442934085663e-05,
+      "loss": 0.5937,
+      "step": 11319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27737881508079e-05,
+      "loss": 0.6039,
+      "step": 11320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2773146960759167e-05,
+      "loss": 0.5569,
+      "step": 11321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277250577071044e-05,
+      "loss": 0.6375,
+      "step": 11322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277186458066171e-05,
+      "loss": 0.4749,
+      "step": 11323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277122339061298e-05,
+      "loss": 0.5635,
+      "step": 11324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2770582200564253e-05,
+      "loss": 0.5578,
+      "step": 11325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276994101051552e-05,
+      "loss": 0.596,
+      "step": 11326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276929982046679e-05,
+      "loss": 0.6014,
+      "step": 11327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2768658630418055e-05,
+      "loss": 0.5758,
+      "step": 11328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276801744036933e-05,
+      "loss": 0.6048,
+      "step": 11329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27673762503206e-05,
+      "loss": 0.5664,
+      "step": 11330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276673506027187e-05,
+      "loss": 0.4611,
+      "step": 11331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2766093870223135e-05,
+      "loss": 0.5072,
+      "step": 11332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276545268017441e-05,
+      "loss": 0.5889,
+      "step": 11333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276481149012568e-05,
+      "loss": 0.5743,
+      "step": 11334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2764170300076943e-05,
+      "loss": 0.6372,
+      "step": 11335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2763529110028215e-05,
+      "loss": 0.5925,
+      "step": 11336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276288791997948e-05,
+      "loss": 0.5643,
+      "step": 11337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276224672993075e-05,
+      "loss": 0.4814,
+      "step": 11338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2761605539882024e-05,
+      "loss": 0.5663,
+      "step": 11339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2760964349833295e-05,
+      "loss": 0.5952,
+      "step": 11340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276032315978456e-05,
+      "loss": 0.589,
+      "step": 11341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275968196973583e-05,
+      "loss": 0.5065,
+      "step": 11342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2759040779687104e-05,
+      "loss": 0.6021,
+      "step": 11343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275839958963837e-05,
+      "loss": 0.5699,
+      "step": 11344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275775839958964e-05,
+      "loss": 0.5276,
+      "step": 11345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275711720954091e-05,
+      "loss": 0.5665,
+      "step": 11346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2756476019492184e-05,
+      "loss": 0.5632,
+      "step": 11347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275583482944345e-05,
+      "loss": 0.5265,
+      "step": 11348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275519363939472e-05,
+      "loss": 0.5902,
+      "step": 11349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2754552449345985e-05,
+      "loss": 0.5701,
+      "step": 11350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275391125929726e-05,
+      "loss": 0.5928,
+      "step": 11351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275327006924853e-05,
+      "loss": 0.539,
+      "step": 11352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2752628879199794e-05,
+      "loss": 0.6312,
+      "step": 11353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2751987689151065e-05,
+      "loss": 0.5332,
+      "step": 11354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275134649910234e-05,
+      "loss": 0.5392,
+      "step": 11355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275070530905361e-05,
+      "loss": 0.5779,
+      "step": 11356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2750064119004874e-05,
+      "loss": 0.6987,
+      "step": 11357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2749422928956146e-05,
+      "loss": 0.4968,
+      "step": 11358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274878173890741e-05,
+      "loss": 0.5755,
+      "step": 11359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274814054885868e-05,
+      "loss": 0.5835,
+      "step": 11360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2747499358809954e-05,
+      "loss": 0.5987,
+      "step": 11361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2746858168761226e-05,
+      "loss": 0.4533,
+      "step": 11362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27462169787125e-05,
+      "loss": 0.6412,
+      "step": 11363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274557578866376e-05,
+      "loss": 0.5339,
+      "step": 11364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2744934598615034e-05,
+      "loss": 0.5425,
+      "step": 11365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27442934085663e-05,
+      "loss": 0.5972,
+      "step": 11366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274365221851757e-05,
+      "loss": 0.5525,
+      "step": 11367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2743011028468836e-05,
+      "loss": 0.5526,
+      "step": 11368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274236983842011e-05,
+      "loss": 0.5649,
+      "step": 11369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274172864837138e-05,
+      "loss": 0.5552,
+      "step": 11370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274108745832265e-05,
+      "loss": 0.5566,
+      "step": 11371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274044626827392e-05,
+      "loss": 0.526,
+      "step": 11372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273980507822519e-05,
+      "loss": 0.5253,
+      "step": 11373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273916388817646e-05,
+      "loss": 0.6633,
+      "step": 11374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2738522698127724e-05,
+      "loss": 0.6506,
+      "step": 11375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2737881508078996e-05,
+      "loss": 0.5733,
+      "step": 11376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273724031803026e-05,
+      "loss": 0.5197,
+      "step": 11377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273659912798154e-05,
+      "loss": 0.6548,
+      "step": 11378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273595793793281e-05,
+      "loss": 0.6122,
+      "step": 11379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2735316747884076e-05,
+      "loss": 0.6094,
+      "step": 11380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273467555783535e-05,
+      "loss": 0.6938,
+      "step": 11381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273403436778661e-05,
+      "loss": 0.5638,
+      "step": 11382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2733393177737884e-05,
+      "loss": 0.5944,
+      "step": 11383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273275198768915e-05,
+      "loss": 0.501,
+      "step": 11384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273211079764042e-05,
+      "loss": 0.5469,
+      "step": 11385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273146960759169e-05,
+      "loss": 0.5736,
+      "step": 11386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2730828417542964e-05,
+      "loss": 0.5883,
+      "step": 11387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2730187227494236e-05,
+      "loss": 0.52,
+      "step": 11388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27295460374455e-05,
+      "loss": 0.5344,
+      "step": 11389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272890484739677e-05,
+      "loss": 0.5847,
+      "step": 11390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272826365734804e-05,
+      "loss": 0.6458,
+      "step": 11391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272762246729931e-05,
+      "loss": 0.475,
+      "step": 11392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2726981277250574e-05,
+      "loss": 0.5855,
+      "step": 11393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2726340087201846e-05,
+      "loss": 0.5574,
+      "step": 11394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272569889715312e-05,
+      "loss": 0.5058,
+      "step": 11395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272505770710439e-05,
+      "loss": 0.5836,
+      "step": 11396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272441651705566e-05,
+      "loss": 0.5023,
+      "step": 11397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2723775327006926e-05,
+      "loss": 0.4969,
+      "step": 11398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27231341369582e-05,
+      "loss": 0.5636,
+      "step": 11399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272249294690946e-05,
+      "loss": 0.5105,
+      "step": 11400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2721851756860734e-05,
+      "loss": 0.5087,
+      "step": 11401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2721210566812006e-05,
+      "loss": 0.5418,
+      "step": 11402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272056937676328e-05,
+      "loss": 0.6475,
+      "step": 11403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271992818671454e-05,
+      "loss": 0.5671,
+      "step": 11404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2719286996665815e-05,
+      "loss": 0.5175,
+      "step": 11405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2718645806617086e-05,
+      "loss": 0.5743,
+      "step": 11406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271800461656835e-05,
+      "loss": 0.5975,
+      "step": 11407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271736342651962e-05,
+      "loss": 0.5599,
+      "step": 11408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271672223647089e-05,
+      "loss": 0.5838,
+      "step": 11409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271608104642216e-05,
+      "loss": 0.5595,
+      "step": 11410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271543985637343e-05,
+      "loss": 0.5926,
+      "step": 11411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27147986663247e-05,
+      "loss": 0.5086,
+      "step": 11412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271415747627597e-05,
+      "loss": 0.5113,
+      "step": 11413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271351628622724e-05,
+      "loss": 0.5933,
+      "step": 11414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271287509617851e-05,
+      "loss": 0.5607,
+      "step": 11415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2712233906129776e-05,
+      "loss": 0.7216,
+      "step": 11416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271159271608105e-05,
+      "loss": 0.565,
+      "step": 11417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271095152603232e-05,
+      "loss": 0.4995,
+      "step": 11418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271031033598359e-05,
+      "loss": 0.6665,
+      "step": 11419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2709669145934856e-05,
+      "loss": 0.5423,
+      "step": 11420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270902795588613e-05,
+      "loss": 0.6656,
+      "step": 11421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270838676583739e-05,
+      "loss": 0.5293,
+      "step": 11422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2707745575788665e-05,
+      "loss": 0.6178,
+      "step": 11423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2707104385739937e-05,
+      "loss": 0.4715,
+      "step": 11424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27064631956912e-05,
+      "loss": 0.5467,
+      "step": 11425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270582200564247e-05,
+      "loss": 0.6648,
+      "step": 11426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2705180815593745e-05,
+      "loss": 0.5012,
+      "step": 11427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270453962554502e-05,
+      "loss": 0.5787,
+      "step": 11428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270389843549628e-05,
+      "loss": 0.5983,
+      "step": 11429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270325724544755e-05,
+      "loss": 0.598,
+      "step": 11430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270261605539882e-05,
+      "loss": 0.6179,
+      "step": 11431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270197486535009e-05,
+      "loss": 0.5475,
+      "step": 11432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270133367530136e-05,
+      "loss": 0.6045,
+      "step": 11433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270069248525263e-05,
+      "loss": 0.6356,
+      "step": 11434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2700051295203905e-05,
+      "loss": 0.5504,
+      "step": 11435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269941010515517e-05,
+      "loss": 0.6235,
+      "step": 11436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269876891510644e-05,
+      "loss": 0.5221,
+      "step": 11437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269812772505771e-05,
+      "loss": 0.6104,
+      "step": 11438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269748653500898e-05,
+      "loss": 0.6464,
+      "step": 11439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269684534496024e-05,
+      "loss": 0.5597,
+      "step": 11440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2696204154911515e-05,
+      "loss": 0.6405,
+      "step": 11441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269556296486279e-05,
+      "loss": 0.5348,
+      "step": 11442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269492177481406e-05,
+      "loss": 0.5663,
+      "step": 11443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269428058476533e-05,
+      "loss": 0.613,
+      "step": 11444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2693639394716595e-05,
+      "loss": 0.546,
+      "step": 11445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269299820466787e-05,
+      "loss": 0.6069,
+      "step": 11446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269235701461913e-05,
+      "loss": 0.5524,
+      "step": 11447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2691715824570404e-05,
+      "loss": 0.5194,
+      "step": 11448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269107463452167e-05,
+      "loss": 0.5705,
+      "step": 11449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269043344447295e-05,
+      "loss": 0.5838,
+      "step": 11450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268979225442422e-05,
+      "loss": 0.4795,
+      "step": 11451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2689151064375484e-05,
+      "loss": 0.5263,
+      "step": 11452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2688509874326755e-05,
+      "loss": 0.5161,
+      "step": 11453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268786868427802e-05,
+      "loss": 0.6349,
+      "step": 11454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268722749422929e-05,
+      "loss": 0.5356,
+      "step": 11455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268658630418056e-05,
+      "loss": 0.5782,
+      "step": 11456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268594511413183e-05,
+      "loss": 0.4503,
+      "step": 11457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26853039240831e-05,
+      "loss": 0.4366,
+      "step": 11458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268466273403437e-05,
+      "loss": 0.6333,
+      "step": 11459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2684021543985644e-05,
+      "loss": 0.5288,
+      "step": 11460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268338035393691e-05,
+      "loss": 0.6327,
+      "step": 11461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268273916388818e-05,
+      "loss": 0.557,
+      "step": 11462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2682097973839445e-05,
+      "loss": 0.5903,
+      "step": 11463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268145678379072e-05,
+      "loss": 0.5534,
+      "step": 11464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268081559374198e-05,
+      "loss": 0.5439,
+      "step": 11465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2680174403693254e-05,
+      "loss": 0.6316,
+      "step": 11466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2679533213644525e-05,
+      "loss": 0.5283,
+      "step": 11467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26788920235958e-05,
+      "loss": 0.5559,
+      "step": 11468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.267825083354707e-05,
+      "loss": 0.5368,
+      "step": 11469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2677609643498334e-05,
+      "loss": 0.5298,
+      "step": 11470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2676968453449606e-05,
+      "loss": 0.6417,
+      "step": 11471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.267632726340087e-05,
+      "loss": 0.5825,
+      "step": 11472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.267568607335214e-05,
+      "loss": 0.5353,
+      "step": 11473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2675044883303414e-05,
+      "loss": 0.6284,
+      "step": 11474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2674403693254686e-05,
+      "loss": 0.4793,
+      "step": 11475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.267376250320595e-05,
+      "loss": 0.5158,
+      "step": 11476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.267312131315722e-05,
+      "loss": 0.5608,
+      "step": 11477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2672480123108494e-05,
+      "loss": 0.5836,
+      "step": 11478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.267183893305976e-05,
+      "loss": 0.4513,
+      "step": 11479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.267119774301103e-05,
+      "loss": 0.5008,
+      "step": 11480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2670556552962296e-05,
+      "loss": 0.4825,
+      "step": 11481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266991536291357e-05,
+      "loss": 0.5625,
+      "step": 11482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266927417286484e-05,
+      "loss": 0.4977,
+      "step": 11483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266863298281611e-05,
+      "loss": 0.5223,
+      "step": 11484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2667991792767376e-05,
+      "loss": 0.4534,
+      "step": 11485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266735060271865e-05,
+      "loss": 0.5603,
+      "step": 11486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266670941266992e-05,
+      "loss": 0.5217,
+      "step": 11487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2666068222621184e-05,
+      "loss": 0.578,
+      "step": 11488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2665427032572456e-05,
+      "loss": 0.6068,
+      "step": 11489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266478584252373e-05,
+      "loss": 0.6068,
+      "step": 11490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2664144652475e-05,
+      "loss": 0.4961,
+      "step": 11491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2663503462426264e-05,
+      "loss": 0.5802,
+      "step": 11492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2662862272377536e-05,
+      "loss": 0.539,
+      "step": 11493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26622210823288e-05,
+      "loss": 0.6264,
+      "step": 11494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266157989228007e-05,
+      "loss": 0.5757,
+      "step": 11495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2660938702231344e-05,
+      "loss": 0.5255,
+      "step": 11496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.266029751218261e-05,
+      "loss": 0.5484,
+      "step": 11497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265965632213388e-05,
+      "loss": 0.4485,
+      "step": 11498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265901513208515e-05,
+      "loss": 0.5017,
+      "step": 11499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2658373942036424e-05,
+      "loss": 0.5103,
+      "step": 11500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265773275198769e-05,
+      "loss": 0.5458,
+      "step": 11501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265709156193896e-05,
+      "loss": 0.5194,
+      "step": 11502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265645037189023e-05,
+      "loss": 0.6201,
+      "step": 11503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26558091818415e-05,
+      "loss": 0.5958,
+      "step": 11504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265516799179277e-05,
+      "loss": 0.4649,
+      "step": 11505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265452680174404e-05,
+      "loss": 0.5735,
+      "step": 11506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265388561169531e-05,
+      "loss": 0.5469,
+      "step": 11507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265324442164658e-05,
+      "loss": 0.4864,
+      "step": 11508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265260323159785e-05,
+      "loss": 0.6713,
+      "step": 11509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2651962041549114e-05,
+      "loss": 0.556,
+      "step": 11510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2651320851500386e-05,
+      "loss": 0.4865,
+      "step": 11511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265067966145166e-05,
+      "loss": 0.5554,
+      "step": 11512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265003847140292e-05,
+      "loss": 0.646,
+      "step": 11513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2649397281354195e-05,
+      "loss": 0.5336,
+      "step": 11514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2648756091305466e-05,
+      "loss": 0.5889,
+      "step": 11515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264811490125674e-05,
+      "loss": 0.5486,
+      "step": 11516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2647473711208e-05,
+      "loss": 0.5828,
+      "step": 11517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2646832521159275e-05,
+      "loss": 0.5101,
+      "step": 11518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264619133111054e-05,
+      "loss": 0.6087,
+      "step": 11519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264555014106181e-05,
+      "loss": 0.4132,
+      "step": 11520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264490895101308e-05,
+      "loss": 0.5472,
+      "step": 11521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2644267760964355e-05,
+      "loss": 0.5719,
+      "step": 11522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2643626570915626e-05,
+      "loss": 0.5141,
+      "step": 11523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264298538086689e-05,
+      "loss": 0.5558,
+      "step": 11524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264234419081816e-05,
+      "loss": 0.6272,
+      "step": 11525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264170300076943e-05,
+      "loss": 0.4997,
+      "step": 11526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26410618107207e-05,
+      "loss": 0.6146,
+      "step": 11527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2640420620671965e-05,
+      "loss": 0.5363,
+      "step": 11528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2639779430623236e-05,
+      "loss": 0.5523,
+      "step": 11529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263913824057451e-05,
+      "loss": 0.5647,
+      "step": 11530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263849705052578e-05,
+      "loss": 0.5604,
+      "step": 11531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263785586047705e-05,
+      "loss": 0.5462,
+      "step": 11532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2637214670428316e-05,
+      "loss": 0.6006,
+      "step": 11533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263657348037959e-05,
+      "loss": 0.4606,
+      "step": 11534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263593229033085e-05,
+      "loss": 0.5277,
+      "step": 11535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2635291100282125e-05,
+      "loss": 0.6486,
+      "step": 11536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263464991023339e-05,
+      "loss": 0.5984,
+      "step": 11537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263400872018467e-05,
+      "loss": 0.5016,
+      "step": 11538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263336753013594e-05,
+      "loss": 0.5235,
+      "step": 11539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2632726340087205e-05,
+      "loss": 0.587,
+      "step": 11540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263208515003848e-05,
+      "loss": 0.6774,
+      "step": 11541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263144395998974e-05,
+      "loss": 0.5156,
+      "step": 11542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263080276994101e-05,
+      "loss": 0.5559,
+      "step": 11543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263016157989228e-05,
+      "loss": 0.5741,
+      "step": 11544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262952038984355e-05,
+      "loss": 0.5959,
+      "step": 11545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262887919979482e-05,
+      "loss": 0.4934,
+      "step": 11546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2628238009746093e-05,
+      "loss": 0.6041,
+      "step": 11547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2627596819697365e-05,
+      "loss": 0.5204,
+      "step": 11548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262695562964863e-05,
+      "loss": 0.5162,
+      "step": 11549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26263144395999e-05,
+      "loss": 0.5722,
+      "step": 11550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262567324955117e-05,
+      "loss": 0.5395,
+      "step": 11551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262503205950244e-05,
+      "loss": 0.6482,
+      "step": 11552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26243908694537e-05,
+      "loss": 0.5668,
+      "step": 11553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2623749679404975e-05,
+      "loss": 0.5611,
+      "step": 11554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262310848935625e-05,
+      "loss": 0.4727,
+      "step": 11555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262246729930752e-05,
+      "loss": 0.5924,
+      "step": 11556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262182610925879e-05,
+      "loss": 0.619,
+      "step": 11557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2621184919210055e-05,
+      "loss": 0.6,
+      "step": 11558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262054372916133e-05,
+      "loss": 0.5667,
+      "step": 11559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261990253911259e-05,
+      "loss": 0.5288,
+      "step": 11560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2619261349063864e-05,
+      "loss": 0.553,
+      "step": 11561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2618620159015135e-05,
+      "loss": 0.5238,
+      "step": 11562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261797896896641e-05,
+      "loss": 0.5495,
+      "step": 11563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261733777891767e-05,
+      "loss": 0.5076,
+      "step": 11564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2616696588868944e-05,
+      "loss": 0.5239,
+      "step": 11565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2616055398820215e-05,
+      "loss": 0.538,
+      "step": 11566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261541420877148e-05,
+      "loss": 0.5485,
+      "step": 11567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261477301872275e-05,
+      "loss": 0.6006,
+      "step": 11568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261413182867402e-05,
+      "loss": 0.5853,
+      "step": 11569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261349063862529e-05,
+      "loss": 0.5803,
+      "step": 11570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261284944857656e-05,
+      "loss": 0.6374,
+      "step": 11571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261220825852783e-05,
+      "loss": 0.5659,
+      "step": 11572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26115670684791e-05,
+      "loss": 0.4802,
+      "step": 11573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.261092587843037e-05,
+      "loss": 0.5668,
+      "step": 11574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.261028468838164e-05,
+      "loss": 0.5816,
+      "step": 11575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2609643498332905e-05,
+      "loss": 0.5038,
+      "step": 11576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260900230828418e-05,
+      "loss": 0.5774,
+      "step": 11577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260836111823545e-05,
+      "loss": 0.6074,
+      "step": 11578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260771992818672e-05,
+      "loss": 0.5301,
+      "step": 11579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2607078738137986e-05,
+      "loss": 0.5942,
+      "step": 11580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260643754808926e-05,
+      "loss": 0.6232,
+      "step": 11581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260579635804052e-05,
+      "loss": 0.558,
+      "step": 11582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2605155167991794e-05,
+      "loss": 0.5925,
+      "step": 11583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2604513977943066e-05,
+      "loss": 0.5599,
+      "step": 11584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260387278789433e-05,
+      "loss": 0.577,
+      "step": 11585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.26032315978456e-05,
+      "loss": 0.681,
+      "step": 11586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2602590407796874e-05,
+      "loss": 0.6119,
+      "step": 11587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2601949217748146e-05,
+      "loss": 0.6478,
+      "step": 11588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260130802769941e-05,
+      "loss": 0.5063,
+      "step": 11589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260066683765068e-05,
+      "loss": 0.5399,
+      "step": 11590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260002564760195e-05,
+      "loss": 0.5959,
+      "step": 11591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259938445755322e-05,
+      "loss": 0.5795,
+      "step": 11592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259874326750449e-05,
+      "loss": 0.5483,
+      "step": 11593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259810207745576e-05,
+      "loss": 0.4664,
+      "step": 11594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2597460887407034e-05,
+      "loss": 0.546,
+      "step": 11595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25968196973583e-05,
+      "loss": 0.4798,
+      "step": 11596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259617850730957e-05,
+      "loss": 0.4918,
+      "step": 11597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2595537317260836e-05,
+      "loss": 0.653,
+      "step": 11598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259489612721211e-05,
+      "loss": 0.5093,
+      "step": 11599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259425493716337e-05,
+      "loss": 0.6154,
+      "step": 11600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2593613747114644e-05,
+      "loss": 0.5482,
+      "step": 11601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2592972557065916e-05,
+      "loss": 0.5536,
+      "step": 11602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259233136701719e-05,
+      "loss": 0.6764,
+      "step": 11603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259169017696846e-05,
+      "loss": 0.4999,
+      "step": 11604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2591048986919724e-05,
+      "loss": 0.5123,
+      "step": 11605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2590407796870996e-05,
+      "loss": 0.6017,
+      "step": 11606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258976660682226e-05,
+      "loss": 0.5257,
+      "step": 11607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258912541677353e-05,
+      "loss": 0.5793,
+      "step": 11608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25884842267248e-05,
+      "loss": 0.4958,
+      "step": 11609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2587843036676076e-05,
+      "loss": 0.6964,
+      "step": 11610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258720184662735e-05,
+      "loss": 0.5854,
+      "step": 11611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258656065657861e-05,
+      "loss": 0.5021,
+      "step": 11612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2585919466529884e-05,
+      "loss": 0.5594,
+      "step": 11613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258527827648115e-05,
+      "loss": 0.5729,
+      "step": 11614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258463708643242e-05,
+      "loss": 0.6048,
+      "step": 11615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2583995896383686e-05,
+      "loss": 0.478,
+      "step": 11616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258335470633496e-05,
+      "loss": 0.5952,
+      "step": 11617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258271351628623e-05,
+      "loss": 0.5209,
+      "step": 11618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25820723262375e-05,
+      "loss": 0.5596,
+      "step": 11619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258143113618877e-05,
+      "loss": 0.5146,
+      "step": 11620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258078994614004e-05,
+      "loss": 0.4271,
+      "step": 11621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258014875609131e-05,
+      "loss": 0.474,
+      "step": 11622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2579507566042574e-05,
+      "loss": 0.5482,
+      "step": 11623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2578866375993846e-05,
+      "loss": 0.5966,
+      "step": 11624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257822518594511e-05,
+      "loss": 0.613,
+      "step": 11625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257758399589639e-05,
+      "loss": 0.4818,
+      "step": 11626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2576942805847655e-05,
+      "loss": 0.4976,
+      "step": 11627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2576301615798926e-05,
+      "loss": 0.594,
+      "step": 11628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25756604257502e-05,
+      "loss": 0.5579,
+      "step": 11629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257501923570146e-05,
+      "loss": 0.6035,
+      "step": 11630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2574378045652735e-05,
+      "loss": 0.5509,
+      "step": 11631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2573736855604e-05,
+      "loss": 0.5089,
+      "step": 11632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257309566555527e-05,
+      "loss": 0.5621,
+      "step": 11633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257245447550654e-05,
+      "loss": 0.5019,
+      "step": 11634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2571813285457815e-05,
+      "loss": 0.6193,
+      "step": 11635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257117209540908e-05,
+      "loss": 0.58,
+      "step": 11636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257053090536035e-05,
+      "loss": 0.593,
+      "step": 11637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256988971531162e-05,
+      "loss": 0.4517,
+      "step": 11638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256924852526289e-05,
+      "loss": 0.6475,
+      "step": 11639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256860733521416e-05,
+      "loss": 0.5312,
+      "step": 11640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2567966145165425e-05,
+      "loss": 0.4827,
+      "step": 11641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2567324955116696e-05,
+      "loss": 0.4806,
+      "step": 11642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256668376506797e-05,
+      "loss": 0.5624,
+      "step": 11643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256604257501924e-05,
+      "loss": 0.5886,
+      "step": 11644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2565401384970505e-05,
+      "loss": 0.5401,
+      "step": 11645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2564760194921777e-05,
+      "loss": 0.5523,
+      "step": 11646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256411900487305e-05,
+      "loss": 0.4712,
+      "step": 11647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256347781482431e-05,
+      "loss": 0.4808,
+      "step": 11648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2562836624775585e-05,
+      "loss": 0.477,
+      "step": 11649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2562195434726857e-05,
+      "loss": 0.5547,
+      "step": 11650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256155424467813e-05,
+      "loss": 0.6424,
+      "step": 11651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256091305462939e-05,
+      "loss": 0.5377,
+      "step": 11652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2560271864580665e-05,
+      "loss": 0.5521,
+      "step": 11653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255963067453193e-05,
+      "loss": 0.5946,
+      "step": 11654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25589894844832e-05,
+      "loss": 0.5657,
+      "step": 11655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255834829443447e-05,
+      "loss": 0.4849,
+      "step": 11656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255770710438574e-05,
+      "loss": 0.603,
+      "step": 11657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255706591433701e-05,
+      "loss": 0.667,
+      "step": 11658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255642472428828e-05,
+      "loss": 0.5223,
+      "step": 11659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2555783534239553e-05,
+      "loss": 0.5169,
+      "step": 11660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255514234419082e-05,
+      "loss": 0.6238,
+      "step": 11661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255450115414209e-05,
+      "loss": 0.519,
+      "step": 11662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2553859964093355e-05,
+      "loss": 0.5907,
+      "step": 11663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255321877404463e-05,
+      "loss": 0.7038,
+      "step": 11664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25525775839959e-05,
+      "loss": 0.5747,
+      "step": 11665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255193639394717e-05,
+      "loss": 0.4685,
+      "step": 11666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255129520389844e-05,
+      "loss": 0.5453,
+      "step": 11667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255065401384971e-05,
+      "loss": 0.4432,
+      "step": 11668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255001282380098e-05,
+      "loss": 0.4656,
+      "step": 11669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2549371633752244e-05,
+      "loss": 0.5801,
+      "step": 11670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2548730443703515e-05,
+      "loss": 0.5013,
+      "step": 11671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254808925365478e-05,
+      "loss": 0.5214,
+      "step": 11672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254744806360605e-05,
+      "loss": 0.563,
+      "step": 11673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2546806873557324e-05,
+      "loss": 0.5561,
+      "step": 11674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2546165683508595e-05,
+      "loss": 0.5252,
+      "step": 11675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254552449345987e-05,
+      "loss": 0.5208,
+      "step": 11676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254488330341113e-05,
+      "loss": 0.5233,
+      "step": 11677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2544242113362404e-05,
+      "loss": 0.5205,
+      "step": 11678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254360092331367e-05,
+      "loss": 0.5646,
+      "step": 11679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254295973326494e-05,
+      "loss": 0.6785,
+      "step": 11680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2542318543216205e-05,
+      "loss": 0.6117,
+      "step": 11681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2541677353167484e-05,
+      "loss": 0.6267,
+      "step": 11682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2541036163118755e-05,
+      "loss": 0.5719,
+      "step": 11683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254039497307002e-05,
+      "loss": 0.5054,
+      "step": 11684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253975378302129e-05,
+      "loss": 0.4985,
+      "step": 11685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253911259297256e-05,
+      "loss": 0.5191,
+      "step": 11686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253847140292383e-05,
+      "loss": 0.4805,
+      "step": 11687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2537830212875094e-05,
+      "loss": 0.6382,
+      "step": 11688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2537189022826365e-05,
+      "loss": 0.5338,
+      "step": 11689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253654783277764e-05,
+      "loss": 0.6392,
+      "step": 11690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253590664272891e-05,
+      "loss": 0.5475,
+      "step": 11691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253526545268018e-05,
+      "loss": 0.5163,
+      "step": 11692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2534624262631446e-05,
+      "loss": 0.5378,
+      "step": 11693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253398307258272e-05,
+      "loss": 0.5251,
+      "step": 11694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253334188253398e-05,
+      "loss": 0.5135,
+      "step": 11695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2532700692485254e-05,
+      "loss": 0.6067,
+      "step": 11696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253205950243652e-05,
+      "loss": 0.5092,
+      "step": 11697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25314183123878e-05,
+      "loss": 0.5678,
+      "step": 11698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253077712233907e-05,
+      "loss": 0.582,
+      "step": 11699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2530135932290334e-05,
+      "loss": 0.6103,
+      "step": 11700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2529494742241606e-05,
+      "loss": 0.5782,
+      "step": 11701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252885355219287e-05,
+      "loss": 0.5426,
+      "step": 11702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252821236214414e-05,
+      "loss": 0.5681,
+      "step": 11703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252757117209541e-05,
+      "loss": 0.517,
+      "step": 11704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252692998204668e-05,
+      "loss": 0.5601,
+      "step": 11705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252628879199795e-05,
+      "loss": 0.654,
+      "step": 11706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252564760194922e-05,
+      "loss": 0.5722,
+      "step": 11707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2525006411900494e-05,
+      "loss": 0.454,
+      "step": 11708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252436522185176e-05,
+      "loss": 0.5828,
+      "step": 11709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252372403180303e-05,
+      "loss": 0.5459,
+      "step": 11710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2523082841754296e-05,
+      "loss": 0.5077,
+      "step": 11711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252244165170557e-05,
+      "loss": 0.5684,
+      "step": 11712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252180046165683e-05,
+      "loss": 0.5644,
+      "step": 11713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2521159271608104e-05,
+      "loss": 0.4768,
+      "step": 11714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2520518081559376e-05,
+      "loss": 0.5259,
+      "step": 11715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251987689151065e-05,
+      "loss": 0.4632,
+      "step": 11716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251923570146192e-05,
+      "loss": 0.5708,
+      "step": 11717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2518594511413184e-05,
+      "loss": 0.4982,
+      "step": 11718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2517953321364456e-05,
+      "loss": 0.6057,
+      "step": 11719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251731213131572e-05,
+      "loss": 0.5652,
+      "step": 11720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251667094126699e-05,
+      "loss": 0.5125,
+      "step": 11721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2516029751218264e-05,
+      "loss": 0.5702,
+      "step": 11722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2515388561169536e-05,
+      "loss": 0.5242,
+      "step": 11723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25147473711208e-05,
+      "loss": 0.5487,
+      "step": 11724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251410618107207e-05,
+      "loss": 0.4265,
+      "step": 11725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2513464991023344e-05,
+      "loss": 0.5626,
+      "step": 11726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251282380097461e-05,
+      "loss": 0.6161,
+      "step": 11727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251218261092588e-05,
+      "loss": 0.6168,
+      "step": 11728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2511541420877146e-05,
+      "loss": 0.4906,
+      "step": 11729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251090023082842e-05,
+      "loss": 0.5288,
+      "step": 11730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251025904077969e-05,
+      "loss": 0.5791,
+      "step": 11731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.250961785073096e-05,
+      "loss": 0.4185,
+      "step": 11732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2508976660682226e-05,
+      "loss": 0.5267,
+      "step": 11733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25083354706335e-05,
+      "loss": 0.5289,
+      "step": 11734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.250769428058477e-05,
+      "loss": 0.5954,
+      "step": 11735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2507053090536034e-05,
+      "loss": 0.6151,
+      "step": 11736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2506411900487306e-05,
+      "loss": 0.6293,
+      "step": 11737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.250577071043858e-05,
+      "loss": 0.665,
+      "step": 11738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.250512952038985e-05,
+      "loss": 0.454,
+      "step": 11739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2504488330341115e-05,
+      "loss": 0.524,
+      "step": 11740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2503847140292386e-05,
+      "loss": 0.6948,
+      "step": 11741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.250320595024365e-05,
+      "loss": 0.6174,
+      "step": 11742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.250256476019492e-05,
+      "loss": 0.6199,
+      "step": 11743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2501923570146195e-05,
+      "loss": 0.4088,
+      "step": 11744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.250128238009746e-05,
+      "loss": 0.5397,
+      "step": 11745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.250064119004873e-05,
+      "loss": 0.5611,
+      "step": 11746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25e-05,
+      "loss": 0.5361,
+      "step": 11747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2499358809951275e-05,
+      "loss": 0.4836,
+      "step": 11748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249871761990254e-05,
+      "loss": 0.5391,
+      "step": 11749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249807642985381e-05,
+      "loss": 0.6718,
+      "step": 11750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2497435239805076e-05,
+      "loss": 0.6121,
+      "step": 11751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249679404975635e-05,
+      "loss": 0.5303,
+      "step": 11752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249615285970762e-05,
+      "loss": 0.5455,
+      "step": 11753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249551166965889e-05,
+      "loss": 0.5736,
+      "step": 11754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249487047961016e-05,
+      "loss": 0.5105,
+      "step": 11755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249422928956143e-05,
+      "loss": 0.5668,
+      "step": 11756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24935880995127e-05,
+      "loss": 0.5375,
+      "step": 11757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2492946909463965e-05,
+      "loss": 0.4993,
+      "step": 11758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2492305719415237e-05,
+      "loss": 0.5617,
+      "step": 11759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24916645293665e-05,
+      "loss": 0.5086,
+      "step": 11760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249102333931777e-05,
+      "loss": 0.5673,
+      "step": 11761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2490382149269045e-05,
+      "loss": 0.5093,
+      "step": 11762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248974095922032e-05,
+      "loss": 0.6268,
+      "step": 11763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248909976917159e-05,
+      "loss": 0.4927,
+      "step": 11764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248845857912285e-05,
+      "loss": 0.6672,
+      "step": 11765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2487817389074125e-05,
+      "loss": 0.4673,
+      "step": 11766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248717619902539e-05,
+      "loss": 0.5272,
+      "step": 11767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248653500897666e-05,
+      "loss": 0.542,
+      "step": 11768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2485893818927927e-05,
+      "loss": 0.5375,
+      "step": 11769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2485252628879205e-05,
+      "loss": 0.5125,
+      "step": 11770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248461143883048e-05,
+      "loss": 0.4654,
+      "step": 11771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248397024878174e-05,
+      "loss": 0.5798,
+      "step": 11772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2483329058733013e-05,
+      "loss": 0.4871,
+      "step": 11773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248268786868428e-05,
+      "loss": 0.5029,
+      "step": 11774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248204667863555e-05,
+      "loss": 0.5427,
+      "step": 11775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2481405488586815e-05,
+      "loss": 0.6867,
+      "step": 11776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248076429853809e-05,
+      "loss": 0.5394,
+      "step": 11777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248012310848936e-05,
+      "loss": 0.5691,
+      "step": 11778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.247948191844063e-05,
+      "loss": 0.5186,
+      "step": 11779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24788407283919e-05,
+      "loss": 0.6305,
+      "step": 11780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.247819953834317e-05,
+      "loss": 0.614,
+      "step": 11781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.247755834829444e-05,
+      "loss": 0.45,
+      "step": 11782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2476917158245704e-05,
+      "loss": 0.6998,
+      "step": 11783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2476275968196975e-05,
+      "loss": 0.4326,
+      "step": 11784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.247563477814824e-05,
+      "loss": 0.5113,
+      "step": 11785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.247499358809952e-05,
+      "loss": 0.4842,
+      "step": 11786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2474352398050784e-05,
+      "loss": 0.4506,
+      "step": 11787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2473711208002055e-05,
+      "loss": 0.5792,
+      "step": 11788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.247307001795333e-05,
+      "loss": 0.5677,
+      "step": 11789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.247242882790459e-05,
+      "loss": 0.553,
+      "step": 11790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2471787637855864e-05,
+      "loss": 0.5646,
+      "step": 11791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.247114644780713e-05,
+      "loss": 0.4576,
+      "step": 11792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24705052577584e-05,
+      "loss": 0.5206,
+      "step": 11793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246986406770967e-05,
+      "loss": 0.4814,
+      "step": 11794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2469222877660944e-05,
+      "loss": 0.5178,
+      "step": 11795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246858168761221e-05,
+      "loss": 0.4351,
+      "step": 11796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246794049756348e-05,
+      "loss": 0.605,
+      "step": 11797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246729930751475e-05,
+      "loss": 0.4908,
+      "step": 11798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246665811746602e-05,
+      "loss": 0.5189,
+      "step": 11799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246601692741729e-05,
+      "loss": 0.4424,
+      "step": 11800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2465375737368554e-05,
+      "loss": 0.5339,
+      "step": 11801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2464734547319825e-05,
+      "loss": 0.5203,
+      "step": 11802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24640933572711e-05,
+      "loss": 0.4814,
+      "step": 11803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246345216722237e-05,
+      "loss": 0.6446,
+      "step": 11804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2462810977173634e-05,
+      "loss": 0.6145,
+      "step": 11805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2462169787124906e-05,
+      "loss": 0.757,
+      "step": 11806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246152859707618e-05,
+      "loss": 0.5988,
+      "step": 11807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246088740702744e-05,
+      "loss": 0.5149,
+      "step": 11808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2460246216978714e-05,
+      "loss": 0.5474,
+      "step": 11809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2459605026929986e-05,
+      "loss": 0.5142,
+      "step": 11810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245896383688126e-05,
+      "loss": 0.5906,
+      "step": 11811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245832264683252e-05,
+      "loss": 0.5102,
+      "step": 11812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2457681456783794e-05,
+      "loss": 0.675,
+      "step": 11813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245704026673506e-05,
+      "loss": 0.5222,
+      "step": 11814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245639907668633e-05,
+      "loss": 0.479,
+      "step": 11815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24557578866376e-05,
+      "loss": 0.5296,
+      "step": 11816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245511669658887e-05,
+      "loss": 0.5193,
+      "step": 11817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245447550654014e-05,
+      "loss": 0.5505,
+      "step": 11818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245383431649141e-05,
+      "loss": 0.5574,
+      "step": 11819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245319312644268e-05,
+      "loss": 0.4831,
+      "step": 11820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245255193639395e-05,
+      "loss": 0.5455,
+      "step": 11821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.245191074634522e-05,
+      "loss": 0.5816,
+      "step": 11822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2451269556296484e-05,
+      "loss": 0.643,
+      "step": 11823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2450628366247756e-05,
+      "loss": 0.736,
+      "step": 11824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.244998717619903e-05,
+      "loss": 0.6683,
+      "step": 11825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24493459861503e-05,
+      "loss": 0.551,
+      "step": 11826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.244870479610157e-05,
+      "loss": 0.6513,
+      "step": 11827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2448063606052836e-05,
+      "loss": 0.581,
+      "step": 11828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.244742241600411e-05,
+      "loss": 0.6264,
+      "step": 11829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.244678122595537e-05,
+      "loss": 0.5528,
+      "step": 11830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2446140035906644e-05,
+      "loss": 0.4682,
+      "step": 11831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.244549884585791e-05,
+      "loss": 0.5614,
+      "step": 11832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.244485765580918e-05,
+      "loss": 0.5428,
+      "step": 11833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.244421646576045e-05,
+      "loss": 0.5146,
+      "step": 11834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2443575275711724e-05,
+      "loss": 0.5817,
+      "step": 11835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2442934085662996e-05,
+      "loss": 0.58,
+      "step": 11836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.244229289561426e-05,
+      "loss": 0.5391,
+      "step": 11837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.244165170556553e-05,
+      "loss": 0.6082,
+      "step": 11838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24410105155168e-05,
+      "loss": 0.6616,
+      "step": 11839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.244036932546807e-05,
+      "loss": 0.5402,
+      "step": 11840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2439728135419334e-05,
+      "loss": 0.5038,
+      "step": 11841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243908694537061e-05,
+      "loss": 0.5683,
+      "step": 11842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2438445755321885e-05,
+      "loss": 0.4961,
+      "step": 11843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243780456527315e-05,
+      "loss": 0.5172,
+      "step": 11844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243716337522442e-05,
+      "loss": 0.6036,
+      "step": 11845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2436522185175686e-05,
+      "loss": 0.5033,
+      "step": 11846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243588099512696e-05,
+      "loss": 0.5091,
+      "step": 11847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243523980507822e-05,
+      "loss": 0.5719,
+      "step": 11848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2434598615029495e-05,
+      "loss": 0.4831,
+      "step": 11849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2433957424980766e-05,
+      "loss": 0.5198,
+      "step": 11850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243331623493204e-05,
+      "loss": 0.582,
+      "step": 11851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243267504488331e-05,
+      "loss": 0.6047,
+      "step": 11852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2432033854834575e-05,
+      "loss": 0.5672,
+      "step": 11853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2431392664785846e-05,
+      "loss": 0.5456,
+      "step": 11854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243075147473711e-05,
+      "loss": 0.5749,
+      "step": 11855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243011028468838e-05,
+      "loss": 0.4949,
+      "step": 11856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242946909463965e-05,
+      "loss": 0.4931,
+      "step": 11857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2428827904590926e-05,
+      "loss": 0.5015,
+      "step": 11858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242818671454219e-05,
+      "loss": 0.5095,
+      "step": 11859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242754552449346e-05,
+      "loss": 0.4272,
+      "step": 11860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2426904334444735e-05,
+      "loss": 0.6071,
+      "step": 11861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2426263144396e-05,
+      "loss": 0.5283,
+      "step": 11862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242562195434727e-05,
+      "loss": 0.5153,
+      "step": 11863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2424980764298536e-05,
+      "loss": 0.4747,
+      "step": 11864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242433957424981e-05,
+      "loss": 0.4986,
+      "step": 11865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242369838420108e-05,
+      "loss": 0.5304,
+      "step": 11866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242305719415235e-05,
+      "loss": 0.728,
+      "step": 11867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2422416004103616e-05,
+      "loss": 0.5278,
+      "step": 11868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242177481405489e-05,
+      "loss": 0.5127,
+      "step": 11869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242113362400616e-05,
+      "loss": 0.5177,
+      "step": 11870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2420492433957425e-05,
+      "loss": 0.4826,
+      "step": 11871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2419851243908697e-05,
+      "loss": 0.4722,
+      "step": 11872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241921005385996e-05,
+      "loss": 0.4906,
+      "step": 11873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241856886381123e-05,
+      "loss": 0.5404,
+      "step": 11874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2417927673762505e-05,
+      "loss": 0.5212,
+      "step": 11875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241728648371378e-05,
+      "loss": 0.5156,
+      "step": 11876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241664529366504e-05,
+      "loss": 0.5457,
+      "step": 11877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241600410361631e-05,
+      "loss": 0.5604,
+      "step": 11878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2415362913567585e-05,
+      "loss": 0.5803,
+      "step": 11879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241472172351885e-05,
+      "loss": 0.483,
+      "step": 11880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241408053347012e-05,
+      "loss": 0.5545,
+      "step": 11881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2413439343421393e-05,
+      "loss": 0.5466,
+      "step": 11882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2412798153372665e-05,
+      "loss": 0.5636,
+      "step": 11883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241215696332393e-05,
+      "loss": 0.5878,
+      "step": 11884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24115157732752e-05,
+      "loss": 0.6499,
+      "step": 11885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2410874583226474e-05,
+      "loss": 0.5546,
+      "step": 11886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.241023339317774e-05,
+      "loss": 0.6116,
+      "step": 11887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240959220312901e-05,
+      "loss": 0.6889,
+      "step": 11888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2408951013080275e-05,
+      "loss": 0.4714,
+      "step": 11889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240830982303155e-05,
+      "loss": 0.5754,
+      "step": 11890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240766863298282e-05,
+      "loss": 0.5744,
+      "step": 11891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240702744293409e-05,
+      "loss": 0.5003,
+      "step": 11892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2406386252885355e-05,
+      "loss": 0.5888,
+      "step": 11893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240574506283663e-05,
+      "loss": 0.5723,
+      "step": 11894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24051038727879e-05,
+      "loss": 0.5869,
+      "step": 11895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2404462682739164e-05,
+      "loss": 0.4705,
+      "step": 11896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2403821492690435e-05,
+      "loss": 0.627,
+      "step": 11897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240318030264171e-05,
+      "loss": 0.5318,
+      "step": 11898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240253911259298e-05,
+      "loss": 0.5168,
+      "step": 11899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2401897922544244e-05,
+      "loss": 0.4641,
+      "step": 11900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2401256732495515e-05,
+      "loss": 0.592,
+      "step": 11901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240061554244678e-05,
+      "loss": 0.5326,
+      "step": 11902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239997435239805e-05,
+      "loss": 0.6424,
+      "step": 11903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2399333162349324e-05,
+      "loss": 0.5492,
+      "step": 11904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239869197230059e-05,
+      "loss": 0.4691,
+      "step": 11905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239805078225186e-05,
+      "loss": 0.4884,
+      "step": 11906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239740959220313e-05,
+      "loss": 0.5722,
+      "step": 11907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2396768402154404e-05,
+      "loss": 0.6658,
+      "step": 11908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239612721210567e-05,
+      "loss": 0.6786,
+      "step": 11909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239548602205694e-05,
+      "loss": 0.4995,
+      "step": 11910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2394844832008205e-05,
+      "loss": 0.5505,
+      "step": 11911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239420364195948e-05,
+      "loss": 0.688,
+      "step": 11912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239356245191075e-05,
+      "loss": 0.5349,
+      "step": 11913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239292126186202e-05,
+      "loss": 0.5766,
+      "step": 11914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239228007181329e-05,
+      "loss": 0.5207,
+      "step": 11915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239163888176456e-05,
+      "loss": 0.5993,
+      "step": 11916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239099769171583e-05,
+      "loss": 0.6287,
+      "step": 11917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2390356501667094e-05,
+      "loss": 0.5549,
+      "step": 11918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2389715311618366e-05,
+      "loss": 0.5197,
+      "step": 11919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238907412156963e-05,
+      "loss": 0.6414,
+      "step": 11920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23884329315209e-05,
+      "loss": 0.5521,
+      "step": 11921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2387791741472174e-05,
+      "loss": 0.5456,
+      "step": 11922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2387150551423446e-05,
+      "loss": 0.512,
+      "step": 11923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238650936137472e-05,
+      "loss": 0.5845,
+      "step": 11924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238586817132598e-05,
+      "loss": 0.578,
+      "step": 11925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2385226981277254e-05,
+      "loss": 0.509,
+      "step": 11926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238458579122852e-05,
+      "loss": 0.5127,
+      "step": 11927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238394460117979e-05,
+      "loss": 0.503,
+      "step": 11928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2383303411131056e-05,
+      "loss": 0.5263,
+      "step": 11929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2382662221082334e-05,
+      "loss": 0.5716,
+      "step": 11930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2382021031033606e-05,
+      "loss": 0.5068,
+      "step": 11931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238137984098487e-05,
+      "loss": 0.5374,
+      "step": 11932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238073865093614e-05,
+      "loss": 0.5951,
+      "step": 11933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238009746088741e-05,
+      "loss": 0.5969,
+      "step": 11934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237945627083868e-05,
+      "loss": 0.5144,
+      "step": 11935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2378815080789944e-05,
+      "loss": 0.6123,
+      "step": 11936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2378173890741216e-05,
+      "loss": 0.5349,
+      "step": 11937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237753270069249e-05,
+      "loss": 0.562,
+      "step": 11938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237689151064376e-05,
+      "loss": 0.5446,
+      "step": 11939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237625032059503e-05,
+      "loss": 0.5813,
+      "step": 11940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2375609130546296e-05,
+      "loss": 0.6389,
+      "step": 11941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237496794049757e-05,
+      "loss": 0.5881,
+      "step": 11942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237432675044883e-05,
+      "loss": 0.5508,
+      "step": 11943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2373685560400104e-05,
+      "loss": 0.6425,
+      "step": 11944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237304437035137e-05,
+      "loss": 0.504,
+      "step": 11945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237240318030265e-05,
+      "loss": 0.5958,
+      "step": 11946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237176199025391e-05,
+      "loss": 0.5394,
+      "step": 11947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2371120800205184e-05,
+      "loss": 0.5438,
+      "step": 11948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2370479610156456e-05,
+      "loss": 0.4463,
+      "step": 11949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236983842010772e-05,
+      "loss": 0.5601,
+      "step": 11950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236919723005899e-05,
+      "loss": 0.4925,
+      "step": 11951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236855604001026e-05,
+      "loss": 0.5785,
+      "step": 11952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236791484996153e-05,
+      "loss": 0.6515,
+      "step": 11953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23672736599128e-05,
+      "loss": 0.6647,
+      "step": 11954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236663246986407e-05,
+      "loss": 0.5445,
+      "step": 11955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236599127981534e-05,
+      "loss": 0.5564,
+      "step": 11956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236535008976661e-05,
+      "loss": 0.4385,
+      "step": 11957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236470889971788e-05,
+      "loss": 0.524,
+      "step": 11958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2364067709669146e-05,
+      "loss": 0.5648,
+      "step": 11959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236342651962042e-05,
+      "loss": 0.5541,
+      "step": 11960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236278532957168e-05,
+      "loss": 0.5671,
+      "step": 11961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2362144139522955e-05,
+      "loss": 0.6022,
+      "step": 11962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2361502949474226e-05,
+      "loss": 0.477,
+      "step": 11963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23608617594255e-05,
+      "loss": 0.5538,
+      "step": 11964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236022056937676e-05,
+      "loss": 0.5054,
+      "step": 11965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2359579379328035e-05,
+      "loss": 0.4291,
+      "step": 11966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2358938189279306e-05,
+      "loss": 0.5205,
+      "step": 11967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235829699923057e-05,
+      "loss": 0.6077,
+      "step": 11968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235765580918184e-05,
+      "loss": 0.5642,
+      "step": 11969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2357014619133115e-05,
+      "loss": 0.5649,
+      "step": 11970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2356373429084386e-05,
+      "loss": 0.4602,
+      "step": 11971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235573223903565e-05,
+      "loss": 0.6458,
+      "step": 11972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235509104898692e-05,
+      "loss": 0.5706,
+      "step": 11973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235444985893819e-05,
+      "loss": 0.6437,
+      "step": 11974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235380866888946e-05,
+      "loss": 0.638,
+      "step": 11975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235316747884073e-05,
+      "loss": 0.421,
+      "step": 11976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2352526288791996e-05,
+      "loss": 0.5538,
+      "step": 11977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235188509874327e-05,
+      "loss": 0.5316,
+      "step": 11978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235124390869454e-05,
+      "loss": 0.5934,
+      "step": 11979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235060271864581e-05,
+      "loss": 0.522,
+      "step": 11980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2349961528597077e-05,
+      "loss": 0.5055,
+      "step": 11981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234932033854835e-05,
+      "loss": 0.5978,
+      "step": 11982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234867914849961e-05,
+      "loss": 0.6822,
+      "step": 11983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2348037958450885e-05,
+      "loss": 0.6214,
+      "step": 11984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234739676840216e-05,
+      "loss": 0.4789,
+      "step": 11985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234675557835343e-05,
+      "loss": 0.5373,
+      "step": 11986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23461143883047e-05,
+      "loss": 0.5347,
+      "step": 11987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2345473198255965e-05,
+      "loss": 0.6236,
+      "step": 11988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234483200820724e-05,
+      "loss": 0.5161,
+      "step": 11989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23441908181585e-05,
+      "loss": 0.5808,
+      "step": 11990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234354962810977e-05,
+      "loss": 0.4663,
+      "step": 11991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234290843806104e-05,
+      "loss": 0.4997,
+      "step": 11992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234226724801231e-05,
+      "loss": 0.5041,
+      "step": 11993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234162605796358e-05,
+      "loss": 0.6365,
+      "step": 11994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2340984867914853e-05,
+      "loss": 0.5879,
+      "step": 11995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2340343677866125e-05,
+      "loss": 0.5504,
+      "step": 11996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233970248781739e-05,
+      "loss": 0.548,
+      "step": 11997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233906129776866e-05,
+      "loss": 0.4776,
+      "step": 11998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233842010771993e-05,
+      "loss": 0.5824,
+      "step": 11999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23377789176712e-05,
+      "loss": 0.5661,
+      "step": 12000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2337137727622463e-05,
+      "loss": 0.6114,
+      "step": 12001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233649653757374e-05,
+      "loss": 0.6086,
+      "step": 12002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2335855347525014e-05,
+      "loss": 0.4698,
+      "step": 12003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233521415747628e-05,
+      "loss": 0.639,
+      "step": 12004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233457296742755e-05,
+      "loss": 0.5525,
+      "step": 12005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2333931777378815e-05,
+      "loss": 0.6398,
+      "step": 12006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233329058733009e-05,
+      "loss": 0.5965,
+      "step": 12007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233264939728135e-05,
+      "loss": 0.592,
+      "step": 12008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2332008207232624e-05,
+      "loss": 0.5183,
+      "step": 12009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2331367017183895e-05,
+      "loss": 0.5765,
+      "step": 12010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233072582713517e-05,
+      "loss": 0.4918,
+      "step": 12011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233008463708644e-05,
+      "loss": 0.5039,
+      "step": 12012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2329443447037704e-05,
+      "loss": 0.5667,
+      "step": 12013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2328802256988975e-05,
+      "loss": 0.5317,
+      "step": 12014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232816106694024e-05,
+      "loss": 0.5013,
+      "step": 12015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232751987689151e-05,
+      "loss": 0.5407,
+      "step": 12016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232687868684278e-05,
+      "loss": 0.5449,
+      "step": 12017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2326237496794055e-05,
+      "loss": 0.501,
+      "step": 12018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232559630674532e-05,
+      "loss": 0.4998,
+      "step": 12019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232495511669659e-05,
+      "loss": 0.5316,
+      "step": 12020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2324313926647864e-05,
+      "loss": 0.5284,
+      "step": 12021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232367273659913e-05,
+      "loss": 0.7048,
+      "step": 12022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23230315465504e-05,
+      "loss": 0.6294,
+      "step": 12023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2322390356501665e-05,
+      "loss": 0.5834,
+      "step": 12024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232174916645294e-05,
+      "loss": 0.4969,
+      "step": 12025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232110797640421e-05,
+      "loss": 0.5904,
+      "step": 12026
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.232046678635548e-05,
+      "loss": 0.5258,
+      "step": 12027
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2319825596306746e-05,
+      "loss": 0.5232,
+      "step": 12028
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231918440625802e-05,
+      "loss": 0.5678,
+      "step": 12029
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231854321620929e-05,
+      "loss": 0.5608,
+      "step": 12030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2317902026160554e-05,
+      "loss": 0.5481,
+      "step": 12031
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2317260836111826e-05,
+      "loss": 0.4797,
+      "step": 12032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231661964606309e-05,
+      "loss": 0.5896,
+      "step": 12033
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231597845601437e-05,
+      "loss": 0.5362,
+      "step": 12034
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2315337265965634e-05,
+      "loss": 0.5766,
+      "step": 12035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2314696075916906e-05,
+      "loss": 0.5898,
+      "step": 12036
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231405488586817e-05,
+      "loss": 0.4385,
+      "step": 12037
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231341369581944e-05,
+      "loss": 0.5906,
+      "step": 12038
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2312772505770714e-05,
+      "loss": 0.5887,
+      "step": 12039
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231213131572198e-05,
+      "loss": 0.5611,
+      "step": 12040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231149012567325e-05,
+      "loss": 0.5651,
+      "step": 12041
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231084893562452e-05,
+      "loss": 0.5935,
+      "step": 12042
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2310207745575794e-05,
+      "loss": 0.5958,
+      "step": 12043
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230956655552706e-05,
+      "loss": 0.5094,
+      "step": 12044
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230892536547833e-05,
+      "loss": 0.5421,
+      "step": 12045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2308284175429596e-05,
+      "loss": 0.5707,
+      "step": 12046
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230764298538087e-05,
+      "loss": 0.5936,
+      "step": 12047
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230700179533214e-05,
+      "loss": 0.5196,
+      "step": 12048
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2306360605283404e-05,
+      "loss": 0.6253,
+      "step": 12049
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2305719415234676e-05,
+      "loss": 0.5177,
+      "step": 12050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230507822518595e-05,
+      "loss": 0.4913,
+      "step": 12051
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230443703513722e-05,
+      "loss": 0.559,
+      "step": 12052
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2303795845088484e-05,
+      "loss": 0.5699,
+      "step": 12053
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2303154655039756e-05,
+      "loss": 0.5652,
+      "step": 12054
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230251346499102e-05,
+      "loss": 0.5802,
+      "step": 12055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230187227494229e-05,
+      "loss": 0.5239,
+      "step": 12056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2301231084893564e-05,
+      "loss": 0.5143,
+      "step": 12057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2300589894844836e-05,
+      "loss": 0.5644,
+      "step": 12058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229994870479611e-05,
+      "loss": 0.5638,
+      "step": 12059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229930751474737e-05,
+      "loss": 0.5364,
+      "step": 12060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2298666324698644e-05,
+      "loss": 0.5569,
+      "step": 12061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229802513464991e-05,
+      "loss": 0.4145,
+      "step": 12062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229738394460118e-05,
+      "loss": 0.6914,
+      "step": 12063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2296742754552446e-05,
+      "loss": 0.4436,
+      "step": 12064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229610156450372e-05,
+      "loss": 0.6146,
+      "step": 12065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229546037445499e-05,
+      "loss": 0.5788,
+      "step": 12066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229481918440626e-05,
+      "loss": 0.4962,
+      "step": 12067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229417799435753e-05,
+      "loss": 0.5441,
+      "step": 12068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.22935368043088e-05,
+      "loss": 0.4701,
+      "step": 12069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229289561426007e-05,
+      "loss": 0.4776,
+      "step": 12070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2292254424211335e-05,
+      "loss": 0.6993,
+      "step": 12071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2291613234162606e-05,
+      "loss": 0.6217,
+      "step": 12072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229097204411388e-05,
+      "loss": 0.5406,
+      "step": 12073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.229033085406515e-05,
+      "loss": 0.6227,
+      "step": 12074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228968966401642e-05,
+      "loss": 0.5225,
+      "step": 12075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2289048473967686e-05,
+      "loss": 0.6884,
+      "step": 12076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228840728391896e-05,
+      "loss": 0.5499,
+      "step": 12077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228776609387022e-05,
+      "loss": 0.6102,
+      "step": 12078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2287124903821495e-05,
+      "loss": 0.5413,
+      "step": 12079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228648371377276e-05,
+      "loss": 0.5778,
+      "step": 12080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228584252372403e-05,
+      "loss": 0.4672,
+      "step": 12081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.22852013336753e-05,
+      "loss": 0.6324,
+      "step": 12082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2284560143626575e-05,
+      "loss": 0.4738,
+      "step": 12083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2283918953577846e-05,
+      "loss": 0.452,
+      "step": 12084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228327776352911e-05,
+      "loss": 0.5389,
+      "step": 12085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228263657348038e-05,
+      "loss": 0.5791,
+      "step": 12086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228199538343165e-05,
+      "loss": 0.6544,
+      "step": 12087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228135419338292e-05,
+      "loss": 0.4793,
+      "step": 12088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2280713003334185e-05,
+      "loss": 0.5271,
+      "step": 12089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.228007181328546e-05,
+      "loss": 0.5539,
+      "step": 12090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2279430623236735e-05,
+      "loss": 0.5564,
+      "step": 12091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2278789433188e-05,
+      "loss": 0.5202,
+      "step": 12092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.227814824313927e-05,
+      "loss": 0.4964,
+      "step": 12093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2277507053090537e-05,
+      "loss": 0.5342,
+      "step": 12094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227686586304181e-05,
+      "loss": 0.5457,
+      "step": 12095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227622467299307e-05,
+      "loss": 0.5309,
+      "step": 12096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2275583482944345e-05,
+      "loss": 0.4863,
+      "step": 12097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227494229289562e-05,
+      "loss": 0.464,
+      "step": 12098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227430110284689e-05,
+      "loss": 0.6255,
+      "step": 12099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227365991279816e-05,
+      "loss": 0.5119,
+      "step": 12100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2273018722749425e-05,
+      "loss": 0.5538,
+      "step": 12101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22723775327007e-05,
+      "loss": 0.5757,
+      "step": 12102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227173634265196e-05,
+      "loss": 0.5902,
+      "step": 12103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2271095152603233e-05,
+      "loss": 0.5905,
+      "step": 12104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22704539625545e-05,
+      "loss": 0.6459,
+      "step": 12105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226981277250578e-05,
+      "loss": 0.4203,
+      "step": 12106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226917158245704e-05,
+      "loss": 0.4745,
+      "step": 12107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2268530392408313e-05,
+      "loss": 0.5733,
+      "step": 12108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2267889202359585e-05,
+      "loss": 0.6422,
+      "step": 12109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226724801231085e-05,
+      "loss": 0.4755,
+      "step": 12110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226660682226212e-05,
+      "loss": 0.5451,
+      "step": 12111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226596563221339e-05,
+      "loss": 0.5173,
+      "step": 12112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226532444216466e-05,
+      "loss": 0.4827,
+      "step": 12113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226468325211593e-05,
+      "loss": 0.5802,
+      "step": 12114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22640420620672e-05,
+      "loss": 0.6085,
+      "step": 12115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226340087201847e-05,
+      "loss": 0.5992,
+      "step": 12116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226275968196974e-05,
+      "loss": 0.5972,
+      "step": 12117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226211849192101e-05,
+      "loss": 0.5816,
+      "step": 12118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2261477301872275e-05,
+      "loss": 0.6695,
+      "step": 12119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226083611182355e-05,
+      "loss": 0.3927,
+      "step": 12120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226019492177481e-05,
+      "loss": 0.539,
+      "step": 12121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2259553731726084e-05,
+      "loss": 0.4812,
+      "step": 12122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2258912541677355e-05,
+      "loss": 0.5395,
+      "step": 12123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225827135162863e-05,
+      "loss": 0.5003,
+      "step": 12124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225763016157989e-05,
+      "loss": 0.465,
+      "step": 12125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2256988971531164e-05,
+      "loss": 0.5143,
+      "step": 12126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2256347781482435e-05,
+      "loss": 0.5276,
+      "step": 12127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22557065914337e-05,
+      "loss": 0.5189,
+      "step": 12128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225506540138497e-05,
+      "loss": 0.6124,
+      "step": 12129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2254424211336244e-05,
+      "loss": 0.6429,
+      "step": 12130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2253783021287516e-05,
+      "loss": 0.5427,
+      "step": 12131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225314183123878e-05,
+      "loss": 0.6214,
+      "step": 12132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225250064119005e-05,
+      "loss": 0.6442,
+      "step": 12133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225185945114132e-05,
+      "loss": 0.4941,
+      "step": 12134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225121826109259e-05,
+      "loss": 0.6157,
+      "step": 12135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225057707104386e-05,
+      "loss": 0.5218,
+      "step": 12136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2249935880995126e-05,
+      "loss": 0.5234,
+      "step": 12137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22492946909464e-05,
+      "loss": 0.5095,
+      "step": 12138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224865350089767e-05,
+      "loss": 0.6407,
+      "step": 12139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224801231084894e-05,
+      "loss": 0.552,
+      "step": 12140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2247371120800206e-05,
+      "loss": 0.5505,
+      "step": 12141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224672993075148e-05,
+      "loss": 0.5304,
+      "step": 12142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224608874070274e-05,
+      "loss": 0.5956,
+      "step": 12143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2245447550654014e-05,
+      "loss": 0.6062,
+      "step": 12144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2244806360605286e-05,
+      "loss": 0.4255,
+      "step": 12145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224416517055656e-05,
+      "loss": 0.6674,
+      "step": 12146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224352398050783e-05,
+      "loss": 0.5427,
+      "step": 12147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2242882790459094e-05,
+      "loss": 0.5663,
+      "step": 12148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2242241600410366e-05,
+      "loss": 0.5659,
+      "step": 12149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224160041036163e-05,
+      "loss": 0.6006,
+      "step": 12150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22409592203129e-05,
+      "loss": 0.4933,
+      "step": 12151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224031803026417e-05,
+      "loss": 0.5108,
+      "step": 12152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223967684021544e-05,
+      "loss": 0.4834,
+      "step": 12153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223903565016671e-05,
+      "loss": 0.6072,
+      "step": 12154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223839446011798e-05,
+      "loss": 0.6264,
+      "step": 12155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2237753270069254e-05,
+      "loss": 0.5402,
+      "step": 12156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223711208002052e-05,
+      "loss": 0.6813,
+      "step": 12157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223647088997179e-05,
+      "loss": 0.4983,
+      "step": 12158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2235829699923056e-05,
+      "loss": 0.6084,
+      "step": 12159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223518850987433e-05,
+      "loss": 0.5975,
+      "step": 12160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223454731982559e-05,
+      "loss": 0.5829,
+      "step": 12161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223390612977687e-05,
+      "loss": 0.647,
+      "step": 12162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223326493972814e-05,
+      "loss": 0.5935,
+      "step": 12163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223262374967941e-05,
+      "loss": 0.5266,
+      "step": 12164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223198255963068e-05,
+      "loss": 0.5338,
+      "step": 12165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2231341369581944e-05,
+      "loss": 0.5583,
+      "step": 12166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2230700179533216e-05,
+      "loss": 0.6014,
+      "step": 12167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.223005898948448e-05,
+      "loss": 0.4814,
+      "step": 12168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222941779943575e-05,
+      "loss": 0.6071,
+      "step": 12169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2228776609387024e-05,
+      "loss": 0.634,
+      "step": 12170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2228135419338296e-05,
+      "loss": 0.6243,
+      "step": 12171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222749422928957e-05,
+      "loss": 0.5913,
+      "step": 12172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222685303924083e-05,
+      "loss": 0.5421,
+      "step": 12173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2226211849192104e-05,
+      "loss": 0.6109,
+      "step": 12174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222557065914337e-05,
+      "loss": 0.5893,
+      "step": 12175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222492946909464e-05,
+      "loss": 0.5172,
+      "step": 12176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2224288279045906e-05,
+      "loss": 0.572,
+      "step": 12177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2223647088997185e-05,
+      "loss": 0.5527,
+      "step": 12178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222300589894845e-05,
+      "loss": 0.5621,
+      "step": 12179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222236470889972e-05,
+      "loss": 0.5875,
+      "step": 12180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222172351885099e-05,
+      "loss": 0.5269,
+      "step": 12181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222108232880226e-05,
+      "loss": 0.5465,
+      "step": 12182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222044113875353e-05,
+      "loss": 0.5632,
+      "step": 12183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2219799948704795e-05,
+      "loss": 0.5407,
+      "step": 12184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2219158758656066e-05,
+      "loss": 0.6259,
+      "step": 12185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221851756860734e-05,
+      "loss": 0.4882,
+      "step": 12186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221787637855861e-05,
+      "loss": 0.4721,
+      "step": 12187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2217235188509875e-05,
+      "loss": 0.5784,
+      "step": 12188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2216593998461146e-05,
+      "loss": 0.5022,
+      "step": 12189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221595280841242e-05,
+      "loss": 0.5291,
+      "step": 12190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221531161836368e-05,
+      "loss": 0.5528,
+      "step": 12191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2214670428314955e-05,
+      "loss": 0.5919,
+      "step": 12192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221402923826622e-05,
+      "loss": 0.5577,
+      "step": 12193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22133880482175e-05,
+      "loss": 0.5958,
+      "step": 12194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221274685816876e-05,
+      "loss": 0.4638,
+      "step": 12195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2212105668120035e-05,
+      "loss": 0.5504,
+      "step": 12196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22114644780713e-05,
+      "loss": 0.5521,
+      "step": 12197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221082328802257e-05,
+      "loss": 0.5545,
+      "step": 12198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221018209797384e-05,
+      "loss": 0.4605,
+      "step": 12199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220954090792511e-05,
+      "loss": 0.5816,
+      "step": 12200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220889971787638e-05,
+      "loss": 0.581,
+      "step": 12201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220825852782765e-05,
+      "loss": 0.5437,
+      "step": 12202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220761733777892e-05,
+      "loss": 0.5475,
+      "step": 12203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220697614773019e-05,
+      "loss": 0.4816,
+      "step": 12204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220633495768146e-05,
+      "loss": 0.5396,
+      "step": 12205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2205693767632725e-05,
+      "loss": 0.5735,
+      "step": 12206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2205052577583997e-05,
+      "loss": 0.5308,
+      "step": 12207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220441138753527e-05,
+      "loss": 0.4971,
+      "step": 12208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220377019748653e-05,
+      "loss": 0.5764,
+      "step": 12209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2203129007437805e-05,
+      "loss": 0.4276,
+      "step": 12210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220248781738908e-05,
+      "loss": 0.5691,
+      "step": 12211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220184662734035e-05,
+      "loss": 0.6666,
+      "step": 12212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220120543729161e-05,
+      "loss": 0.4881,
+      "step": 12213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2200564247242885e-05,
+      "loss": 0.5682,
+      "step": 12214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219992305719415e-05,
+      "loss": 0.5581,
+      "step": 12215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219928186714542e-05,
+      "loss": 0.5414,
+      "step": 12216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2198640677096693e-05,
+      "loss": 0.5141,
+      "step": 12217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2197999487047965e-05,
+      "loss": 0.4749,
+      "step": 12218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219735829699924e-05,
+      "loss": 0.6847,
+      "step": 12219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21967171069505e-05,
+      "loss": 0.5843,
+      "step": 12220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2196075916901774e-05,
+      "loss": 0.4779,
+      "step": 12221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219543472685304e-05,
+      "loss": 0.5331,
+      "step": 12222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219479353680431e-05,
+      "loss": 0.5647,
+      "step": 12223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2194152346755575e-05,
+      "loss": 0.5174,
+      "step": 12224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219351115670685e-05,
+      "loss": 0.5578,
+      "step": 12225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219286996665812e-05,
+      "loss": 0.5125,
+      "step": 12226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219222877660939e-05,
+      "loss": 0.5468,
+      "step": 12227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219158758656066e-05,
+      "loss": 0.5459,
+      "step": 12228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219094639651193e-05,
+      "loss": 0.5439,
+      "step": 12229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21903052064632e-05,
+      "loss": 0.5242,
+      "step": 12230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2189664016414464e-05,
+      "loss": 0.5683,
+      "step": 12231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2189022826365735e-05,
+      "loss": 0.5235,
+      "step": 12232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218838163631701e-05,
+      "loss": 0.6344,
+      "step": 12233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218774044626828e-05,
+      "loss": 0.5321,
+      "step": 12234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218709925621955e-05,
+      "loss": 0.5508,
+      "step": 12235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2186458066170815e-05,
+      "loss": 0.5834,
+      "step": 12236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218581687612209e-05,
+      "loss": 0.5341,
+      "step": 12237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218517568607335e-05,
+      "loss": 0.4933,
+      "step": 12238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2184534496024624e-05,
+      "loss": 0.582,
+      "step": 12239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218389330597589e-05,
+      "loss": 0.6801,
+      "step": 12240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218325211592716e-05,
+      "loss": 0.5893,
+      "step": 12241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218261092587843e-05,
+      "loss": 0.482,
+      "step": 12242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2181969735829704e-05,
+      "loss": 0.5299,
+      "step": 12243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2181328545780976e-05,
+      "loss": 0.6225,
+      "step": 12244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218068735573224e-05,
+      "loss": 0.5973,
+      "step": 12245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218004616568351e-05,
+      "loss": 0.5832,
+      "step": 12246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217940497563478e-05,
+      "loss": 0.6499,
+      "step": 12247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217876378558605e-05,
+      "loss": 0.5256,
+      "step": 12248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2178122595537314e-05,
+      "loss": 0.5736,
+      "step": 12249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217748140548859e-05,
+      "loss": 0.5879,
+      "step": 12250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217684021543986e-05,
+      "loss": 0.5638,
+      "step": 12251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217619902539113e-05,
+      "loss": 0.5488,
+      "step": 12252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21755578353424e-05,
+      "loss": 0.6537,
+      "step": 12253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2174916645293666e-05,
+      "loss": 0.6209,
+      "step": 12254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217427545524494e-05,
+      "loss": 0.5392,
+      "step": 12255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21736342651962e-05,
+      "loss": 0.585,
+      "step": 12256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2172993075147474e-05,
+      "loss": 0.4906,
+      "step": 12257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2172351885098746e-05,
+      "loss": 0.5955,
+      "step": 12258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217171069505002e-05,
+      "loss": 0.575,
+      "step": 12259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217106950500129e-05,
+      "loss": 0.5988,
+      "step": 12260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2170428314952554e-05,
+      "loss": 0.4937,
+      "step": 12261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2169787124903826e-05,
+      "loss": 0.4956,
+      "step": 12262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216914593485509e-05,
+      "loss": 0.5,
+      "step": 12263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216850474480636e-05,
+      "loss": 0.5117,
+      "step": 12264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216786355475763e-05,
+      "loss": 0.4871,
+      "step": 12265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2167222364708906e-05,
+      "loss": 0.5353,
+      "step": 12266
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216658117466017e-05,
+      "loss": 0.5091,
+      "step": 12267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216593998461144e-05,
+      "loss": 0.6402,
+      "step": 12268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2165298794562714e-05,
+      "loss": 0.6363,
+      "step": 12269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216465760451398e-05,
+      "loss": 0.5462,
+      "step": 12270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216401641446525e-05,
+      "loss": 0.6112,
+      "step": 12271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2163375224416516e-05,
+      "loss": 0.5152,
+      "step": 12272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216273403436779e-05,
+      "loss": 0.5278,
+      "step": 12273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216209284431906e-05,
+      "loss": 0.5366,
+      "step": 12274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216145165427033e-05,
+      "loss": 0.5947,
+      "step": 12275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2160810464221596e-05,
+      "loss": 0.7104,
+      "step": 12276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216016927417287e-05,
+      "loss": 0.6322,
+      "step": 12277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215952808412414e-05,
+      "loss": 0.5937,
+      "step": 12278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2158886894075404e-05,
+      "loss": 0.5521,
+      "step": 12279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2158245704026676e-05,
+      "loss": 0.5783,
+      "step": 12280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215760451397794e-05,
+      "loss": 0.564,
+      "step": 12281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215696332392922e-05,
+      "loss": 0.485,
+      "step": 12282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2156322133880484e-05,
+      "loss": 0.6022,
+      "step": 12283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2155680943831756e-05,
+      "loss": 0.4773,
+      "step": 12284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215503975378302e-05,
+      "loss": 0.5027,
+      "step": 12285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215439856373429e-05,
+      "loss": 0.56,
+      "step": 12286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2153757373685565e-05,
+      "loss": 0.5485,
+      "step": 12287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215311618363683e-05,
+      "loss": 0.5243,
+      "step": 12288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21524749935881e-05,
+      "loss": 0.5265,
+      "step": 12289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215183380353937e-05,
+      "loss": 0.6111,
+      "step": 12290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2151192613490645e-05,
+      "loss": 0.516,
+      "step": 12291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215055142344191e-05,
+      "loss": 0.5087,
+      "step": 12292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214991023339318e-05,
+      "loss": 0.5885,
+      "step": 12293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2149269043344446e-05,
+      "loss": 0.569,
+      "step": 12294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214862785329572e-05,
+      "loss": 0.5339,
+      "step": 12295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214798666324699e-05,
+      "loss": 0.4566,
+      "step": 12296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2147345473198255e-05,
+      "loss": 0.6076,
+      "step": 12297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2146704283149526e-05,
+      "loss": 0.5199,
+      "step": 12298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21460630931008e-05,
+      "loss": 0.609,
+      "step": 12299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214542190305207e-05,
+      "loss": 0.5794,
+      "step": 12300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2144780713003335e-05,
+      "loss": 0.4939,
+      "step": 12301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2144139522954606e-05,
+      "loss": 0.5884,
+      "step": 12302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214349833290587e-05,
+      "loss": 0.7111,
+      "step": 12303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214285714285714e-05,
+      "loss": 0.6247,
+      "step": 12304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2142215952808415e-05,
+      "loss": 0.5459,
+      "step": 12305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2141574762759686e-05,
+      "loss": 0.4454,
+      "step": 12306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214093357271096e-05,
+      "loss": 0.6927,
+      "step": 12307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.214029238266222e-05,
+      "loss": 0.5172,
+      "step": 12308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2139651192613495e-05,
+      "loss": 0.5717,
+      "step": 12309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213901000256476e-05,
+      "loss": 0.5442,
+      "step": 12310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213836881251603e-05,
+      "loss": 0.5771,
+      "step": 12311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2137727622467296e-05,
+      "loss": 0.5761,
+      "step": 12312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213708643241857e-05,
+      "loss": 0.4679,
+      "step": 12313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213644524236984e-05,
+      "loss": 0.5485,
+      "step": 12314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213580405232111e-05,
+      "loss": 0.5058,
+      "step": 12315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213516286227238e-05,
+      "loss": 0.4025,
+      "step": 12316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213452167222365e-05,
+      "loss": 0.5833,
+      "step": 12317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213388048217492e-05,
+      "loss": 0.6246,
+      "step": 12318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2133239292126185e-05,
+      "loss": 0.5836,
+      "step": 12319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213259810207746e-05,
+      "loss": 0.5812,
+      "step": 12320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213195691202873e-05,
+      "loss": 0.491,
+      "step": 12321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213131572198e-05,
+      "loss": 0.6128,
+      "step": 12322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213067453193127e-05,
+      "loss": 0.5199,
+      "step": 12323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213003334188254e-05,
+      "loss": 0.478,
+      "step": 12324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.212939215183381e-05,
+      "loss": 0.5038,
+      "step": 12325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.212875096178507e-05,
+      "loss": 0.4848,
+      "step": 12326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2128109771736345e-05,
+      "loss": 0.425,
+      "step": 12327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.212746858168761e-05,
+      "loss": 0.619,
+      "step": 12328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.212682739163888e-05,
+      "loss": 0.5105,
+      "step": 12329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2126186201590153e-05,
+      "loss": 0.6648,
+      "step": 12330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2125545011541425e-05,
+      "loss": 0.5328,
+      "step": 12331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21249038214927e-05,
+      "loss": 0.7082,
+      "step": 12332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.212426263144396e-05,
+      "loss": 0.5765,
+      "step": 12333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2123621441395234e-05,
+      "loss": 0.6075,
+      "step": 12334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21229802513465e-05,
+      "loss": 0.5132,
+      "step": 12335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.212233906129777e-05,
+      "loss": 0.5204,
+      "step": 12336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2121697871249035e-05,
+      "loss": 0.4683,
+      "step": 12337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2121056681200314e-05,
+      "loss": 0.662,
+      "step": 12338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.212041549115158e-05,
+      "loss": 0.5442,
+      "step": 12339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211977430110285e-05,
+      "loss": 0.6129,
+      "step": 12340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211913311105412e-05,
+      "loss": 0.5489,
+      "step": 12341
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211849192100539e-05,
+      "loss": 0.6382,
+      "step": 12342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211785073095666e-05,
+      "loss": 0.6035,
+      "step": 12343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2117209540907924e-05,
+      "loss": 0.5983,
+      "step": 12344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2116568350859195e-05,
+      "loss": 0.5833,
+      "step": 12345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211592716081047e-05,
+      "loss": 0.5742,
+      "step": 12346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211528597076174e-05,
+      "loss": 0.5177,
+      "step": 12347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2114644780713004e-05,
+      "loss": 0.5395,
+      "step": 12348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2114003590664275e-05,
+      "loss": 0.563,
+      "step": 12349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211336240061555e-05,
+      "loss": 0.6256,
+      "step": 12350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211272121056681e-05,
+      "loss": 0.5626,
+      "step": 12351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2112080020518084e-05,
+      "loss": 0.5268,
+      "step": 12352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211143883046935e-05,
+      "loss": 0.5381,
+      "step": 12353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211079764042063e-05,
+      "loss": 0.5455,
+      "step": 12354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.211015645037189e-05,
+      "loss": 0.5822,
+      "step": 12355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2109515260323164e-05,
+      "loss": 0.5234,
+      "step": 12356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210887407027443e-05,
+      "loss": 0.5251,
+      "step": 12357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.21082328802257e-05,
+      "loss": 0.4861,
+      "step": 12358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210759169017697e-05,
+      "loss": 0.5322,
+      "step": 12359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210695050012824e-05,
+      "loss": 0.5235,
+      "step": 12360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210630931007951e-05,
+      "loss": 0.5409,
+      "step": 12361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210566812003078e-05,
+      "loss": 0.5597,
+      "step": 12362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210502692998205e-05,
+      "loss": 0.4881,
+      "step": 12363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210438573993332e-05,
+      "loss": 0.5911,
+      "step": 12364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210374454988459e-05,
+      "loss": 0.5624,
+      "step": 12365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2103103359835854e-05,
+      "loss": 0.5823,
+      "step": 12366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2102462169787126e-05,
+      "loss": 0.6175,
+      "step": 12367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.21018209797384e-05,
+      "loss": 0.4981,
+      "step": 12368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210117978968966e-05,
+      "loss": 0.6046,
+      "step": 12369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2100538599640934e-05,
+      "loss": 0.5718,
+      "step": 12370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2099897409592206e-05,
+      "loss": 0.6384,
+      "step": 12371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209925621954348e-05,
+      "loss": 0.5032,
+      "step": 12372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209861502949474e-05,
+      "loss": 0.5351,
+      "step": 12373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2097973839446014e-05,
+      "loss": 0.5679,
+      "step": 12374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209733264939728e-05,
+      "loss": 0.5039,
+      "step": 12375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209669145934855e-05,
+      "loss": 0.5633,
+      "step": 12376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209605026929982e-05,
+      "loss": 0.5156,
+      "step": 12377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2095409079251094e-05,
+      "loss": 0.4492,
+      "step": 12378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2094767889202366e-05,
+      "loss": 0.598,
+      "step": 12379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209412669915363e-05,
+      "loss": 0.5956,
+      "step": 12380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20934855091049e-05,
+      "loss": 0.5236,
+      "step": 12381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209284431905617e-05,
+      "loss": 0.5805,
+      "step": 12382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209220312900744e-05,
+      "loss": 0.6414,
+      "step": 12383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2091561938958704e-05,
+      "loss": 0.5263,
+      "step": 12384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2090920748909976e-05,
+      "loss": 0.5527,
+      "step": 12385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.209027955886125e-05,
+      "loss": 0.53,
+      "step": 12386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208963836881252e-05,
+      "loss": 0.5023,
+      "step": 12387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208899717876379e-05,
+      "loss": 0.6101,
+      "step": 12388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2088355988715056e-05,
+      "loss": 0.6111,
+      "step": 12389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208771479866633e-05,
+      "loss": 0.6324,
+      "step": 12390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208707360861759e-05,
+      "loss": 0.6266,
+      "step": 12391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2086432418568864e-05,
+      "loss": 0.6428,
+      "step": 12392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2085791228520136e-05,
+      "loss": 0.5513,
+      "step": 12393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208515003847141e-05,
+      "loss": 0.5898,
+      "step": 12394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208450884842268e-05,
+      "loss": 0.4978,
+      "step": 12395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2083867658373944e-05,
+      "loss": 0.5454,
+      "step": 12396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2083226468325216e-05,
+      "loss": 0.4975,
+      "step": 12397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208258527827648e-05,
+      "loss": 0.627,
+      "step": 12398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208194408822775e-05,
+      "loss": 0.4933,
+      "step": 12399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208130289817902e-05,
+      "loss": 0.5291,
+      "step": 12400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208066170813029e-05,
+      "loss": 0.5966,
+      "step": 12401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208002051808156e-05,
+      "loss": 0.4404,
+      "step": 12402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207937932803283e-05,
+      "loss": 0.5217,
+      "step": 12403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2078738137984105e-05,
+      "loss": 0.5364,
+      "step": 12404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207809694793537e-05,
+      "loss": 0.5084,
+      "step": 12405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207745575788664e-05,
+      "loss": 0.5881,
+      "step": 12406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2076814567837906e-05,
+      "loss": 0.5691,
+      "step": 12407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207617337778918e-05,
+      "loss": 0.5293,
+      "step": 12408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207553218774044e-05,
+      "loss": 0.5767,
+      "step": 12409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207489099769172e-05,
+      "loss": 0.5535,
+      "step": 12410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2074249807642986e-05,
+      "loss": 0.5768,
+      "step": 12411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207360861759426e-05,
+      "loss": 0.5462,
+      "step": 12412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207296742754553e-05,
+      "loss": 0.4989,
+      "step": 12413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2072326237496795e-05,
+      "loss": 0.4899,
+      "step": 12414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2071685047448066e-05,
+      "loss": 0.4617,
+      "step": 12415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207104385739933e-05,
+      "loss": 0.4662,
+      "step": 12416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20704026673506e-05,
+      "loss": 0.5623,
+      "step": 12417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2069761477301875e-05,
+      "loss": 0.4798,
+      "step": 12418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2069120287253147e-05,
+      "loss": 0.5607,
+      "step": 12419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206847909720441e-05,
+      "loss": 0.5453,
+      "step": 12420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206783790715568e-05,
+      "loss": 0.4799,
+      "step": 12421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2067196717106955e-05,
+      "loss": 0.537,
+      "step": 12422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206655552705822e-05,
+      "loss": 0.5861,
+      "step": 12423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206591433700949e-05,
+      "loss": 0.5242,
+      "step": 12424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2065273146960756e-05,
+      "loss": 0.479,
+      "step": 12425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2064631956912035e-05,
+      "loss": 0.698,
+      "step": 12426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20639907668633e-05,
+      "loss": 0.5487,
+      "step": 12427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206334957681457e-05,
+      "loss": 0.5773,
+      "step": 12428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2062708386765837e-05,
+      "loss": 0.5434,
+      "step": 12429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206206719671711e-05,
+      "loss": 0.5888,
+      "step": 12430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206142600666838e-05,
+      "loss": 0.4912,
+      "step": 12431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2060784816619645e-05,
+      "loss": 0.604,
+      "step": 12432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206014362657092e-05,
+      "loss": 0.5345,
+      "step": 12433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205950243652219e-05,
+      "loss": 0.5597,
+      "step": 12434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205886124647346e-05,
+      "loss": 0.5792,
+      "step": 12435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2058220056424725e-05,
+      "loss": 0.5995,
+      "step": 12436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2057578866376e-05,
+      "loss": 0.6262,
+      "step": 12437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205693767632726e-05,
+      "loss": 0.5436,
+      "step": 12438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2056296486278533e-05,
+      "loss": 0.6337,
+      "step": 12439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2055655296229805e-05,
+      "loss": 0.5081,
+      "step": 12440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205501410618107e-05,
+      "loss": 0.4923,
+      "step": 12441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205437291613235e-05,
+      "loss": 0.5526,
+      "step": 12442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2053731726083613e-05,
+      "loss": 0.5757,
+      "step": 12443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2053090536034885e-05,
+      "loss": 0.5214,
+      "step": 12444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205244934598615e-05,
+      "loss": 0.7004,
+      "step": 12445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205180815593742e-05,
+      "loss": 0.648,
+      "step": 12446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205116696588869e-05,
+      "loss": 0.5758,
+      "step": 12447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205052577583996e-05,
+      "loss": 0.443,
+      "step": 12448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204988458579123e-05,
+      "loss": 0.4642,
+      "step": 12449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20492433957425e-05,
+      "loss": 0.6328,
+      "step": 12450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2048602205693774e-05,
+      "loss": 0.5039,
+      "step": 12451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204796101564504e-05,
+      "loss": 0.4763,
+      "step": 12452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204731982559631e-05,
+      "loss": 0.5959,
+      "step": 12453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2046678635547575e-05,
+      "loss": 0.571,
+      "step": 12454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204603744549885e-05,
+      "loss": 0.5966,
+      "step": 12455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204539625545012e-05,
+      "loss": 0.5418,
+      "step": 12456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2044755065401384e-05,
+      "loss": 0.5661,
+      "step": 12457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2044113875352655e-05,
+      "loss": 0.474,
+      "step": 12458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204347268530393e-05,
+      "loss": 0.5644,
+      "step": 12459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20428314952552e-05,
+      "loss": 0.5976,
+      "step": 12460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2042190305206464e-05,
+      "loss": 0.5401,
+      "step": 12461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2041549115157735e-05,
+      "loss": 0.5042,
+      "step": 12462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2040907925109e-05,
+      "loss": 0.6832,
+      "step": 12463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204026673506027e-05,
+      "loss": 0.5869,
+      "step": 12464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2039625545011544e-05,
+      "loss": 0.5615,
+      "step": 12465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2038984354962816e-05,
+      "loss": 0.6085,
+      "step": 12466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203834316491409e-05,
+      "loss": 0.6825,
+      "step": 12467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203770197486535e-05,
+      "loss": 0.5656,
+      "step": 12468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2037060784816624e-05,
+      "loss": 0.4337,
+      "step": 12469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203641959476789e-05,
+      "loss": 0.5492,
+      "step": 12470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203577840471916e-05,
+      "loss": 0.5326,
+      "step": 12471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2035137214670426e-05,
+      "loss": 0.5117,
+      "step": 12472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20344960246217e-05,
+      "loss": 0.5699,
+      "step": 12473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203385483457297e-05,
+      "loss": 0.573,
+      "step": 12474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203321364452424e-05,
+      "loss": 0.633,
+      "step": 12475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203257245447551e-05,
+      "loss": 0.56,
+      "step": 12476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203193126442678e-05,
+      "loss": 0.5308,
+      "step": 12477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203129007437805e-05,
+      "loss": 0.5395,
+      "step": 12478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2030648884329314e-05,
+      "loss": 0.4798,
+      "step": 12479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2030007694280586e-05,
+      "loss": 0.501,
+      "step": 12480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202936650423186e-05,
+      "loss": 0.5759,
+      "step": 12481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202872531418313e-05,
+      "loss": 0.613,
+      "step": 12482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20280841241344e-05,
+      "loss": 0.4861,
+      "step": 12483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2027442934085666e-05,
+      "loss": 0.5782,
+      "step": 12484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202680174403694e-05,
+      "loss": 0.4559,
+      "step": 12485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20261605539882e-05,
+      "loss": 0.5213,
+      "step": 12486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2025519363939474e-05,
+      "loss": 0.4805,
+      "step": 12487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202487817389074e-05,
+      "loss": 0.5608,
+      "step": 12488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202423698384201e-05,
+      "loss": 0.5826,
+      "step": 12489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202359579379328e-05,
+      "loss": 0.5008,
+      "step": 12490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2022954603744554e-05,
+      "loss": 0.556,
+      "step": 12491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2022313413695826e-05,
+      "loss": 0.57,
+      "step": 12492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202167222364709e-05,
+      "loss": 0.5534,
+      "step": 12493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202103103359836e-05,
+      "loss": 0.5438,
+      "step": 12494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202038984354963e-05,
+      "loss": 0.556,
+      "step": 12495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20197486535009e-05,
+      "loss": 0.6346,
+      "step": 12496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2019107463452164e-05,
+      "loss": 0.4868,
+      "step": 12497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201846627340344e-05,
+      "loss": 0.5643,
+      "step": 12498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201782508335471e-05,
+      "loss": 0.4755,
+      "step": 12499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201718389330598e-05,
+      "loss": 0.5911,
+      "step": 12500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201654270325725e-05,
+      "loss": 0.6122,
+      "step": 12501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2015901513208516e-05,
+      "loss": 0.5051,
+      "step": 12502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201526032315979e-05,
+      "loss": 0.6022,
+      "step": 12503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201461913311105e-05,
+      "loss": 0.5101,
+      "step": 12504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2013977943062324e-05,
+      "loss": 0.51,
+      "step": 12505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2013336753013596e-05,
+      "loss": 0.4906,
+      "step": 12506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201269556296487e-05,
+      "loss": 0.423,
+      "step": 12507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201205437291613e-05,
+      "loss": 0.5407,
+      "step": 12508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2011413182867404e-05,
+      "loss": 0.6137,
+      "step": 12509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2010771992818676e-05,
+      "loss": 0.5301,
+      "step": 12510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.201013080276994e-05,
+      "loss": 0.4944,
+      "step": 12511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200948961272121e-05,
+      "loss": 0.5227,
+      "step": 12512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200884842267248e-05,
+      "loss": 0.7119,
+      "step": 12513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2008207232623756e-05,
+      "loss": 0.4835,
+      "step": 12514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200756604257502e-05,
+      "loss": 0.6287,
+      "step": 12515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200692485252629e-05,
+      "loss": 0.6933,
+      "step": 12516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200628366247756e-05,
+      "loss": 0.5561,
+      "step": 12517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200564247242883e-05,
+      "loss": 0.5729,
+      "step": 12518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20050012823801e-05,
+      "loss": 0.5111,
+      "step": 12519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2004360092331366e-05,
+      "loss": 0.4003,
+      "step": 12520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200371890228264e-05,
+      "loss": 0.4293,
+      "step": 12521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200307771223391e-05,
+      "loss": 0.6479,
+      "step": 12522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200243652218518e-05,
+      "loss": 0.4819,
+      "step": 12523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2001795332136446e-05,
+      "loss": 0.6011,
+      "step": 12524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200115414208772e-05,
+      "loss": 0.4285,
+      "step": 12525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.200051295203898e-05,
+      "loss": 0.4767,
+      "step": 12526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1999871761990255e-05,
+      "loss": 0.4648,
+      "step": 12527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1999230571941526e-05,
+      "loss": 0.6184,
+      "step": 12528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199858938189279e-05,
+      "loss": 0.5733,
+      "step": 12529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199794819184406e-05,
+      "loss": 0.5873,
+      "step": 12530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1997307001795335e-05,
+      "loss": 0.5154,
+      "step": 12531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1996665811746607e-05,
+      "loss": 0.5726,
+      "step": 12532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199602462169787e-05,
+      "loss": 0.5555,
+      "step": 12533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199538343164914e-05,
+      "loss": 0.4518,
+      "step": 12534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199474224160041e-05,
+      "loss": 0.5006,
+      "step": 12535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199410105155168e-05,
+      "loss": 0.5557,
+      "step": 12536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199345986150295e-05,
+      "loss": 0.5699,
+      "step": 12537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199281867145422e-05,
+      "loss": 0.5042,
+      "step": 12538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1992177481405495e-05,
+      "loss": 0.5144,
+      "step": 12539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199153629135676e-05,
+      "loss": 0.5678,
+      "step": 12540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.199089510130803e-05,
+      "loss": 0.4623,
+      "step": 12541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1990253911259297e-05,
+      "loss": 0.5198,
+      "step": 12542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198961272121057e-05,
+      "loss": 0.578,
+      "step": 12543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198897153116183e-05,
+      "loss": 0.5196,
+      "step": 12544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1988330341113105e-05,
+      "loss": 0.5772,
+      "step": 12545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198768915106438e-05,
+      "loss": 0.5527,
+      "step": 12546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198704796101565e-05,
+      "loss": 0.4631,
+      "step": 12547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198640677096692e-05,
+      "loss": 0.6901,
+      "step": 12548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1985765580918185e-05,
+      "loss": 0.4901,
+      "step": 12549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198512439086946e-05,
+      "loss": 0.6017,
+      "step": 12550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198448320082072e-05,
+      "loss": 0.6677,
+      "step": 12551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1983842010771993e-05,
+      "loss": 0.6808,
+      "step": 12552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1983200820723265e-05,
+      "loss": 0.545,
+      "step": 12553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198255963067454e-05,
+      "loss": 0.5778,
+      "step": 12554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198191844062581e-05,
+      "loss": 0.5561,
+      "step": 12555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1981277250577074e-05,
+      "loss": 0.6532,
+      "step": 12556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1980636060528345e-05,
+      "loss": 0.5537,
+      "step": 12557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197999487047961e-05,
+      "loss": 0.577,
+      "step": 12558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197935368043088e-05,
+      "loss": 0.5563,
+      "step": 12559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197871249038215e-05,
+      "loss": 0.5337,
+      "step": 12560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197807130033342e-05,
+      "loss": 0.5664,
+      "step": 12561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197743011028469e-05,
+      "loss": 0.617,
+      "step": 12562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197678892023596e-05,
+      "loss": 0.5628,
+      "step": 12563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1976147730187234e-05,
+      "loss": 0.6059,
+      "step": 12564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.19755065401385e-05,
+      "loss": 0.5945,
+      "step": 12565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197486535008977e-05,
+      "loss": 0.678,
+      "step": 12566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1974224160041035e-05,
+      "loss": 0.6038,
+      "step": 12567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197358296999231e-05,
+      "loss": 0.5828,
+      "step": 12568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197294177994358e-05,
+      "loss": 0.621,
+      "step": 12569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197230058989485e-05,
+      "loss": 0.5974,
+      "step": 12570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1971659399846115e-05,
+      "loss": 0.4678,
+      "step": 12571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197101820979739e-05,
+      "loss": 0.5057,
+      "step": 12572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.197037701974866e-05,
+      "loss": 0.5845,
+      "step": 12573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1969735829699924e-05,
+      "loss": 0.6281,
+      "step": 12574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1969094639651195e-05,
+      "loss": 0.5538,
+      "step": 12575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196845344960246e-05,
+      "loss": 0.5802,
+      "step": 12576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196781225955373e-05,
+      "loss": 0.5324,
+      "step": 12577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1967171069505004e-05,
+      "loss": 0.609,
+      "step": 12578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1966529879456276e-05,
+      "loss": 0.598,
+      "step": 12579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196588868940754e-05,
+      "loss": 0.6565,
+      "step": 12580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196524749935881e-05,
+      "loss": 0.519,
+      "step": 12581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1964606309310084e-05,
+      "loss": 0.5275,
+      "step": 12582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196396511926135e-05,
+      "loss": 0.5645,
+      "step": 12583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196332392921262e-05,
+      "loss": 0.4953,
+      "step": 12584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1962682739163886e-05,
+      "loss": 0.5902,
+      "step": 12585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1962041549115164e-05,
+      "loss": 0.5894,
+      "step": 12586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196140035906643e-05,
+      "loss": 0.5109,
+      "step": 12587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.19607591690177e-05,
+      "loss": 0.6071,
+      "step": 12588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1960117978968966e-05,
+      "loss": 0.5331,
+      "step": 12589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.195947678892024e-05,
+      "loss": 0.4763,
+      "step": 12590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.195883559887151e-05,
+      "loss": 0.6596,
+      "step": 12591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1958194408822774e-05,
+      "loss": 0.5173,
+      "step": 12592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1957553218774046e-05,
+      "loss": 0.5015,
+      "step": 12593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.195691202872532e-05,
+      "loss": 0.5453,
+      "step": 12594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.195627083867659e-05,
+      "loss": 0.539,
+      "step": 12595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1955629648627854e-05,
+      "loss": 0.5369,
+      "step": 12596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1954988458579126e-05,
+      "loss": 0.6236,
+      "step": 12597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.195434726853039e-05,
+      "loss": 0.5233,
+      "step": 12598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.195370607848166e-05,
+      "loss": 0.4461,
+      "step": 12599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1953064888432934e-05,
+      "loss": 0.5283,
+      "step": 12600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.19524236983842e-05,
+      "loss": 0.5316,
+      "step": 12601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.195178250833548e-05,
+      "loss": 0.6468,
+      "step": 12602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.195114131828674e-05,
+      "loss": 0.5617,
+      "step": 12603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1950500128238014e-05,
+      "loss": 0.5049,
+      "step": 12604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194985893818928e-05,
+      "loss": 0.6082,
+      "step": 12605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194921774814055e-05,
+      "loss": 0.4425,
+      "step": 12606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1948576558091816e-05,
+      "loss": 0.4867,
+      "step": 12607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194793536804309e-05,
+      "loss": 0.6075,
+      "step": 12608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194729417799436e-05,
+      "loss": 0.5384,
+      "step": 12609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194665298794563e-05,
+      "loss": 0.6104,
+      "step": 12610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.19460117978969e-05,
+      "loss": 0.4917,
+      "step": 12611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194537060784817e-05,
+      "loss": 0.563,
+      "step": 12612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194472941779944e-05,
+      "loss": 0.4954,
+      "step": 12613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1944088227750704e-05,
+      "loss": 0.5153,
+      "step": 12614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1943447037701976e-05,
+      "loss": 0.4423,
+      "step": 12615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194280584765324e-05,
+      "loss": 0.5892,
+      "step": 12616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194216465760451e-05,
+      "loss": 0.5486,
+      "step": 12617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1941523467555784e-05,
+      "loss": 0.4446,
+      "step": 12618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1940882277507056e-05,
+      "loss": 0.4511,
+      "step": 12619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194024108745833e-05,
+      "loss": 0.5866,
+      "step": 12620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193959989740959e-05,
+      "loss": 0.5393,
+      "step": 12621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1938958707360865e-05,
+      "loss": 0.4939,
+      "step": 12622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193831751731213e-05,
+      "loss": 0.5167,
+      "step": 12623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.19376763272634e-05,
+      "loss": 0.5715,
+      "step": 12624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193703513721467e-05,
+      "loss": 0.4915,
+      "step": 12625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1936393947165945e-05,
+      "loss": 0.6603,
+      "step": 12626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1935752757117216e-05,
+      "loss": 0.5756,
+      "step": 12627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193511156706848e-05,
+      "loss": 0.5537,
+      "step": 12628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193447037701975e-05,
+      "loss": 0.4528,
+      "step": 12629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193382918697102e-05,
+      "loss": 0.4852,
+      "step": 12630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193318799692229e-05,
+      "loss": 0.5473,
+      "step": 12631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1932546806873555e-05,
+      "loss": 0.5856,
+      "step": 12632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1931905616824826e-05,
+      "loss": 0.5668,
+      "step": 12633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.19312644267761e-05,
+      "loss": 0.5007,
+      "step": 12634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193062323672737e-05,
+      "loss": 0.5687,
+      "step": 12635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192998204667864e-05,
+      "loss": 0.5294,
+      "step": 12636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1929340856629906e-05,
+      "loss": 0.5132,
+      "step": 12637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192869966658118e-05,
+      "loss": 0.566,
+      "step": 12638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192805847653244e-05,
+      "loss": 0.5521,
+      "step": 12639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1927417286483715e-05,
+      "loss": 0.5363,
+      "step": 12640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1926776096434986e-05,
+      "loss": 0.5001,
+      "step": 12641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192613490638626e-05,
+      "loss": 0.45,
+      "step": 12642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192549371633753e-05,
+      "loss": 0.5504,
+      "step": 12643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1924852526288795e-05,
+      "loss": 0.5751,
+      "step": 12644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1924211336240067e-05,
+      "loss": 0.4559,
+      "step": 12645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192357014619133e-05,
+      "loss": 0.6372,
+      "step": 12646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.19229289561426e-05,
+      "loss": 0.4999,
+      "step": 12647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192228776609387e-05,
+      "loss": 0.534,
+      "step": 12648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192164657604514e-05,
+      "loss": 0.5888,
+      "step": 12649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192100538599641e-05,
+      "loss": 0.4246,
+      "step": 12650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192036419594768e-05,
+      "loss": 0.6861,
+      "step": 12651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1919723005898955e-05,
+      "loss": 0.6271,
+      "step": 12652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191908181585022e-05,
+      "loss": 0.5554,
+      "step": 12653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191844062580149e-05,
+      "loss": 0.5679,
+      "step": 12654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191779943575276e-05,
+      "loss": 0.5957,
+      "step": 12655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191715824570403e-05,
+      "loss": 0.462,
+      "step": 12656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191651705565529e-05,
+      "loss": 0.4477,
+      "step": 12657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191587586560657e-05,
+      "loss": 0.6565,
+      "step": 12658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191523467555784e-05,
+      "loss": 0.5765,
+      "step": 12659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191459348550911e-05,
+      "loss": 0.6415,
+      "step": 12660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191395229546038e-05,
+      "loss": 0.5816,
+      "step": 12661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1913311105411645e-05,
+      "loss": 0.6732,
+      "step": 12662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191266991536292e-05,
+      "loss": 0.4366,
+      "step": 12663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191202872531418e-05,
+      "loss": 0.6171,
+      "step": 12664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1911387535265453e-05,
+      "loss": 0.6257,
+      "step": 12665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1910746345216725e-05,
+      "loss": 0.5655,
+      "step": 12666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1910105155168e-05,
+      "loss": 0.5822,
+      "step": 12667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190946396511926e-05,
+      "loss": 0.5898,
+      "step": 12668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1908822775070534e-05,
+      "loss": 0.5193,
+      "step": 12669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1908181585021805e-05,
+      "loss": 0.5374,
+      "step": 12670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190754039497307e-05,
+      "loss": 0.5841,
+      "step": 12671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190689920492434e-05,
+      "loss": 0.5248,
+      "step": 12672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190625801487561e-05,
+      "loss": 0.516,
+      "step": 12673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1905616824826885e-05,
+      "loss": 0.5765,
+      "step": 12674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190497563477815e-05,
+      "loss": 0.6361,
+      "step": 12675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190433444472942e-05,
+      "loss": 0.517,
+      "step": 12676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190369325468069e-05,
+      "loss": 0.5153,
+      "step": 12677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190305206463196e-05,
+      "loss": 0.4602,
+      "step": 12678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190241087458323e-05,
+      "loss": 0.5514,
+      "step": 12679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1901769684534495e-05,
+      "loss": 0.5636,
+      "step": 12680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190112849448577e-05,
+      "loss": 0.5801,
+      "step": 12681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190048730443704e-05,
+      "loss": 0.5243,
+      "step": 12682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189984611438831e-05,
+      "loss": 0.5027,
+      "step": 12683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1899204924339575e-05,
+      "loss": 0.5609,
+      "step": 12684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189856373429085e-05,
+      "loss": 0.586,
+      "step": 12685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189792254424211e-05,
+      "loss": 0.4124,
+      "step": 12686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1897281354193384e-05,
+      "loss": 0.4539,
+      "step": 12687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1896640164144656e-05,
+      "loss": 0.4993,
+      "step": 12688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189599897409592e-05,
+      "loss": 0.59,
+      "step": 12689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18953577840472e-05,
+      "loss": 0.5296,
+      "step": 12690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1894716593998464e-05,
+      "loss": 0.4846,
+      "step": 12691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1894075403949736e-05,
+      "loss": 0.4759,
+      "step": 12692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1893434213901e-05,
+      "loss": 0.612,
+      "step": 12693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189279302385227e-05,
+      "loss": 0.6812,
+      "step": 12694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189215183380354e-05,
+      "loss": 0.5749,
+      "step": 12695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189151064375481e-05,
+      "loss": 0.4672,
+      "step": 12696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189086945370608e-05,
+      "loss": 0.5021,
+      "step": 12697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189022826365735e-05,
+      "loss": 0.4717,
+      "step": 12698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1889587073608624e-05,
+      "loss": 0.516,
+      "step": 12699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188894588355989e-05,
+      "loss": 0.4953,
+      "step": 12700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188830469351116e-05,
+      "loss": 0.5508,
+      "step": 12701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1887663503462426e-05,
+      "loss": 0.6251,
+      "step": 12702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18870223134137e-05,
+      "loss": 0.5181,
+      "step": 12703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188638112336496e-05,
+      "loss": 0.5387,
+      "step": 12704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1885739933316234e-05,
+      "loss": 0.4928,
+      "step": 12705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1885098743267506e-05,
+      "loss": 0.7365,
+      "step": 12706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188445755321878e-05,
+      "loss": 0.5597,
+      "step": 12707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188381636317005e-05,
+      "loss": 0.5983,
+      "step": 12708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1883175173121314e-05,
+      "loss": 0.5265,
+      "step": 12709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1882533983072586e-05,
+      "loss": 0.5406,
+      "step": 12710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188189279302385e-05,
+      "loss": 0.5253,
+      "step": 12711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188125160297512e-05,
+      "loss": 0.5856,
+      "step": 12712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1880610412926394e-05,
+      "loss": 0.569,
+      "step": 12713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1879969222877666e-05,
+      "loss": 0.487,
+      "step": 12714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187932803282894e-05,
+      "loss": 0.5671,
+      "step": 12715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18786868427802e-05,
+      "loss": 0.546,
+      "step": 12716
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1878045652731474e-05,
+      "loss": 0.5051,
+      "step": 12717
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187740446268274e-05,
+      "loss": 0.6141,
+      "step": 12718
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187676327263401e-05,
+      "loss": 0.5938,
+      "step": 12719
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1876122082585276e-05,
+      "loss": 0.5677,
+      "step": 12720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187548089253655e-05,
+      "loss": 0.555,
+      "step": 12721
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187483970248782e-05,
+      "loss": 0.5793,
+      "step": 12722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187419851243909e-05,
+      "loss": 0.5065,
+      "step": 12723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187355732239036e-05,
+      "loss": 0.576,
+      "step": 12724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187291613234163e-05,
+      "loss": 0.5304,
+      "step": 12725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18722749422929e-05,
+      "loss": 0.5327,
+      "step": 12726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1871633752244164e-05,
+      "loss": 0.552,
+      "step": 12727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1870992562195436e-05,
+      "loss": 0.5855,
+      "step": 12728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187035137214671e-05,
+      "loss": 0.4428,
+      "step": 12729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186971018209798e-05,
+      "loss": 0.4779,
+      "step": 12730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1869068992049244e-05,
+      "loss": 0.5029,
+      "step": 12731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1868427802000516e-05,
+      "loss": 0.537,
+      "step": 12732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186778661195179e-05,
+      "loss": 0.503,
+      "step": 12733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186714542190305e-05,
+      "loss": 0.4692,
+      "step": 12734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1866504231854325e-05,
+      "loss": 0.5221,
+      "step": 12735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186586304180559e-05,
+      "loss": 0.7074,
+      "step": 12736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186522185175686e-05,
+      "loss": 0.5617,
+      "step": 12737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186458066170813e-05,
+      "loss": 0.5833,
+      "step": 12738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1863939471659405e-05,
+      "loss": 0.5841,
+      "step": 12739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186329828161067e-05,
+      "loss": 0.5668,
+      "step": 12740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186265709156194e-05,
+      "loss": 0.5435,
+      "step": 12741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186201590151321e-05,
+      "loss": 0.5235,
+      "step": 12742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186137471146448e-05,
+      "loss": 0.4221,
+      "step": 12743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186073352141575e-05,
+      "loss": 0.5832,
+      "step": 12744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1860092331367015e-05,
+      "loss": 0.4941,
+      "step": 12745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185945114131829e-05,
+      "loss": 0.5817,
+      "step": 12746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185880995126956e-05,
+      "loss": 0.6827,
+      "step": 12747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185816876122083e-05,
+      "loss": 0.5367,
+      "step": 12748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1857527571172095e-05,
+      "loss": 0.4222,
+      "step": 12749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1856886381123366e-05,
+      "loss": 0.589,
+      "step": 12750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185624519107464e-05,
+      "loss": 0.4973,
+      "step": 12751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18556040010259e-05,
+      "loss": 0.4923,
+      "step": 12752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1854962810977175e-05,
+      "loss": 0.5777,
+      "step": 12753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1854321620928447e-05,
+      "loss": 0.4868,
+      "step": 12754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185368043087972e-05,
+      "loss": 0.4954,
+      "step": 12755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185303924083098e-05,
+      "loss": 0.6656,
+      "step": 12756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1852398050782255e-05,
+      "loss": 0.5853,
+      "step": 12757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185175686073352e-05,
+      "loss": 0.5817,
+      "step": 12758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185111567068479e-05,
+      "loss": 0.5248,
+      "step": 12759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185047448063606e-05,
+      "loss": 0.4831,
+      "step": 12760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184983329058733e-05,
+      "loss": 0.6435,
+      "step": 12761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184919210053861e-05,
+      "loss": 0.5112,
+      "step": 12762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184855091048987e-05,
+      "loss": 0.4387,
+      "step": 12763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184790972044114e-05,
+      "loss": 0.4961,
+      "step": 12764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184726853039241e-05,
+      "loss": 0.5167,
+      "step": 12765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184662734034368e-05,
+      "loss": 0.5408,
+      "step": 12766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1845986150294945e-05,
+      "loss": 0.5668,
+      "step": 12767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184534496024622e-05,
+      "loss": 0.6723,
+      "step": 12768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184470377019749e-05,
+      "loss": 0.5877,
+      "step": 12769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184406258014876e-05,
+      "loss": 0.4503,
+      "step": 12770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184342139010003e-05,
+      "loss": 0.5289,
+      "step": 12771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18427802000513e-05,
+      "loss": 0.4868,
+      "step": 12772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184213901000257e-05,
+      "loss": 0.4722,
+      "step": 12773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1841497819953833e-05,
+      "loss": 0.4573,
+      "step": 12774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1840856629905105e-05,
+      "loss": 0.6117,
+      "step": 12775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.184021543985637e-05,
+      "loss": 0.6844,
+      "step": 12776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183957424980764e-05,
+      "loss": 0.5371,
+      "step": 12777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1838933059758914e-05,
+      "loss": 0.5578,
+      "step": 12778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1838291869710185e-05,
+      "loss": 0.4895,
+      "step": 12779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183765067966146e-05,
+      "loss": 0.6065,
+      "step": 12780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183700948961272e-05,
+      "loss": 0.5022,
+      "step": 12781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1836368299563994e-05,
+      "loss": 0.5219,
+      "step": 12782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183572710951526e-05,
+      "loss": 0.5188,
+      "step": 12783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183508591946653e-05,
+      "loss": 0.4612,
+      "step": 12784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18344447294178e-05,
+      "loss": 0.4863,
+      "step": 12785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1833803539369074e-05,
+      "loss": 0.5703,
+      "step": 12786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1833162349320345e-05,
+      "loss": 0.6066,
+      "step": 12787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183252115927161e-05,
+      "loss": 0.4712,
+      "step": 12788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183187996922288e-05,
+      "loss": 0.6362,
+      "step": 12789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183123877917415e-05,
+      "loss": 0.5355,
+      "step": 12790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183059758912542e-05,
+      "loss": 0.5119,
+      "step": 12791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1829956399076684e-05,
+      "loss": 0.5742,
+      "step": 12792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1829315209027955e-05,
+      "loss": 0.61,
+      "step": 12793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182867401897923e-05,
+      "loss": 0.7572,
+      "step": 12794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18280328289305e-05,
+      "loss": 0.5627,
+      "step": 12795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182739163888177e-05,
+      "loss": 0.5247,
+      "step": 12796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1826750448833035e-05,
+      "loss": 0.532,
+      "step": 12797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182610925878431e-05,
+      "loss": 0.5735,
+      "step": 12798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182546806873557e-05,
+      "loss": 0.4582,
+      "step": 12799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1824826878686844e-05,
+      "loss": 0.5449,
+      "step": 12800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1824185688638116e-05,
+      "loss": 0.6108,
+      "step": 12801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182354449858939e-05,
+      "loss": 0.5189,
+      "step": 12802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182290330854065e-05,
+      "loss": 0.4757,
+      "step": 12803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1822262118491924e-05,
+      "loss": 0.5657,
+      "step": 12804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1821620928443196e-05,
+      "loss": 0.5706,
+      "step": 12805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182097973839446e-05,
+      "loss": 0.4918,
+      "step": 12806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182033854834573e-05,
+      "loss": 0.5354,
+      "step": 12807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1819697358297e-05,
+      "loss": 0.5175,
+      "step": 12808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181905616824827e-05,
+      "loss": 0.5047,
+      "step": 12809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181841497819954e-05,
+      "loss": 0.527,
+      "step": 12810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181777378815081e-05,
+      "loss": 0.4653,
+      "step": 12811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181713259810208e-05,
+      "loss": 0.496,
+      "step": 12812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181649140805335e-05,
+      "loss": 0.4777,
+      "step": 12813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181585021800462e-05,
+      "loss": 0.5567,
+      "step": 12814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1815209027955886e-05,
+      "loss": 0.5187,
+      "step": 12815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181456783790716e-05,
+      "loss": 0.5495,
+      "step": 12816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181392664785842e-05,
+      "loss": 0.548,
+      "step": 12817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18132854578097e-05,
+      "loss": 0.6947,
+      "step": 12818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1812644267760966e-05,
+      "loss": 0.479,
+      "step": 12819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.181200307771224e-05,
+      "loss": 0.4702,
+      "step": 12820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18113618876635e-05,
+      "loss": 0.5823,
+      "step": 12821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1810720697614774e-05,
+      "loss": 0.6175,
+      "step": 12822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1810079507566046e-05,
+      "loss": 0.4953,
+      "step": 12823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180943831751731e-05,
+      "loss": 0.5339,
+      "step": 12824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180879712746858e-05,
+      "loss": 0.4323,
+      "step": 12825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1808155937419854e-05,
+      "loss": 0.6503,
+      "step": 12826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1807514747371126e-05,
+      "loss": 0.5574,
+      "step": 12827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180687355732239e-05,
+      "loss": 0.561,
+      "step": 12828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180623236727366e-05,
+      "loss": 0.5008,
+      "step": 12829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1805591177224934e-05,
+      "loss": 0.483,
+      "step": 12830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18049499871762e-05,
+      "loss": 0.7044,
+      "step": 12831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180430879712747e-05,
+      "loss": 0.5208,
+      "step": 12832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1803667607078736e-05,
+      "loss": 0.5509,
+      "step": 12833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1803026417030014e-05,
+      "loss": 0.3417,
+      "step": 12834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180238522698128e-05,
+      "loss": 0.5653,
+      "step": 12835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180174403693255e-05,
+      "loss": 0.6629,
+      "step": 12836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1801102846883816e-05,
+      "loss": 0.6457,
+      "step": 12837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180046165683509e-05,
+      "loss": 0.4722,
+      "step": 12838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179982046678636e-05,
+      "loss": 0.5164,
+      "step": 12839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1799179276737624e-05,
+      "loss": 0.5383,
+      "step": 12840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1798538086688896e-05,
+      "loss": 0.6425,
+      "step": 12841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179789689664017e-05,
+      "loss": 0.5812,
+      "step": 12842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179725570659144e-05,
+      "loss": 0.5338,
+      "step": 12843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1796614516542704e-05,
+      "loss": 0.5372,
+      "step": 12844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1795973326493976e-05,
+      "loss": 0.5507,
+      "step": 12845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179533213644524e-05,
+      "loss": 0.6367,
+      "step": 12846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179469094639651e-05,
+      "loss": 0.6109,
+      "step": 12847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1794049756347785e-05,
+      "loss": 0.5069,
+      "step": 12848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179340856629905e-05,
+      "loss": 0.5483,
+      "step": 12849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179276737625033e-05,
+      "loss": 0.4352,
+      "step": 12850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179212618620159e-05,
+      "loss": 0.5058,
+      "step": 12851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1791484996152865e-05,
+      "loss": 0.6761,
+      "step": 12852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179084380610413e-05,
+      "loss": 0.5455,
+      "step": 12853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.17902026160554e-05,
+      "loss": 0.5144,
+      "step": 12854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1789561426006666e-05,
+      "loss": 0.5477,
+      "step": 12855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178892023595794e-05,
+      "loss": 0.523,
+      "step": 12856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178827904590921e-05,
+      "loss": 0.5139,
+      "step": 12857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178763785586048e-05,
+      "loss": 0.5174,
+      "step": 12858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178699666581175e-05,
+      "loss": 0.4726,
+      "step": 12859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178635547576302e-05,
+      "loss": 0.4793,
+      "step": 12860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178571428571429e-05,
+      "loss": 0.5138,
+      "step": 12861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1785073095665555e-05,
+      "loss": 0.487,
+      "step": 12862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1784431905616826e-05,
+      "loss": 0.5834,
+      "step": 12863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178379071556809e-05,
+      "loss": 0.5173,
+      "step": 12864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178314952551936e-05,
+      "loss": 0.6262,
+      "step": 12865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1782508335470635e-05,
+      "loss": 0.566,
+      "step": 12866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1781867145421907e-05,
+      "loss": 0.4565,
+      "step": 12867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178122595537318e-05,
+      "loss": 0.5536,
+      "step": 12868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.178058476532444e-05,
+      "loss": 0.6524,
+      "step": 12869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1779943575275715e-05,
+      "loss": 0.4717,
+      "step": 12870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.177930238522698e-05,
+      "loss": 0.5238,
+      "step": 12871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.177866119517825e-05,
+      "loss": 0.4694,
+      "step": 12872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.177802000512952e-05,
+      "loss": 0.5186,
+      "step": 12873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1777378815080795e-05,
+      "loss": 0.5501,
+      "step": 12874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177673762503207e-05,
+      "loss": 0.5468,
+      "step": 12875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177609643498333e-05,
+      "loss": 0.5965,
+      "step": 12876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17754552449346e-05,
+      "loss": 0.4955,
+      "step": 12877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177481405488587e-05,
+      "loss": 0.6256,
+      "step": 12878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177417286483714e-05,
+      "loss": 0.5998,
+      "step": 12879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1773531674788405e-05,
+      "loss": 0.4673,
+      "step": 12880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177289048473968e-05,
+      "loss": 0.5082,
+      "step": 12881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177224929469095e-05,
+      "loss": 0.493,
+      "step": 12882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177160810464222e-05,
+      "loss": 0.4594,
+      "step": 12883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177096691459349e-05,
+      "loss": 0.5238,
+      "step": 12884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177032572454476e-05,
+      "loss": 0.5495,
+      "step": 12885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176968453449603e-05,
+      "loss": 0.568,
+      "step": 12886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1769043344447293e-05,
+      "loss": 0.5229,
+      "step": 12887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1768402154398565e-05,
+      "loss": 0.5344,
+      "step": 12888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176776096434984e-05,
+      "loss": 0.528,
+      "step": 12889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176711977430111e-05,
+      "loss": 0.5353,
+      "step": 12890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1766478584252374e-05,
+      "loss": 0.6542,
+      "step": 12891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1765837394203645e-05,
+      "loss": 0.5307,
+      "step": 12892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176519620415492e-05,
+      "loss": 0.544,
+      "step": 12893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176455501410618e-05,
+      "loss": 0.5521,
+      "step": 12894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1763913824057454e-05,
+      "loss": 0.576,
+      "step": 12895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176327263400872e-05,
+      "loss": 0.5257,
+      "step": 12896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176263144395999e-05,
+      "loss": 0.4757,
+      "step": 12897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176199025391126e-05,
+      "loss": 0.5523,
+      "step": 12898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1761349063862534e-05,
+      "loss": 0.5002,
+      "step": 12899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17607078738138e-05,
+      "loss": 0.6141,
+      "step": 12900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176006668376507e-05,
+      "loss": 0.5334,
+      "step": 12901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175942549371634e-05,
+      "loss": 0.6195,
+      "step": 12902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175878430366761e-05,
+      "loss": 0.5669,
+      "step": 12903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175814311361888e-05,
+      "loss": 0.5236,
+      "step": 12904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1757501923570144e-05,
+      "loss": 0.5576,
+      "step": 12905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175686073352142e-05,
+      "loss": 0.5933,
+      "step": 12906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175621954347269e-05,
+      "loss": 0.4587,
+      "step": 12907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175557835342396e-05,
+      "loss": 0.466,
+      "step": 12908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1754937163375224e-05,
+      "loss": 0.4703,
+      "step": 12909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1754295973326495e-05,
+      "loss": 0.5098,
+      "step": 12910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175365478327777e-05,
+      "loss": 0.5839,
+      "step": 12911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175301359322903e-05,
+      "loss": 0.5388,
+      "step": 12912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1752372403180304e-05,
+      "loss": 0.5129,
+      "step": 12913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1751731213131576e-05,
+      "loss": 0.643,
+      "step": 12914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175109002308285e-05,
+      "loss": 0.6929,
+      "step": 12915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175044883303411e-05,
+      "loss": 0.5525,
+      "step": 12916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1749807642985384e-05,
+      "loss": 0.489,
+      "step": 12917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174916645293665e-05,
+      "loss": 0.4997,
+      "step": 12918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174852526288792e-05,
+      "loss": 0.6618,
+      "step": 12919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174788407283919e-05,
+      "loss": 0.5219,
+      "step": 12920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174724288279046e-05,
+      "loss": 0.612,
+      "step": 12921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1746601692741736e-05,
+      "loss": 0.6051,
+      "step": 12922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1745960502693e-05,
+      "loss": 0.5795,
+      "step": 12923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174531931264427e-05,
+      "loss": 0.5907,
+      "step": 12924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174467812259554e-05,
+      "loss": 0.5139,
+      "step": 12925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174403693254681e-05,
+      "loss": 0.5615,
+      "step": 12926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1743395742498074e-05,
+      "loss": 0.516,
+      "step": 12927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1742754552449346e-05,
+      "loss": 0.577,
+      "step": 12928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174211336240062e-05,
+      "loss": 0.5963,
+      "step": 12929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174147217235189e-05,
+      "loss": 0.5544,
+      "step": 12930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174083098230316e-05,
+      "loss": 0.542,
+      "step": 12931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1740189792254426e-05,
+      "loss": 0.5256,
+      "step": 12932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17395486022057e-05,
+      "loss": 0.5966,
+      "step": 12933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173890741215696e-05,
+      "loss": 0.5652,
+      "step": 12934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1738266222108234e-05,
+      "loss": 0.4978,
+      "step": 12935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17376250320595e-05,
+      "loss": 0.4791,
+      "step": 12936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173698384201077e-05,
+      "loss": 0.5468,
+      "step": 12937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173634265196204e-05,
+      "loss": 0.4869,
+      "step": 12938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1735701461913314e-05,
+      "loss": 0.5161,
+      "step": 12939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1735060271864586e-05,
+      "loss": 0.5816,
+      "step": 12940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173441908181585e-05,
+      "loss": 0.5715,
+      "step": 12941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173377789176712e-05,
+      "loss": 0.4635,
+      "step": 12942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173313670171839e-05,
+      "loss": 0.5803,
+      "step": 12943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173249551166966e-05,
+      "loss": 0.4533,
+      "step": 12944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173185432162093e-05,
+      "loss": 0.6316,
+      "step": 12945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17312131315722e-05,
+      "loss": 0.5982,
+      "step": 12946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1730571941523474e-05,
+      "loss": 0.654,
+      "step": 12947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172993075147474e-05,
+      "loss": 0.5925,
+      "step": 12948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172928956142601e-05,
+      "loss": 0.6498,
+      "step": 12949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1728648371377276e-05,
+      "loss": 0.5761,
+      "step": 12950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172800718132855e-05,
+      "loss": 0.5533,
+      "step": 12951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172736599127981e-05,
+      "loss": 0.5736,
+      "step": 12952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1726724801231084e-05,
+      "loss": 0.4485,
+      "step": 12953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1726083611182356e-05,
+      "loss": 0.4917,
+      "step": 12954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172544242113363e-05,
+      "loss": 0.5596,
+      "step": 12955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17248012310849e-05,
+      "loss": 0.4968,
+      "step": 12956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1724160041036165e-05,
+      "loss": 0.4793,
+      "step": 12957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1723518850987436e-05,
+      "loss": 0.5781,
+      "step": 12958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17228776609387e-05,
+      "loss": 0.5408,
+      "step": 12959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172223647088997e-05,
+      "loss": 0.4407,
+      "step": 12960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1721595280841245e-05,
+      "loss": 0.5137,
+      "step": 12961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1720954090792516e-05,
+      "loss": 0.5081,
+      "step": 12962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172031290074378e-05,
+      "loss": 0.6407,
+      "step": 12963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171967171069505e-05,
+      "loss": 0.53,
+      "step": 12964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1719030520646325e-05,
+      "loss": 0.6911,
+      "step": 12965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171838933059759e-05,
+      "loss": 0.503,
+      "step": 12966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171774814054886e-05,
+      "loss": 0.5731,
+      "step": 12967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1717106950500126e-05,
+      "loss": 0.4614,
+      "step": 12968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17164657604514e-05,
+      "loss": 0.4651,
+      "step": 12969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171582457040267e-05,
+      "loss": 0.5203,
+      "step": 12970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171518338035394e-05,
+      "loss": 0.6492,
+      "step": 12971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1714542190305206e-05,
+      "loss": 0.5071,
+      "step": 12972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171390100025648e-05,
+      "loss": 0.5039,
+      "step": 12973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171325981020775e-05,
+      "loss": 0.4577,
+      "step": 12974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1712618620159015e-05,
+      "loss": 0.4899,
+      "step": 12975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1711977430110286e-05,
+      "loss": 0.5474,
+      "step": 12976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171133624006156e-05,
+      "loss": 0.531,
+      "step": 12977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171069505001283e-05,
+      "loss": 0.5222,
+      "step": 12978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1710053859964095e-05,
+      "loss": 0.6202,
+      "step": 12979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1709412669915367e-05,
+      "loss": 0.4873,
+      "step": 12980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170877147986663e-05,
+      "loss": 0.5297,
+      "step": 12981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17081302898179e-05,
+      "loss": 0.5618,
+      "step": 12982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1707489099769175e-05,
+      "loss": 0.6336,
+      "step": 12983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170684790972044e-05,
+      "loss": 0.4993,
+      "step": 12984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170620671967171e-05,
+      "loss": 0.611,
+      "step": 12985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170556552962298e-05,
+      "loss": 0.5657,
+      "step": 12986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1704924339574255e-05,
+      "loss": 0.6416,
+      "step": 12987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170428314952552e-05,
+      "loss": 0.6594,
+      "step": 12988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170364195947679e-05,
+      "loss": 0.5135,
+      "step": 12989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170300076942806e-05,
+      "loss": 0.4339,
+      "step": 12990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170235957937933e-05,
+      "loss": 0.5104,
+      "step": 12991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17017183893306e-05,
+      "loss": 0.554,
+      "step": 12992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1701077199281865e-05,
+      "loss": 0.6675,
+      "step": 12993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1700436009233144e-05,
+      "loss": 0.5739,
+      "step": 12994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169979481918441e-05,
+      "loss": 0.4475,
+      "step": 12995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169915362913568e-05,
+      "loss": 0.5406,
+      "step": 12996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1698512439086945e-05,
+      "loss": 0.5068,
+      "step": 12997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169787124903822e-05,
+      "loss": 0.5942,
+      "step": 12998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169723005898948e-05,
+      "loss": 0.5443,
+      "step": 12999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1696588868940753e-05,
+      "loss": 0.7364,
+      "step": 13000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1695947678892025e-05,
+      "loss": 0.5459,
+      "step": 13001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16953064888433e-05,
+      "loss": 0.5616,
+      "step": 13002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169466529879457e-05,
+      "loss": 0.5753,
+      "step": 13003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1694024108745834e-05,
+      "loss": 0.5252,
+      "step": 13004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1693382918697105e-05,
+      "loss": 0.6536,
+      "step": 13005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169274172864837e-05,
+      "loss": 0.5211,
+      "step": 13006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169210053859964e-05,
+      "loss": 0.4824,
+      "step": 13007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169145934855091e-05,
+      "loss": 0.5679,
+      "step": 13008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169081815850218e-05,
+      "loss": 0.5993,
+      "step": 13009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169017696845346e-05,
+      "loss": 0.5617,
+      "step": 13010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168953577840472e-05,
+      "loss": 0.6223,
+      "step": 13011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1688894588355994e-05,
+      "loss": 0.5891,
+      "step": 13012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168825339830726e-05,
+      "loss": 0.4513,
+      "step": 13013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168761220825853e-05,
+      "loss": 0.4068,
+      "step": 13014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1686971018209795e-05,
+      "loss": 0.5536,
+      "step": 13015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168632982816107e-05,
+      "loss": 0.5315,
+      "step": 13016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168568863811234e-05,
+      "loss": 0.5487,
+      "step": 13017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168504744806361e-05,
+      "loss": 0.4771,
+      "step": 13018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168440625801488e-05,
+      "loss": 0.6794,
+      "step": 13019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168376506796615e-05,
+      "loss": 0.5625,
+      "step": 13020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168312387791742e-05,
+      "loss": 0.6859,
+      "step": 13021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1682482687868684e-05,
+      "loss": 0.4762,
+      "step": 13022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1681841497819956e-05,
+      "loss": 0.5818,
+      "step": 13023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168120030777122e-05,
+      "loss": 0.6515,
+      "step": 13024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168055911772249e-05,
+      "loss": 0.5097,
+      "step": 13025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1679917927673764e-05,
+      "loss": 0.605,
+      "step": 13026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1679276737625036e-05,
+      "loss": 0.593,
+      "step": 13027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167863554757631e-05,
+      "loss": 0.5103,
+      "step": 13028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167799435752757e-05,
+      "loss": 0.5142,
+      "step": 13029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1677353167478844e-05,
+      "loss": 0.6449,
+      "step": 13030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167671197743011e-05,
+      "loss": 0.4398,
+      "step": 13031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167607078738138e-05,
+      "loss": 0.6279,
+      "step": 13032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167542959733265e-05,
+      "loss": 0.5675,
+      "step": 13033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1674788407283924e-05,
+      "loss": 0.628,
+      "step": 13034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1674147217235196e-05,
+      "loss": 0.5997,
+      "step": 13035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167350602718646e-05,
+      "loss": 0.5586,
+      "step": 13036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167286483713773e-05,
+      "loss": 0.5015,
+      "step": 13037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1672223647089e-05,
+      "loss": 0.6281,
+      "step": 13038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167158245704027e-05,
+      "loss": 0.5382,
+      "step": 13039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1670941266991534e-05,
+      "loss": 0.6194,
+      "step": 13040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1670300076942806e-05,
+      "loss": 0.4788,
+      "step": 13041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166965888689408e-05,
+      "loss": 0.506,
+      "step": 13042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166901769684535e-05,
+      "loss": 0.5788,
+      "step": 13043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166837650679662e-05,
+      "loss": 0.5648,
+      "step": 13044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1667735316747886e-05,
+      "loss": 0.5587,
+      "step": 13045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166709412669916e-05,
+      "loss": 0.573,
+      "step": 13046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166645293665042e-05,
+      "loss": 0.5958,
+      "step": 13047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1665811746601694e-05,
+      "loss": 0.5842,
+      "step": 13048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1665170556552966e-05,
+      "loss": 0.5646,
+      "step": 13049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166452936650424e-05,
+      "loss": 0.5496,
+      "step": 13050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16638881764555e-05,
+      "loss": 0.5847,
+      "step": 13051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1663246986406774e-05,
+      "loss": 0.5111,
+      "step": 13052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1662605796358046e-05,
+      "loss": 0.5017,
+      "step": 13053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166196460630931e-05,
+      "loss": 0.5624,
+      "step": 13054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166132341626058e-05,
+      "loss": 0.4698,
+      "step": 13055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166068222621185e-05,
+      "loss": 0.4701,
+      "step": 13056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.166004103616312e-05,
+      "loss": 0.5216,
+      "step": 13057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165939984611439e-05,
+      "loss": 0.5629,
+      "step": 13058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165875865606566e-05,
+      "loss": 0.6451,
+      "step": 13059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165811746601693e-05,
+      "loss": 0.4826,
+      "step": 13060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16574762759682e-05,
+      "loss": 0.6196,
+      "step": 13061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165683508591947e-05,
+      "loss": 0.6359,
+      "step": 13062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1656193895870736e-05,
+      "loss": 0.5082,
+      "step": 13063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165555270582201e-05,
+      "loss": 0.4867,
+      "step": 13064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165491151577327e-05,
+      "loss": 0.5397,
+      "step": 13065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165427032572455e-05,
+      "loss": 0.5618,
+      "step": 13066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1653629135675816e-05,
+      "loss": 0.6416,
+      "step": 13067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165298794562709e-05,
+      "loss": 0.4654,
+      "step": 13068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165234675557835e-05,
+      "loss": 0.6199,
+      "step": 13069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1651705565529625e-05,
+      "loss": 0.4616,
+      "step": 13070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1651064375480896e-05,
+      "loss": 0.6444,
+      "step": 13071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165042318543216e-05,
+      "loss": 0.5947,
+      "step": 13072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164978199538343e-05,
+      "loss": 0.6258,
+      "step": 13073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1649140805334705e-05,
+      "loss": 0.5212,
+      "step": 13074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1648499615285976e-05,
+      "loss": 0.5615,
+      "step": 13075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164785842523724e-05,
+      "loss": 0.4811,
+      "step": 13076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164721723518851e-05,
+      "loss": 0.5466,
+      "step": 13077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164657604513978e-05,
+      "loss": 0.5795,
+      "step": 13078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164593485509105e-05,
+      "loss": 0.636,
+      "step": 13079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164529366504232e-05,
+      "loss": 0.4425,
+      "step": 13080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1644652474993586e-05,
+      "loss": 0.5212,
+      "step": 13081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1644011284944865e-05,
+      "loss": 0.5164,
+      "step": 13082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164337009489613e-05,
+      "loss": 0.5415,
+      "step": 13083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16427289048474e-05,
+      "loss": 0.5345,
+      "step": 13084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1642087714798666e-05,
+      "loss": 0.4879,
+      "step": 13085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164144652474994e-05,
+      "loss": 0.4452,
+      "step": 13086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16408053347012e-05,
+      "loss": 0.492,
+      "step": 13087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1640164144652475e-05,
+      "loss": 0.6486,
+      "step": 13088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1639522954603747e-05,
+      "loss": 0.5373,
+      "step": 13089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163888176455502e-05,
+      "loss": 0.4852,
+      "step": 13090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163824057450629e-05,
+      "loss": 0.4556,
+      "step": 13091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1637599384457555e-05,
+      "loss": 0.493,
+      "step": 13092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1636958194408827e-05,
+      "loss": 0.5973,
+      "step": 13093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163631700436009e-05,
+      "loss": 0.5214,
+      "step": 13094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163567581431136e-05,
+      "loss": 0.5008,
+      "step": 13095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163503462426263e-05,
+      "loss": 0.5595,
+      "step": 13096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16343934342139e-05,
+      "loss": 0.6055,
+      "step": 13097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163375224416518e-05,
+      "loss": 0.516,
+      "step": 13098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163311105411644e-05,
+      "loss": 0.5879,
+      "step": 13099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1632469864067715e-05,
+      "loss": 0.6319,
+      "step": 13100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163182867401898e-05,
+      "loss": 0.4987,
+      "step": 13101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163118748397025e-05,
+      "loss": 0.5232,
+      "step": 13102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.163054629392152e-05,
+      "loss": 0.5983,
+      "step": 13103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162990510387279e-05,
+      "loss": 0.6152,
+      "step": 13104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162926391382406e-05,
+      "loss": 0.5112,
+      "step": 13105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162862272377533e-05,
+      "loss": 0.5469,
+      "step": 13106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1627981533726604e-05,
+      "loss": 0.4329,
+      "step": 13107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162734034367787e-05,
+      "loss": 0.7064,
+      "step": 13108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162669915362914e-05,
+      "loss": 0.5475,
+      "step": 13109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1626057963580405e-05,
+      "loss": 0.5503,
+      "step": 13110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162541677353168e-05,
+      "loss": 0.56,
+      "step": 13111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162477558348294e-05,
+      "loss": 0.5271,
+      "step": 13112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1624134393434214e-05,
+      "loss": 0.4435,
+      "step": 13113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1623493203385485e-05,
+      "loss": 0.4521,
+      "step": 13114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162285201333676e-05,
+      "loss": 0.5289,
+      "step": 13115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162221082328803e-05,
+      "loss": 0.5181,
+      "step": 13116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1621569633239294e-05,
+      "loss": 0.5076,
+      "step": 13117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1620928443190565e-05,
+      "loss": 0.6174,
+      "step": 13118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162028725314183e-05,
+      "loss": 0.7447,
+      "step": 13119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16196460630931e-05,
+      "loss": 0.5253,
+      "step": 13120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1619004873044374e-05,
+      "loss": 0.573,
+      "step": 13121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1618363682995645e-05,
+      "loss": 0.6337,
+      "step": 13122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161772249294691e-05,
+      "loss": 0.6291,
+      "step": 13123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161708130289818e-05,
+      "loss": 0.5717,
+      "step": 13124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1616440112849454e-05,
+      "loss": 0.5883,
+      "step": 13125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161579892280072e-05,
+      "loss": 0.5968,
+      "step": 13126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161515773275199e-05,
+      "loss": 0.5076,
+      "step": 13127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1614516542703255e-05,
+      "loss": 0.5595,
+      "step": 13128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161387535265453e-05,
+      "loss": 0.5011,
+      "step": 13129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16132341626058e-05,
+      "loss": 0.5863,
+      "step": 13130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161259297255707e-05,
+      "loss": 0.6104,
+      "step": 13131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1611951782508335e-05,
+      "loss": 0.5007,
+      "step": 13132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161131059245961e-05,
+      "loss": 0.5462,
+      "step": 13133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161066940241088e-05,
+      "loss": 0.6064,
+      "step": 13134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1610028212362144e-05,
+      "loss": 0.4944,
+      "step": 13135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1609387022313416e-05,
+      "loss": 0.59,
+      "step": 13136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160874583226469e-05,
+      "loss": 0.5402,
+      "step": 13137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160810464221596e-05,
+      "loss": 0.5326,
+      "step": 13138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1607463452167224e-05,
+      "loss": 0.4514,
+      "step": 13139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1606822262118496e-05,
+      "loss": 0.5482,
+      "step": 13140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160618107206976e-05,
+      "loss": 0.569,
+      "step": 13141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160553988202103e-05,
+      "loss": 0.4744,
+      "step": 13142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1604898691972304e-05,
+      "loss": 0.5793,
+      "step": 13143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160425750192357e-05,
+      "loss": 0.7002,
+      "step": 13144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160361631187484e-05,
+      "loss": 0.4753,
+      "step": 13145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160297512182611e-05,
+      "loss": 0.5583,
+      "step": 13146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1602333931777384e-05,
+      "loss": 0.5245,
+      "step": 13147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160169274172865e-05,
+      "loss": 0.5371,
+      "step": 13148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.160105155167992e-05,
+      "loss": 0.5542,
+      "step": 13149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1600410361631186e-05,
+      "loss": 0.5228,
+      "step": 13150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159976917158246e-05,
+      "loss": 0.6298,
+      "step": 13151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159912798153373e-05,
+      "loss": 0.5225,
+      "step": 13152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1598486791484994e-05,
+      "loss": 0.6283,
+      "step": 13153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159784560143627e-05,
+      "loss": 0.4784,
+      "step": 13154
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159720441138754e-05,
+      "loss": 0.5255,
+      "step": 13155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159656322133881e-05,
+      "loss": 0.5424,
+      "step": 13156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1595922031290074e-05,
+      "loss": 0.5824,
+      "step": 13157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1595280841241346e-05,
+      "loss": 0.4859,
+      "step": 13158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159463965119261e-05,
+      "loss": 0.5587,
+      "step": 13159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159399846114388e-05,
+      "loss": 0.5216,
+      "step": 13160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1593357271095154e-05,
+      "loss": 0.6098,
+      "step": 13161
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1592716081046426e-05,
+      "loss": 0.6317,
+      "step": 13162
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15920748909977e-05,
+      "loss": 0.447,
+      "step": 13163
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159143370094896e-05,
+      "loss": 0.5823,
+      "step": 13164
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1590792510900234e-05,
+      "loss": 0.5423,
+      "step": 13165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15901513208515e-05,
+      "loss": 0.6286,
+      "step": 13166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158951013080277e-05,
+      "loss": 0.4182,
+      "step": 13167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1588868940754036e-05,
+      "loss": 0.5793,
+      "step": 13168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158822775070531e-05,
+      "loss": 0.4317,
+      "step": 13169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1587586560656586e-05,
+      "loss": 0.5042,
+      "step": 13170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158694537060785e-05,
+      "loss": 0.5399,
+      "step": 13171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158630418055912e-05,
+      "loss": 0.4732,
+      "step": 13172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158566299051039e-05,
+      "loss": 0.5525,
+      "step": 13173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158502180046166e-05,
+      "loss": 0.5815,
+      "step": 13174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1584380610412924e-05,
+      "loss": 0.5488,
+      "step": 13175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1583739420364196e-05,
+      "loss": 0.4538,
+      "step": 13176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158309823031547e-05,
+      "loss": 0.5767,
+      "step": 13177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158245704026674e-05,
+      "loss": 0.4654,
+      "step": 13178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158181585021801e-05,
+      "loss": 0.5615,
+      "step": 13179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1581174660169276e-05,
+      "loss": 0.6393,
+      "step": 13180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158053347012055e-05,
+      "loss": 0.5185,
+      "step": 13181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157989228007181e-05,
+      "loss": 0.5808,
+      "step": 13182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1579251090023085e-05,
+      "loss": 0.5025,
+      "step": 13183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157860989997435e-05,
+      "loss": 0.6408,
+      "step": 13184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157796870992562e-05,
+      "loss": 0.582,
+      "step": 13185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157732751987689e-05,
+      "loss": 0.6433,
+      "step": 13186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1576686329828165e-05,
+      "loss": 0.5272,
+      "step": 13187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1576045139779436e-05,
+      "loss": 0.5881,
+      "step": 13188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15754039497307e-05,
+      "loss": 0.5043,
+      "step": 13189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157476275968197e-05,
+      "loss": 0.6382,
+      "step": 13190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157412156963324e-05,
+      "loss": 0.4768,
+      "step": 13191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157348037958451e-05,
+      "loss": 0.5976,
+      "step": 13192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157283918953578e-05,
+      "loss": 0.5001,
+      "step": 13193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157219799948705e-05,
+      "loss": 0.5361,
+      "step": 13194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157155680943832e-05,
+      "loss": 0.5319,
+      "step": 13195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157091561938959e-05,
+      "loss": 0.4934,
+      "step": 13196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157027442934086e-05,
+      "loss": 0.5428,
+      "step": 13197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1569633239292126e-05,
+      "loss": 0.4729,
+      "step": 13198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15689920492434e-05,
+      "loss": 0.4842,
+      "step": 13199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156835085919466e-05,
+      "loss": 0.5954,
+      "step": 13200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1567709669145935e-05,
+      "loss": 0.5759,
+      "step": 13201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1567068479097207e-05,
+      "loss": 0.5118,
+      "step": 13202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156642728904848e-05,
+      "loss": 0.4718,
+      "step": 13203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156578609899974e-05,
+      "loss": 0.5186,
+      "step": 13204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1565144908951015e-05,
+      "loss": 0.5312,
+      "step": 13205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156450371890229e-05,
+      "loss": 0.5364,
+      "step": 13206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156386252885355e-05,
+      "loss": 0.5479,
+      "step": 13207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156322133880482e-05,
+      "loss": 0.6735,
+      "step": 13208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1562580148756095e-05,
+      "loss": 0.4888,
+      "step": 13209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156193895870737e-05,
+      "loss": 0.5341,
+      "step": 13210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156129776865863e-05,
+      "loss": 0.5603,
+      "step": 13211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1560656578609903e-05,
+      "loss": 0.4675,
+      "step": 13212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1560015388561175e-05,
+      "loss": 0.5244,
+      "step": 13213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155937419851244e-05,
+      "loss": 0.6002,
+      "step": 13214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155873300846371e-05,
+      "loss": 0.5581,
+      "step": 13215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155809181841498e-05,
+      "loss": 0.5165,
+      "step": 13216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155745062836625e-05,
+      "loss": 0.6209,
+      "step": 13217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155680943831752e-05,
+      "loss": 0.5122,
+      "step": 13218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155616824826879e-05,
+      "loss": 0.5112,
+      "step": 13219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155552705822006e-05,
+      "loss": 0.529,
+      "step": 13220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155488586817133e-05,
+      "loss": 0.5325,
+      "step": 13221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15542446781226e-05,
+      "loss": 0.5902,
+      "step": 13222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1553603488073865e-05,
+      "loss": 0.6792,
+      "step": 13223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155296229802514e-05,
+      "loss": 0.6788,
+      "step": 13224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15523211079764e-05,
+      "loss": 0.5995,
+      "step": 13225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155167991792768e-05,
+      "loss": 0.4769,
+      "step": 13226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1551038727878945e-05,
+      "loss": 0.5272,
+      "step": 13227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155039753783022e-05,
+      "loss": 0.631,
+      "step": 13228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154975634778148e-05,
+      "loss": 0.5645,
+      "step": 13229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1549115157732754e-05,
+      "loss": 0.4865,
+      "step": 13230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1548473967684025e-05,
+      "loss": 0.6237,
+      "step": 13231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154783277763529e-05,
+      "loss": 0.5596,
+      "step": 13232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154719158758656e-05,
+      "loss": 0.5631,
+      "step": 13233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1546550397537834e-05,
+      "loss": 0.5487,
+      "step": 13234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1545909207489105e-05,
+      "loss": 0.524,
+      "step": 13235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154526801744037e-05,
+      "loss": 0.6427,
+      "step": 13236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154462682739164e-05,
+      "loss": 0.6846,
+      "step": 13237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154398563734291e-05,
+      "loss": 0.6009,
+      "step": 13238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154334444729418e-05,
+      "loss": 0.6549,
+      "step": 13239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154270325724545e-05,
+      "loss": 0.5728,
+      "step": 13240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1542062067196715e-05,
+      "loss": 0.568,
+      "step": 13241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1541420877147994e-05,
+      "loss": 0.584,
+      "step": 13242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154077968709926e-05,
+      "loss": 0.5776,
+      "step": 13243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154013849705053e-05,
+      "loss": 0.6098,
+      "step": 13244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1539497307001796e-05,
+      "loss": 0.598,
+      "step": 13245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153885611695307e-05,
+      "loss": 0.582,
+      "step": 13246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153821492690433e-05,
+      "loss": 0.5371,
+      "step": 13247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1537573736855604e-05,
+      "loss": 0.5373,
+      "step": 13248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1536932546806876e-05,
+      "loss": 0.5208,
+      "step": 13249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153629135675815e-05,
+      "loss": 0.5131,
+      "step": 13250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153565016670942e-05,
+      "loss": 0.5789,
+      "step": 13251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1535008976660684e-05,
+      "loss": 0.5303,
+      "step": 13252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1534367786611956e-05,
+      "loss": 0.5793,
+      "step": 13253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153372659656322e-05,
+      "loss": 0.5332,
+      "step": 13254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153308540651449e-05,
+      "loss": 0.5626,
+      "step": 13255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153244421646576e-05,
+      "loss": 0.5704,
+      "step": 13256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153180302641703e-05,
+      "loss": 0.5117,
+      "step": 13257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153116183636831e-05,
+      "loss": 0.4769,
+      "step": 13258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153052064631957e-05,
+      "loss": 0.5188,
+      "step": 13259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1529879456270844e-05,
+      "loss": 0.509,
+      "step": 13260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152923826622211e-05,
+      "loss": 0.595,
+      "step": 13261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152859707617338e-05,
+      "loss": 0.4577,
+      "step": 13262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1527955886124646e-05,
+      "loss": 0.6526,
+      "step": 13263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152731469607592e-05,
+      "loss": 0.5901,
+      "step": 13264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152667350602719e-05,
+      "loss": 0.6153,
+      "step": 13265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152603231597846e-05,
+      "loss": 0.4506,
+      "step": 13266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152539112592973e-05,
+      "loss": 0.5161,
+      "step": 13267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1524749935881e-05,
+      "loss": 0.5763,
+      "step": 13268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152410874583227e-05,
+      "loss": 0.5859,
+      "step": 13269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1523467555783534e-05,
+      "loss": 0.5646,
+      "step": 13270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1522826365734806e-05,
+      "loss": 0.5223,
+      "step": 13271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152218517568607e-05,
+      "loss": 0.6599,
+      "step": 13272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152154398563734e-05,
+      "loss": 0.6245,
+      "step": 13273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1520902795588614e-05,
+      "loss": 0.623,
+      "step": 13274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1520261605539886e-05,
+      "loss": 0.5075,
+      "step": 13275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151962041549116e-05,
+      "loss": 0.5332,
+      "step": 13276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151897922544242e-05,
+      "loss": 0.6047,
+      "step": 13277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1518338035393694e-05,
+      "loss": 0.5597,
+      "step": 13278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151769684534496e-05,
+      "loss": 0.5901,
+      "step": 13279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151705565529623e-05,
+      "loss": 0.5666,
+      "step": 13280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15164144652475e-05,
+      "loss": 0.5182,
+      "step": 13281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1515773275198774e-05,
+      "loss": 0.53,
+      "step": 13282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151513208515004e-05,
+      "loss": 0.5974,
+      "step": 13283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151449089510131e-05,
+      "loss": 0.4799,
+      "step": 13284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151384970505258e-05,
+      "loss": 0.5969,
+      "step": 13285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151320851500385e-05,
+      "loss": 0.5395,
+      "step": 13286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151256732495512e-05,
+      "loss": 0.5897,
+      "step": 13287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1511926134906384e-05,
+      "loss": 0.5669,
+      "step": 13288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1511284944857656e-05,
+      "loss": 0.5323,
+      "step": 13289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151064375480893e-05,
+      "loss": 0.5031,
+      "step": 13290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15100025647602e-05,
+      "loss": 0.5548,
+      "step": 13291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1509361374711465e-05,
+      "loss": 0.5512,
+      "step": 13292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1508720184662736e-05,
+      "loss": 0.5478,
+      "step": 13293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150807899461401e-05,
+      "loss": 0.4891,
+      "step": 13294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150743780456527e-05,
+      "loss": 0.6286,
+      "step": 13295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1506796614516545e-05,
+      "loss": 0.5071,
+      "step": 13296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1506155424467816e-05,
+      "loss": 0.4979,
+      "step": 13297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150551423441909e-05,
+      "loss": 0.6326,
+      "step": 13298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150487304437035e-05,
+      "loss": 0.5976,
+      "step": 13299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1504231854321625e-05,
+      "loss": 0.5545,
+      "step": 13300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150359066427289e-05,
+      "loss": 0.532,
+      "step": 13301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150294947422416e-05,
+      "loss": 0.6217,
+      "step": 13302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150230828417543e-05,
+      "loss": 0.4753,
+      "step": 13303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15016670941267e-05,
+      "loss": 0.5227,
+      "step": 13304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150102590407797e-05,
+      "loss": 0.5659,
+      "step": 13305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.150038471402924e-05,
+      "loss": 0.4978,
+      "step": 13306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149974352398051e-05,
+      "loss": 0.5955,
+      "step": 13307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149910233393178e-05,
+      "loss": 0.6348,
+      "step": 13308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149846114388305e-05,
+      "loss": 0.6607,
+      "step": 13309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1497819953834315e-05,
+      "loss": 0.5804,
+      "step": 13310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1497178763785586e-05,
+      "loss": 0.5003,
+      "step": 13311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149653757373686e-05,
+      "loss": 0.4883,
+      "step": 13312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149589638368812e-05,
+      "loss": 0.5334,
+      "step": 13313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.14952551936394e-05,
+      "loss": 0.4947,
+      "step": 13314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1494614003590667e-05,
+      "loss": 0.5305,
+      "step": 13315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149397281354194e-05,
+      "loss": 0.6269,
+      "step": 13316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.14933316234932e-05,
+      "loss": 0.5145,
+      "step": 13317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1492690433444475e-05,
+      "loss": 0.458,
+      "step": 13318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149204924339574e-05,
+      "loss": 0.5877,
+      "step": 13319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149140805334701e-05,
+      "loss": 0.5472,
+      "step": 13320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.149076686329828e-05,
+      "loss": 0.4859,
+      "step": 13321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1490125673249555e-05,
+      "loss": 0.6502,
+      "step": 13322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148948448320083e-05,
+      "loss": 0.5189,
+      "step": 13323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148884329315209e-05,
+      "loss": 0.5124,
+      "step": 13324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1488202103103363e-05,
+      "loss": 0.5993,
+      "step": 13325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148756091305463e-05,
+      "loss": 0.52,
+      "step": 13326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.14869197230059e-05,
+      "loss": 0.4692,
+      "step": 13327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1486278532957165e-05,
+      "loss": 0.505,
+      "step": 13328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148563734290844e-05,
+      "loss": 0.6179,
+      "step": 13329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1484996152859715e-05,
+      "loss": 0.6315,
+      "step": 13330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148435496281098e-05,
+      "loss": 0.6218,
+      "step": 13331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148371377276225e-05,
+      "loss": 0.6045,
+      "step": 13332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148307258271352e-05,
+      "loss": 0.5691,
+      "step": 13333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148243139266479e-05,
+      "loss": 0.6415,
+      "step": 13334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1481790202616053e-05,
+      "loss": 0.5144,
+      "step": 13335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1481149012567325e-05,
+      "loss": 0.551,
+      "step": 13336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.14805078225186e-05,
+      "loss": 0.5388,
+      "step": 13337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147986663246987e-05,
+      "loss": 0.5369,
+      "step": 13338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147922544242114e-05,
+      "loss": 0.5109,
+      "step": 13339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1478584252372405e-05,
+      "loss": 0.556,
+      "step": 13340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147794306232368e-05,
+      "loss": 0.5493,
+      "step": 13341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147730187227494e-05,
+      "loss": 0.5118,
+      "step": 13342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1476660682226214e-05,
+      "loss": 0.5557,
+      "step": 13343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147601949217748e-05,
+      "loss": 0.541,
+      "step": 13344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147537830212875e-05,
+      "loss": 0.5321,
+      "step": 13345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147473711208002e-05,
+      "loss": 0.5573,
+      "step": 13346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1474095922031294e-05,
+      "loss": 0.5225,
+      "step": 13347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1473454731982565e-05,
+      "loss": 0.4571,
+      "step": 13348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147281354193383e-05,
+      "loss": 0.5623,
+      "step": 13349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.14721723518851e-05,
+      "loss": 0.5345,
+      "step": 13350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147153116183637e-05,
+      "loss": 0.4836,
+      "step": 13351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147088997178764e-05,
+      "loss": 0.4548,
+      "step": 13352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147024878173891e-05,
+      "loss": 0.509,
+      "step": 13353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146960759169018e-05,
+      "loss": 0.5693,
+      "step": 13354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146896640164145e-05,
+      "loss": 0.5753,
+      "step": 13355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146832521159272e-05,
+      "loss": 0.4153,
+      "step": 13356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146768402154399e-05,
+      "loss": 0.646,
+      "step": 13357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1467042831495256e-05,
+      "loss": 0.5428,
+      "step": 13358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146640164144653e-05,
+      "loss": 0.518,
+      "step": 13359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146576045139779e-05,
+      "loss": 0.6475,
+      "step": 13360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1465119261349064e-05,
+      "loss": 0.5372,
+      "step": 13361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1464478071300336e-05,
+      "loss": 0.615,
+      "step": 13362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146383688125161e-05,
+      "loss": 0.6274,
+      "step": 13363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146319569120287e-05,
+      "loss": 0.645,
+      "step": 13364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1462554501154144e-05,
+      "loss": 0.4871,
+      "step": 13365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1461913311105416e-05,
+      "loss": 0.54,
+      "step": 13366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146127212105668e-05,
+      "loss": 0.6233,
+      "step": 13367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.146063093100795e-05,
+      "loss": 0.4047,
+      "step": 13368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1459989740959224e-05,
+      "loss": 0.5519,
+      "step": 13369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1459348550910496e-05,
+      "loss": 0.583,
+      "step": 13370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145870736086176e-05,
+      "loss": 0.5433,
+      "step": 13371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145806617081303e-05,
+      "loss": 0.5623,
+      "step": 13372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.14574249807643e-05,
+      "loss": 0.5672,
+      "step": 13373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145678379071557e-05,
+      "loss": 0.6667,
+      "step": 13374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145614260066684e-05,
+      "loss": 0.5463,
+      "step": 13375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1455501410618106e-05,
+      "loss": 0.5096,
+      "step": 13376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145486022056938e-05,
+      "loss": 0.5221,
+      "step": 13377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145421903052065e-05,
+      "loss": 0.5685,
+      "step": 13378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145357784047192e-05,
+      "loss": 0.5747,
+      "step": 13379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1452936650423186e-05,
+      "loss": 0.5864,
+      "step": 13380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145229546037446e-05,
+      "loss": 0.6063,
+      "step": 13381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.145165427032572e-05,
+      "loss": 0.5574,
+      "step": 13382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1451013080276994e-05,
+      "loss": 0.5485,
+      "step": 13383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1450371890228266e-05,
+      "loss": 0.4669,
+      "step": 13384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.144973070017954e-05,
+      "loss": 0.5691,
+      "step": 13385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.144908951013081e-05,
+      "loss": 0.5998,
+      "step": 13386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1448448320082074e-05,
+      "loss": 0.5978,
+      "step": 13387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1447807130033346e-05,
+      "loss": 0.5867,
+      "step": 13388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.144716593998461e-05,
+      "loss": 0.5183,
+      "step": 13389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.144652474993588e-05,
+      "loss": 0.5071,
+      "step": 13390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.144588355988715e-05,
+      "loss": 0.613,
+      "step": 13391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.144524236983842e-05,
+      "loss": 0.5133,
+      "step": 13392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.144460117978969e-05,
+      "loss": 0.4935,
+      "step": 13393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.144395998974096e-05,
+      "loss": 0.5578,
+      "step": 13394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1443318799692235e-05,
+      "loss": 0.532,
+      "step": 13395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.14426776096435e-05,
+      "loss": 0.5613,
+      "step": 13396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.144203641959477e-05,
+      "loss": 0.5212,
+      "step": 13397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1441395229546036e-05,
+      "loss": 0.61,
+      "step": 13398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.144075403949731e-05,
+      "loss": 0.5407,
+      "step": 13399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.144011284944858e-05,
+      "loss": 0.4929,
+      "step": 13400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1439471659399844e-05,
+      "loss": 0.553,
+      "step": 13401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143883046935112e-05,
+      "loss": 0.5393,
+      "step": 13402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143818927930239e-05,
+      "loss": 0.4884,
+      "step": 13403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143754808925366e-05,
+      "loss": 0.5629,
+      "step": 13404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1436906899204925e-05,
+      "loss": 0.5784,
+      "step": 13405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1436265709156196e-05,
+      "loss": 0.5393,
+      "step": 13406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143562451910746e-05,
+      "loss": 0.5259,
+      "step": 13407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143498332905873e-05,
+      "loss": 0.5714,
+      "step": 13408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1434342139010005e-05,
+      "loss": 0.589,
+      "step": 13409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1433700948961276e-05,
+      "loss": 0.6164,
+      "step": 13410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143305975891255e-05,
+      "loss": 0.5232,
+      "step": 13411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143241856886381e-05,
+      "loss": 0.5699,
+      "step": 13412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1431777378815085e-05,
+      "loss": 0.5153,
+      "step": 13413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143113618876635e-05,
+      "loss": 0.5896,
+      "step": 13414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143049499871762e-05,
+      "loss": 0.5595,
+      "step": 13415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1429853808668886e-05,
+      "loss": 0.5167,
+      "step": 13416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142921261862016e-05,
+      "loss": 0.5837,
+      "step": 13417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1428571428571437e-05,
+      "loss": 0.5412,
+      "step": 13418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.14279302385227e-05,
+      "loss": 0.59,
+      "step": 13419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142728904847397e-05,
+      "loss": 0.5103,
+      "step": 13420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142664785842524e-05,
+      "loss": 0.5684,
+      "step": 13421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142600666837651e-05,
+      "loss": 0.5054,
+      "step": 13422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1425365478327775e-05,
+      "loss": 0.4907,
+      "step": 13423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1424724288279047e-05,
+      "loss": 0.6558,
+      "step": 13424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142408309823032e-05,
+      "loss": 0.5662,
+      "step": 13425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142344190818159e-05,
+      "loss": 0.5505,
+      "step": 13426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142280071813286e-05,
+      "loss": 0.5618,
+      "step": 13427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142215952808413e-05,
+      "loss": 0.5265,
+      "step": 13428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.14215183380354e-05,
+      "loss": 0.4589,
+      "step": 13429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142087714798666e-05,
+      "loss": 0.5108,
+      "step": 13430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1420235957937935e-05,
+      "loss": 0.5164,
+      "step": 13431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.14195947678892e-05,
+      "loss": 0.4718,
+      "step": 13432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141895357784047e-05,
+      "loss": 0.4601,
+      "step": 13433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141831238779174e-05,
+      "loss": 0.4946,
+      "step": 13434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1417671197743015e-05,
+      "loss": 0.6225,
+      "step": 13435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141703000769429e-05,
+      "loss": 0.6665,
+      "step": 13436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141638881764555e-05,
+      "loss": 0.5091,
+      "step": 13437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1415747627596823e-05,
+      "loss": 0.6065,
+      "step": 13438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141510643754809e-05,
+      "loss": 0.4809,
+      "step": 13439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141446524749936e-05,
+      "loss": 0.549,
+      "step": 13440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141382405745063e-05,
+      "loss": 0.6257,
+      "step": 13441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1413182867401904e-05,
+      "loss": 0.4801,
+      "step": 13442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141254167735317e-05,
+      "loss": 0.4583,
+      "step": 13443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141190048730444e-05,
+      "loss": 0.5177,
+      "step": 13444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141125929725571e-05,
+      "loss": 0.5876,
+      "step": 13445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141061810720698e-05,
+      "loss": 0.5635,
+      "step": 13446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140997691715825e-05,
+      "loss": 0.454,
+      "step": 13447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1409335727109514e-05,
+      "loss": 0.5711,
+      "step": 13448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1408694537060785e-05,
+      "loss": 0.4838,
+      "step": 13449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140805334701206e-05,
+      "loss": 0.4614,
+      "step": 13450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140741215696333e-05,
+      "loss": 0.564,
+      "step": 13451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1406770966914594e-05,
+      "loss": 0.4145,
+      "step": 13452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1406129776865865e-05,
+      "loss": 0.5326,
+      "step": 13453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140548858681714e-05,
+      "loss": 0.5336,
+      "step": 13454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.14048473967684e-05,
+      "loss": 0.5267,
+      "step": 13455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1404206206719674e-05,
+      "loss": 0.5463,
+      "step": 13456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1403565016670945e-05,
+      "loss": 0.5689,
+      "step": 13457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140292382662222e-05,
+      "loss": 0.5351,
+      "step": 13458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140228263657348e-05,
+      "loss": 0.4585,
+      "step": 13459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1401641446524754e-05,
+      "loss": 0.5282,
+      "step": 13460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140100025647602e-05,
+      "loss": 0.563,
+      "step": 13461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140035906642729e-05,
+      "loss": 0.5318,
+      "step": 13462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139971787637856e-05,
+      "loss": 0.5782,
+      "step": 13463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139907668632983e-05,
+      "loss": 0.4916,
+      "step": 13464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13984354962811e-05,
+      "loss": 0.5091,
+      "step": 13465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139779430623237e-05,
+      "loss": 0.6457,
+      "step": 13466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139715311618364e-05,
+      "loss": 0.46,
+      "step": 13467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139651192613491e-05,
+      "loss": 0.5759,
+      "step": 13468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139587073608618e-05,
+      "loss": 0.5252,
+      "step": 13469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1395229546037444e-05,
+      "loss": 0.5117,
+      "step": 13470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1394588355988716e-05,
+      "loss": 0.6272,
+      "step": 13471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139394716593999e-05,
+      "loss": 0.5185,
+      "step": 13472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139330597589125e-05,
+      "loss": 0.6034,
+      "step": 13473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139266478584253e-05,
+      "loss": 0.5615,
+      "step": 13474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1392023595793796e-05,
+      "loss": 0.6651,
+      "step": 13475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139138240574507e-05,
+      "loss": 0.5127,
+      "step": 13476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139074121569633e-05,
+      "loss": 0.527,
+      "step": 13477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1390100025647604e-05,
+      "loss": 0.5343,
+      "step": 13478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138945883559887e-05,
+      "loss": 0.4908,
+      "step": 13479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138881764555014e-05,
+      "loss": 0.4996,
+      "step": 13480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138817645550141e-05,
+      "loss": 0.5008,
+      "step": 13481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1387535265452684e-05,
+      "loss": 0.5272,
+      "step": 13482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1386894075403956e-05,
+      "loss": 0.5412,
+      "step": 13483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138625288535522e-05,
+      "loss": 0.4739,
+      "step": 13484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138561169530649e-05,
+      "loss": 0.5306,
+      "step": 13485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138497050525776e-05,
+      "loss": 0.5228,
+      "step": 13486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138432931520903e-05,
+      "loss": 0.5741,
+      "step": 13487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1383688125160294e-05,
+      "loss": 0.5146,
+      "step": 13488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1383046935111566e-05,
+      "loss": 0.5207,
+      "step": 13489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1382405745062844e-05,
+      "loss": 0.5887,
+      "step": 13490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138176455501411e-05,
+      "loss": 0.4704,
+      "step": 13491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138112336496538e-05,
+      "loss": 0.5639,
+      "step": 13492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1380482174916646e-05,
+      "loss": 0.4192,
+      "step": 13493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137984098486792e-05,
+      "loss": 0.5249,
+      "step": 13494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137919979481918e-05,
+      "loss": 0.4658,
+      "step": 13495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1378558604770454e-05,
+      "loss": 0.6041,
+      "step": 13496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1377917414721726e-05,
+      "loss": 0.592,
+      "step": 13497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1377276224673e-05,
+      "loss": 0.5823,
+      "step": 13498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137663503462427e-05,
+      "loss": 0.4648,
+      "step": 13499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1375993844575534e-05,
+      "loss": 0.6019,
+      "step": 13500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1375352654526806e-05,
+      "loss": 0.5929,
+      "step": 13501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137471146447807e-05,
+      "loss": 0.5231,
+      "step": 13502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137407027442934e-05,
+      "loss": 0.5733,
+      "step": 13503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137342908438061e-05,
+      "loss": 0.4934,
+      "step": 13504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137278789433188e-05,
+      "loss": 0.5816,
+      "step": 13505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137214670428315e-05,
+      "loss": 0.5997,
+      "step": 13506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137150551423442e-05,
+      "loss": 0.5203,
+      "step": 13507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1370864324185695e-05,
+      "loss": 0.4716,
+      "step": 13508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137022313413696e-05,
+      "loss": 0.4933,
+      "step": 13509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136958194408823e-05,
+      "loss": 0.4445,
+      "step": 13510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1368940754039496e-05,
+      "loss": 0.4663,
+      "step": 13511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136829956399077e-05,
+      "loss": 0.6356,
+      "step": 13512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136765837394204e-05,
+      "loss": 0.584,
+      "step": 13513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136701718389331e-05,
+      "loss": 0.5875,
+      "step": 13514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1366375993844576e-05,
+      "loss": 0.6627,
+      "step": 13515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136573480379585e-05,
+      "loss": 0.5689,
+      "step": 13516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136509361374712e-05,
+      "loss": 0.6332,
+      "step": 13517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1364452423698385e-05,
+      "loss": 0.5981,
+      "step": 13518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1363811233649656e-05,
+      "loss": 0.597,
+      "step": 13519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136317004360092e-05,
+      "loss": 0.5264,
+      "step": 13520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136252885355219e-05,
+      "loss": 0.6163,
+      "step": 13521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1361887663503465e-05,
+      "loss": 0.4957,
+      "step": 13522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1361246473454736e-05,
+      "loss": 0.5647,
+      "step": 13523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1360605283406e-05,
+      "loss": 0.5052,
+      "step": 13524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135996409335727e-05,
+      "loss": 0.6524,
+      "step": 13525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1359322903308545e-05,
+      "loss": 0.4734,
+      "step": 13526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135868171325981e-05,
+      "loss": 0.547,
+      "step": 13527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135804052321108e-05,
+      "loss": 0.4778,
+      "step": 13528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135739933316235e-05,
+      "loss": 0.5618,
+      "step": 13529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1356758143113625e-05,
+      "loss": 0.5447,
+      "step": 13530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135611695306489e-05,
+      "loss": 0.4776,
+      "step": 13531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135547576301616e-05,
+      "loss": 0.5739,
+      "step": 13532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1354834572967426e-05,
+      "loss": 0.5622,
+      "step": 13533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13541933829187e-05,
+      "loss": 0.5877,
+      "step": 13534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135355219286997e-05,
+      "loss": 0.5163,
+      "step": 13535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1352911002821235e-05,
+      "loss": 0.4955,
+      "step": 13536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1352269812772507e-05,
+      "loss": 0.5248,
+      "step": 13537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135162862272378e-05,
+      "loss": 0.5797,
+      "step": 13538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.135098743267505e-05,
+      "loss": 0.6175,
+      "step": 13539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1350346242626315e-05,
+      "loss": 0.5574,
+      "step": 13540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134970505257759e-05,
+      "loss": 0.5443,
+      "step": 13541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134906386252885e-05,
+      "loss": 0.5915,
+      "step": 13542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134842267248012e-05,
+      "loss": 0.6431,
+      "step": 13543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1347781482431395e-05,
+      "loss": 0.5814,
+      "step": 13544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134714029238267e-05,
+      "loss": 0.5677,
+      "step": 13545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134649910233394e-05,
+      "loss": 0.569,
+      "step": 13546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1345857912285203e-05,
+      "loss": 0.5497,
+      "step": 13547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1345216722236475e-05,
+      "loss": 0.6019,
+      "step": 13548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134457553218774e-05,
+      "loss": 0.5319,
+      "step": 13549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134393434213901e-05,
+      "loss": 0.5081,
+      "step": 13550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134329315209028e-05,
+      "loss": 0.5159,
+      "step": 13551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134265196204155e-05,
+      "loss": 0.5895,
+      "step": 13552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134201077199282e-05,
+      "loss": 0.5479,
+      "step": 13553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134136958194409e-05,
+      "loss": 0.5427,
+      "step": 13554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1340728391895364e-05,
+      "loss": 0.4339,
+      "step": 13555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.134008720184663e-05,
+      "loss": 0.5607,
+      "step": 13556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13394460117979e-05,
+      "loss": 0.5181,
+      "step": 13557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1338804821749165e-05,
+      "loss": 0.4744,
+      "step": 13558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133816363170044e-05,
+      "loss": 0.5444,
+      "step": 13559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13375224416517e-05,
+      "loss": 0.4738,
+      "step": 13560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1336881251602974e-05,
+      "loss": 0.5445,
+      "step": 13561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133624006155425e-05,
+      "loss": 0.4952,
+      "step": 13562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133559887150552e-05,
+      "loss": 0.5123,
+      "step": 13563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133495768145679e-05,
+      "loss": 0.5673,
+      "step": 13564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1334316491408054e-05,
+      "loss": 0.5712,
+      "step": 13565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1333675301359325e-05,
+      "loss": 0.5487,
+      "step": 13566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133303411131059e-05,
+      "loss": 0.5369,
+      "step": 13567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133239292126186e-05,
+      "loss": 0.5234,
+      "step": 13568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1331751731213134e-05,
+      "loss": 0.606,
+      "step": 13569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1331110541164405e-05,
+      "loss": 0.5445,
+      "step": 13570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133046935111568e-05,
+      "loss": 0.4175,
+      "step": 13571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132982816106694e-05,
+      "loss": 0.5535,
+      "step": 13572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1329186971018214e-05,
+      "loss": 0.5124,
+      "step": 13573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132854578096948e-05,
+      "loss": 0.5313,
+      "step": 13574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132790459092075e-05,
+      "loss": 0.4994,
+      "step": 13575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1327263400872015e-05,
+      "loss": 0.5431,
+      "step": 13576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132662221082329e-05,
+      "loss": 0.4434,
+      "step": 13577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132598102077456e-05,
+      "loss": 0.6533,
+      "step": 13578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132533983072583e-05,
+      "loss": 0.5446,
+      "step": 13579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13246986406771e-05,
+      "loss": 0.4901,
+      "step": 13580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132405745062837e-05,
+      "loss": 0.5938,
+      "step": 13581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132341626057964e-05,
+      "loss": 0.3878,
+      "step": 13582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1322775070530904e-05,
+      "loss": 0.5911,
+      "step": 13583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1322133880482176e-05,
+      "loss": 0.5277,
+      "step": 13584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132149269043345e-05,
+      "loss": 0.6027,
+      "step": 13585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132085150038472e-05,
+      "loss": 0.4819,
+      "step": 13586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132021031033599e-05,
+      "loss": 0.6423,
+      "step": 13587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1319569120287256e-05,
+      "loss": 0.5658,
+      "step": 13588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131892793023853e-05,
+      "loss": 0.511,
+      "step": 13589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131828674018979e-05,
+      "loss": 0.5093,
+      "step": 13590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1317645550141064e-05,
+      "loss": 0.5116,
+      "step": 13591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131700436009233e-05,
+      "loss": 0.5874,
+      "step": 13592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13163631700436e-05,
+      "loss": 0.5659,
+      "step": 13593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131572197999487e-05,
+      "loss": 0.5843,
+      "step": 13594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1315080789946144e-05,
+      "loss": 0.535,
+      "step": 13595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1314439599897416e-05,
+      "loss": 0.6391,
+      "step": 13596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131379840984868e-05,
+      "loss": 0.4932,
+      "step": 13597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131315721979995e-05,
+      "loss": 0.637,
+      "step": 13598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131251602975122e-05,
+      "loss": 0.6492,
+      "step": 13599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131187483970249e-05,
+      "loss": 0.5145,
+      "step": 13600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131123364965376e-05,
+      "loss": 0.5551,
+      "step": 13601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.131059245960503e-05,
+      "loss": 0.4889,
+      "step": 13602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13099512695563e-05,
+      "loss": 0.542,
+      "step": 13603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130931007950757e-05,
+      "loss": 0.5713,
+      "step": 13604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130866888945884e-05,
+      "loss": 0.5788,
+      "step": 13605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1308027699410106e-05,
+      "loss": 0.4694,
+      "step": 13606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130738650936138e-05,
+      "loss": 0.5028,
+      "step": 13607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130674531931264e-05,
+      "loss": 0.5286,
+      "step": 13608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1306104129263914e-05,
+      "loss": 0.4758,
+      "step": 13609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1305462939215186e-05,
+      "loss": 0.5468,
+      "step": 13610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130482174916646e-05,
+      "loss": 0.4956,
+      "step": 13611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130418055911772e-05,
+      "loss": 0.4847,
+      "step": 13612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1303539369068994e-05,
+      "loss": 0.477,
+      "step": 13613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1302898179020266e-05,
+      "loss": 0.5169,
+      "step": 13614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.130225698897153e-05,
+      "loss": 0.4847,
+      "step": 13615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.13016157989228e-05,
+      "loss": 0.4406,
+      "step": 13616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1300974608874074e-05,
+      "loss": 0.5504,
+      "step": 13617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1300333418825346e-05,
+      "loss": 0.5806,
+      "step": 13618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129969222877661e-05,
+      "loss": 0.6304,
+      "step": 13619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129905103872788e-05,
+      "loss": 0.5445,
+      "step": 13620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129840984867915e-05,
+      "loss": 0.5708,
+      "step": 13621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129776865863042e-05,
+      "loss": 0.5488,
+      "step": 13622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129712746858169e-05,
+      "loss": 0.4858,
+      "step": 13623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1296486278532956e-05,
+      "loss": 0.6456,
+      "step": 13624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129584508848423e-05,
+      "loss": 0.5917,
+      "step": 13625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.12952038984355e-05,
+      "loss": 0.5653,
+      "step": 13626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129456270838677e-05,
+      "loss": 0.5417,
+      "step": 13627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1293921518338036e-05,
+      "loss": 0.6646,
+      "step": 13628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129328032828931e-05,
+      "loss": 0.587,
+      "step": 13629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129263913824057e-05,
+      "loss": 0.4727,
+      "step": 13630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1291997948191845e-05,
+      "loss": 0.5445,
+      "step": 13631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1291356758143116e-05,
+      "loss": 0.4667,
+      "step": 13632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129071556809439e-05,
+      "loss": 0.598,
+      "step": 13633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129007437804566e-05,
+      "loss": 0.5171,
+      "step": 13634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1289433187996925e-05,
+      "loss": 0.4829,
+      "step": 13635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1288791997948196e-05,
+      "loss": 0.5179,
+      "step": 13636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128815080789946e-05,
+      "loss": 0.5408,
+      "step": 13637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128750961785073e-05,
+      "loss": 0.6036,
+      "step": 13638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1286868427802e-05,
+      "loss": 0.5905,
+      "step": 13639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128622723775327e-05,
+      "loss": 0.516,
+      "step": 13640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128558604770454e-05,
+      "loss": 0.5956,
+      "step": 13641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128494485765581e-05,
+      "loss": 0.5856,
+      "step": 13642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1284303667607085e-05,
+      "loss": 0.5192,
+      "step": 13643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128366247755835e-05,
+      "loss": 0.479,
+      "step": 13644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128302128750962e-05,
+      "loss": 0.4986,
+      "step": 13645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1282380097460887e-05,
+      "loss": 0.4943,
+      "step": 13646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128173890741216e-05,
+      "loss": 0.4298,
+      "step": 13647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.128109771736342e-05,
+      "loss": 0.644,
+      "step": 13648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1280456527314695e-05,
+      "loss": 0.5959,
+      "step": 13649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.127981533726597e-05,
+      "loss": 0.6782,
+      "step": 13650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.127917414721724e-05,
+      "loss": 0.5744,
+      "step": 13651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.127853295716851e-05,
+      "loss": 0.5001,
+      "step": 13652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1277891767119775e-05,
+      "loss": 0.4527,
+      "step": 13653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.127725057707105e-05,
+      "loss": 0.5331,
+      "step": 13654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.127660938702231e-05,
+      "loss": 0.5717,
+      "step": 13655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127596819697358e-05,
+      "loss": 0.579,
+      "step": 13656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1275327006924855e-05,
+      "loss": 0.4002,
+      "step": 13657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127468581687613e-05,
+      "loss": 0.5329,
+      "step": 13658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12740446268274e-05,
+      "loss": 0.5615,
+      "step": 13659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1273403436778663e-05,
+      "loss": 0.6599,
+      "step": 13660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1272762246729935e-05,
+      "loss": 0.4674,
+      "step": 13661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12721210566812e-05,
+      "loss": 0.4976,
+      "step": 13662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127147986663247e-05,
+      "loss": 0.63,
+      "step": 13663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127083867658374e-05,
+      "loss": 0.6626,
+      "step": 13664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127019748653501e-05,
+      "loss": 0.5605,
+      "step": 13665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126955629648628e-05,
+      "loss": 0.4675,
+      "step": 13666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126891510643755e-05,
+      "loss": 0.545,
+      "step": 13667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1268273916388824e-05,
+      "loss": 0.4907,
+      "step": 13668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126763272634009e-05,
+      "loss": 0.5086,
+      "step": 13669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126699153629136e-05,
+      "loss": 0.5639,
+      "step": 13670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1266350346242625e-05,
+      "loss": 0.5191,
+      "step": 13671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12657091561939e-05,
+      "loss": 0.4772,
+      "step": 13672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126506796614517e-05,
+      "loss": 0.5236,
+      "step": 13673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126442677609644e-05,
+      "loss": 0.5704,
+      "step": 13674
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1263785586047705e-05,
+      "loss": 0.4479,
+      "step": 13675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126314439599898e-05,
+      "loss": 0.5214,
+      "step": 13676
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126250320595025e-05,
+      "loss": 0.5675,
+      "step": 13677
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1261862015901514e-05,
+      "loss": 0.5949,
+      "step": 13678
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1261220825852785e-05,
+      "loss": 0.6083,
+      "step": 13679
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126057963580405e-05,
+      "loss": 0.5812,
+      "step": 13680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125993844575532e-05,
+      "loss": 0.6152,
+      "step": 13681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1259297255706594e-05,
+      "loss": 0.5428,
+      "step": 13682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1258656065657865e-05,
+      "loss": 0.5723,
+      "step": 13683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125801487560913e-05,
+      "loss": 0.5723,
+      "step": 13684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12573736855604e-05,
+      "loss": 0.5002,
+      "step": 13685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1256732495511674e-05,
+      "loss": 0.5052,
+      "step": 13686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125609130546294e-05,
+      "loss": 0.5542,
+      "step": 13687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125545011541421e-05,
+      "loss": 0.6027,
+      "step": 13688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125480892536548e-05,
+      "loss": 0.5428,
+      "step": 13689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1254167735316754e-05,
+      "loss": 0.5689,
+      "step": 13690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125352654526802e-05,
+      "loss": 0.5703,
+      "step": 13691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125288535521929e-05,
+      "loss": 0.4656,
+      "step": 13692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1252244165170556e-05,
+      "loss": 0.6045,
+      "step": 13693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125160297512183e-05,
+      "loss": 0.672,
+      "step": 13694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12509617850731e-05,
+      "loss": 0.5575,
+      "step": 13695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1250320595024364e-05,
+      "loss": 0.4471,
+      "step": 13696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1249679404975636e-05,
+      "loss": 0.5607,
+      "step": 13697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124903821492691e-05,
+      "loss": 0.5052,
+      "step": 13698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124839702487818e-05,
+      "loss": 0.549,
+      "step": 13699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1247755834829444e-05,
+      "loss": 0.5724,
+      "step": 13700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1247114644780716e-05,
+      "loss": 0.5059,
+      "step": 13701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124647345473198e-05,
+      "loss": 0.5725,
+      "step": 13702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124583226468325e-05,
+      "loss": 0.5097,
+      "step": 13703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1245191074634524e-05,
+      "loss": 0.4755,
+      "step": 13704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1244549884585796e-05,
+      "loss": 0.6354,
+      "step": 13705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124390869453707e-05,
+      "loss": 0.5803,
+      "step": 13706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124326750448833e-05,
+      "loss": 0.4899,
+      "step": 13707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1242626314439604e-05,
+      "loss": 0.4354,
+      "step": 13708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124198512439087e-05,
+      "loss": 0.5182,
+      "step": 13709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124134393434214e-05,
+      "loss": 0.5155,
+      "step": 13710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1240702744293406e-05,
+      "loss": 0.5481,
+      "step": 13711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124006155424468e-05,
+      "loss": 0.5153,
+      "step": 13712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123942036419595e-05,
+      "loss": 0.4977,
+      "step": 13713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123877917414722e-05,
+      "loss": 0.561,
+      "step": 13714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123813798409849e-05,
+      "loss": 0.6113,
+      "step": 13715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123749679404976e-05,
+      "loss": 0.3978,
+      "step": 13716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123685560400103e-05,
+      "loss": 0.6106,
+      "step": 13717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1236214413952294e-05,
+      "loss": 0.4766,
+      "step": 13718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1235573223903566e-05,
+      "loss": 0.5047,
+      "step": 13719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123493203385483e-05,
+      "loss": 0.4795,
+      "step": 13720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12342908438061e-05,
+      "loss": 0.4782,
+      "step": 13721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123364965375738e-05,
+      "loss": 0.5757,
+      "step": 13722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1233008463708646e-05,
+      "loss": 0.6264,
+      "step": 13723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123236727365992e-05,
+      "loss": 0.5136,
+      "step": 13724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123172608361118e-05,
+      "loss": 0.5096,
+      "step": 13725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1231084893562454e-05,
+      "loss": 0.6092,
+      "step": 13726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123044370351372e-05,
+      "loss": 0.5453,
+      "step": 13727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122980251346499e-05,
+      "loss": 0.5136,
+      "step": 13728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122916132341626e-05,
+      "loss": 0.5107,
+      "step": 13729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1228520133367535e-05,
+      "loss": 0.4979,
+      "step": 13730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1227878943318806e-05,
+      "loss": 0.5722,
+      "step": 13731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122723775327007e-05,
+      "loss": 0.5701,
+      "step": 13732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122659656322134e-05,
+      "loss": 0.6122,
+      "step": 13733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122595537317261e-05,
+      "loss": 0.6538,
+      "step": 13734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122531418312388e-05,
+      "loss": 0.6525,
+      "step": 13735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1224672993075144e-05,
+      "loss": 0.5508,
+      "step": 13736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1224031803026416e-05,
+      "loss": 0.5304,
+      "step": 13737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122339061297769e-05,
+      "loss": 0.553,
+      "step": 13738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122274942292896e-05,
+      "loss": 0.6109,
+      "step": 13739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122210823288023e-05,
+      "loss": 0.5415,
+      "step": 13740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1221467042831496e-05,
+      "loss": 0.549,
+      "step": 13741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122082585278277e-05,
+      "loss": 0.5835,
+      "step": 13742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122018466273403e-05,
+      "loss": 0.6688,
+      "step": 13743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1219543472685305e-05,
+      "loss": 0.5243,
+      "step": 13744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1218902282636576e-05,
+      "loss": 0.5462,
+      "step": 13745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121826109258785e-05,
+      "loss": 0.5075,
+      "step": 13746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121761990253911e-05,
+      "loss": 0.5021,
+      "step": 13747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1216978712490385e-05,
+      "loss": 0.6322,
+      "step": 13748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1216337522441656e-05,
+      "loss": 0.525,
+      "step": 13749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121569633239292e-05,
+      "loss": 0.5324,
+      "step": 13750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121505514234419e-05,
+      "loss": 0.5739,
+      "step": 13751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121441395229546e-05,
+      "loss": 0.5258,
+      "step": 13752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121377276224673e-05,
+      "loss": 0.5556,
+      "step": 13753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1213131572198e-05,
+      "loss": 0.4938,
+      "step": 13754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121249038214927e-05,
+      "loss": 0.5137,
+      "step": 13755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121184919210054e-05,
+      "loss": 0.4712,
+      "step": 13756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121120800205181e-05,
+      "loss": 0.5768,
+      "step": 13757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121056681200308e-05,
+      "loss": 0.5458,
+      "step": 13758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1209925621954347e-05,
+      "loss": 0.5071,
+      "step": 13759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120928443190562e-05,
+      "loss": 0.5385,
+      "step": 13760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120864324185689e-05,
+      "loss": 0.568,
+      "step": 13761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120800205180816e-05,
+      "loss": 0.4747,
+      "step": 13762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120736086175943e-05,
+      "loss": 0.5077,
+      "step": 13763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12067196717107e-05,
+      "loss": 0.5885,
+      "step": 13764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120607848166196e-05,
+      "loss": 0.6294,
+      "step": 13765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1205437291613235e-05,
+      "loss": 0.4821,
+      "step": 13766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120479610156451e-05,
+      "loss": 0.6099,
+      "step": 13767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120415491151577e-05,
+      "loss": 0.5393,
+      "step": 13768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120351372146704e-05,
+      "loss": 0.6633,
+      "step": 13769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1202872531418315e-05,
+      "loss": 0.5803,
+      "step": 13770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120223134136959e-05,
+      "loss": 0.5277,
+      "step": 13771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120159015132085e-05,
+      "loss": 0.5344,
+      "step": 13772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1200948961272123e-05,
+      "loss": 0.6554,
+      "step": 13773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1200307771223395e-05,
+      "loss": 0.4029,
+      "step": 13774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119966658117466e-05,
+      "loss": 0.5434,
+      "step": 13775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119902539112593e-05,
+      "loss": 0.6626,
+      "step": 13776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1198384201077204e-05,
+      "loss": 0.4903,
+      "step": 13777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1197743011028475e-05,
+      "loss": 0.6364,
+      "step": 13778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119710182097974e-05,
+      "loss": 0.5071,
+      "step": 13779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119646063093101e-05,
+      "loss": 0.5829,
+      "step": 13780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119581944088228e-05,
+      "loss": 0.5726,
+      "step": 13781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119517825083355e-05,
+      "loss": 0.5018,
+      "step": 13782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119453706078482e-05,
+      "loss": 0.5407,
+      "step": 13783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1193895870736085e-05,
+      "loss": 0.5964,
+      "step": 13784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119325468068736e-05,
+      "loss": 0.5254,
+      "step": 13785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119261349063863e-05,
+      "loss": 0.6187,
+      "step": 13786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11919723005899e-05,
+      "loss": 0.5228,
+      "step": 13787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1191331110541165e-05,
+      "loss": 0.5822,
+      "step": 13788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119068992049244e-05,
+      "loss": 0.4653,
+      "step": 13789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11900487304437e-05,
+      "loss": 0.5572,
+      "step": 13790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1189407540394974e-05,
+      "loss": 0.6305,
+      "step": 13791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1188766350346245e-05,
+      "loss": 0.5288,
+      "step": 13792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118812516029752e-05,
+      "loss": 0.5836,
+      "step": 13793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118748397024879e-05,
+      "loss": 0.4721,
+      "step": 13794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1186842780200054e-05,
+      "loss": 0.6638,
+      "step": 13795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1186201590151326e-05,
+      "loss": 0.4857,
+      "step": 13796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118556040010259e-05,
+      "loss": 0.5066,
+      "step": 13797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118491921005386e-05,
+      "loss": 0.5895,
+      "step": 13798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118427802000513e-05,
+      "loss": 0.5158,
+      "step": 13799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11836368299564e-05,
+      "loss": 0.5551,
+      "step": 13800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118299563990767e-05,
+      "loss": 0.5221,
+      "step": 13801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118235444985894e-05,
+      "loss": 0.5268,
+      "step": 13802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1181713259810214e-05,
+      "loss": 0.5917,
+      "step": 13803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118107206976148e-05,
+      "loss": 0.4441,
+      "step": 13804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118043087971275e-05,
+      "loss": 0.5147,
+      "step": 13805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1179789689664016e-05,
+      "loss": 0.6467,
+      "step": 13806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117914849961529e-05,
+      "loss": 0.5824,
+      "step": 13807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117850730956655e-05,
+      "loss": 0.4098,
+      "step": 13808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1177866119517824e-05,
+      "loss": 0.5886,
+      "step": 13809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11772249294691e-05,
+      "loss": 0.4908,
+      "step": 13810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117658373942037e-05,
+      "loss": 0.4282,
+      "step": 13811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117594254937164e-05,
+      "loss": 0.6381,
+      "step": 13812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1175301359322904e-05,
+      "loss": 0.5049,
+      "step": 13813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1174660169274176e-05,
+      "loss": 0.5535,
+      "step": 13814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117401897922544e-05,
+      "loss": 0.5922,
+      "step": 13815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117337778917671e-05,
+      "loss": 0.4447,
+      "step": 13816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1172736599127984e-05,
+      "loss": 0.4161,
+      "step": 13817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1172095409079256e-05,
+      "loss": 0.5701,
+      "step": 13818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117145421903053e-05,
+      "loss": 0.5988,
+      "step": 13819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.117081302898179e-05,
+      "loss": 0.4797,
+      "step": 13820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1170171838933064e-05,
+      "loss": 0.5562,
+      "step": 13821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116953064888433e-05,
+      "loss": 0.5042,
+      "step": 13822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11688894588356e-05,
+      "loss": 0.419,
+      "step": 13823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1168248268786866e-05,
+      "loss": 0.5923,
+      "step": 13824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116760707873814e-05,
+      "loss": 0.6396,
+      "step": 13825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116696588868941e-05,
+      "loss": 0.5031,
+      "step": 13826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116632469864068e-05,
+      "loss": 0.6747,
+      "step": 13827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116568350859195e-05,
+      "loss": 0.5481,
+      "step": 13828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116504231854322e-05,
+      "loss": 0.6573,
+      "step": 13829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116440112849449e-05,
+      "loss": 0.6479,
+      "step": 13830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1163759938445754e-05,
+      "loss": 0.5425,
+      "step": 13831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1163118748397026e-05,
+      "loss": 0.6054,
+      "step": 13832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11624775583483e-05,
+      "loss": 0.5908,
+      "step": 13833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116183636829957e-05,
+      "loss": 0.5734,
+      "step": 13834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1161195178250834e-05,
+      "loss": 0.4696,
+      "step": 13835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1160553988202106e-05,
+      "loss": 0.5212,
+      "step": 13836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115991279815338e-05,
+      "loss": 0.4801,
+      "step": 13837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115927160810464e-05,
+      "loss": 0.583,
+      "step": 13838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1158630418055914e-05,
+      "loss": 0.5427,
+      "step": 13839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115798922800718e-05,
+      "loss": 0.532,
+      "step": 13840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115734803795845e-05,
+      "loss": 0.4005,
+      "step": 13841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115670684790972e-05,
+      "loss": 0.4798,
+      "step": 13842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1156065657860995e-05,
+      "loss": 0.4659,
+      "step": 13843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115542446781226e-05,
+      "loss": 0.5249,
+      "step": 13844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115478327776353e-05,
+      "loss": 0.5819,
+      "step": 13845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11541420877148e-05,
+      "loss": 0.6137,
+      "step": 13846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115350089766607e-05,
+      "loss": 0.4725,
+      "step": 13847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115285970761734e-05,
+      "loss": 0.5179,
+      "step": 13848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115221851756861e-05,
+      "loss": 0.5518,
+      "step": 13849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115157732751988e-05,
+      "loss": 0.5507,
+      "step": 13850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115093613747115e-05,
+      "loss": 0.5695,
+      "step": 13851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.115029494742242e-05,
+      "loss": 0.5116,
+      "step": 13852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1149653757373685e-05,
+      "loss": 0.5004,
+      "step": 13853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1149012567324956e-05,
+      "loss": 0.5755,
+      "step": 13854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114837137727623e-05,
+      "loss": 0.5896,
+      "step": 13855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114773018722749e-05,
+      "loss": 0.4692,
+      "step": 13856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1147088997178765e-05,
+      "loss": 0.5388,
+      "step": 13857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1146447807130036e-05,
+      "loss": 0.4517,
+      "step": 13858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114580661708131e-05,
+      "loss": 0.5095,
+      "step": 13859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114516542703257e-05,
+      "loss": 0.5502,
+      "step": 13860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1144524236983845e-05,
+      "loss": 0.545,
+      "step": 13861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114388304693511e-05,
+      "loss": 0.5444,
+      "step": 13862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114324185688638e-05,
+      "loss": 0.4857,
+      "step": 13863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114260066683765e-05,
+      "loss": 0.6171,
+      "step": 13864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1141959476788925e-05,
+      "loss": 0.5852,
+      "step": 13865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1141318286740197e-05,
+      "loss": 0.5337,
+      "step": 13866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114067709669146e-05,
+      "loss": 0.4832,
+      "step": 13867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114003590664273e-05,
+      "loss": 0.5621,
+      "step": 13868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1139394716594e-05,
+      "loss": 0.4715,
+      "step": 13869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113875352654527e-05,
+      "loss": 0.3625,
+      "step": 13870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1138112336496535e-05,
+      "loss": 0.5619,
+      "step": 13871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1137471146447807e-05,
+      "loss": 0.5245,
+      "step": 13872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113682995639908e-05,
+      "loss": 0.4928,
+      "step": 13873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113618876635035e-05,
+      "loss": 0.4971,
+      "step": 13874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113554757630162e-05,
+      "loss": 0.7615,
+      "step": 13875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113490638625289e-05,
+      "loss": 0.55,
+      "step": 13876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113426519620416e-05,
+      "loss": 0.4905,
+      "step": 13877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113362400615542e-05,
+      "loss": 0.6104,
+      "step": 13878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1132982816106695e-05,
+      "loss": 0.47,
+      "step": 13879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113234162605796e-05,
+      "loss": 0.5016,
+      "step": 13880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113170043600923e-05,
+      "loss": 0.5439,
+      "step": 13881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.113105924596051e-05,
+      "loss": 0.5685,
+      "step": 13882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1130418055911775e-05,
+      "loss": 0.6398,
+      "step": 13883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112977686586305e-05,
+      "loss": 0.5907,
+      "step": 13884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112913567581431e-05,
+      "loss": 0.4173,
+      "step": 13885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1128494485765583e-05,
+      "loss": 0.5118,
+      "step": 13886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112785329571685e-05,
+      "loss": 0.487,
+      "step": 13887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112721210566812e-05,
+      "loss": 0.4324,
+      "step": 13888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112657091561939e-05,
+      "loss": 0.5837,
+      "step": 13889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1125929725570664e-05,
+      "loss": 0.6375,
+      "step": 13890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1125288535521935e-05,
+      "loss": 0.5551,
+      "step": 13891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11246473454732e-05,
+      "loss": 0.4525,
+      "step": 13892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112400615542447e-05,
+      "loss": 0.5924,
+      "step": 13893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112336496537574e-05,
+      "loss": 0.557,
+      "step": 13894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112272377532701e-05,
+      "loss": 0.4867,
+      "step": 13895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1122082585278274e-05,
+      "loss": 0.3705,
+      "step": 13896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1121441395229545e-05,
+      "loss": 0.5629,
+      "step": 13897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112080020518082e-05,
+      "loss": 0.4749,
+      "step": 13898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.112015901513209e-05,
+      "loss": 0.6564,
+      "step": 13899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111951782508336e-05,
+      "loss": 0.382,
+      "step": 13900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1118876635034625e-05,
+      "loss": 0.5029,
+      "step": 13901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11182354449859e-05,
+      "loss": 0.5333,
+      "step": 13902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111759425493716e-05,
+      "loss": 0.5087,
+      "step": 13903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1116953064888434e-05,
+      "loss": 0.5302,
+      "step": 13904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1116311874839705e-05,
+      "loss": 0.5507,
+      "step": 13905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111567068479098e-05,
+      "loss": 0.5792,
+      "step": 13906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111502949474224e-05,
+      "loss": 0.5698,
+      "step": 13907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1114388304693514e-05,
+      "loss": 0.4552,
+      "step": 13908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1113747114644786e-05,
+      "loss": 0.5352,
+      "step": 13909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111310592459605e-05,
+      "loss": 0.5981,
+      "step": 13910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111246473454732e-05,
+      "loss": 0.5049,
+      "step": 13911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111182354449859e-05,
+      "loss": 0.4848,
+      "step": 13912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111118235444986e-05,
+      "loss": 0.5615,
+      "step": 13913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111054116440113e-05,
+      "loss": 0.5501,
+      "step": 13914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.11098999743524e-05,
+      "loss": 0.4761,
+      "step": 13915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110925878430367e-05,
+      "loss": 0.505,
+      "step": 13916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110861759425494e-05,
+      "loss": 0.5888,
+      "step": 13917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110797640420621e-05,
+      "loss": 0.515,
+      "step": 13918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1107335214157476e-05,
+      "loss": 0.6044,
+      "step": 13919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110669402410875e-05,
+      "loss": 0.5261,
+      "step": 13920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110605283406002e-05,
+      "loss": 0.4797,
+      "step": 13921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110541164401129e-05,
+      "loss": 0.4723,
+      "step": 13922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1104770453962556e-05,
+      "loss": 0.5608,
+      "step": 13923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110412926391383e-05,
+      "loss": 0.5088,
+      "step": 13924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110348807386509e-05,
+      "loss": 0.5096,
+      "step": 13925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1102846883816364e-05,
+      "loss": 0.4688,
+      "step": 13926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1102205693767636e-05,
+      "loss": 0.618,
+      "step": 13927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.11015645037189e-05,
+      "loss": 0.4292,
+      "step": 13928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110092331367017e-05,
+      "loss": 0.5592,
+      "step": 13929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1100282123621444e-05,
+      "loss": 0.531,
+      "step": 13930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1099640933572716e-05,
+      "loss": 0.6271,
+      "step": 13931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109899974352398e-05,
+      "loss": 0.5762,
+      "step": 13932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109835855347525e-05,
+      "loss": 0.5217,
+      "step": 13933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109771736342652e-05,
+      "loss": 0.608,
+      "step": 13934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109707617337779e-05,
+      "loss": 0.6721,
+      "step": 13935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109643498332906e-05,
+      "loss": 0.5643,
+      "step": 13936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109579379328033e-05,
+      "loss": 0.6004,
+      "step": 13937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1095152603231604e-05,
+      "loss": 0.5459,
+      "step": 13938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109451141318287e-05,
+      "loss": 0.5953,
+      "step": 13939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109387022313414e-05,
+      "loss": 0.5416,
+      "step": 13940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1093229033085406e-05,
+      "loss": 0.5441,
+      "step": 13941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109258784303668e-05,
+      "loss": 0.6103,
+      "step": 13942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109194665298794e-05,
+      "loss": 0.4647,
+      "step": 13943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1091305462939214e-05,
+      "loss": 0.5731,
+      "step": 13944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1090664272890486e-05,
+      "loss": 0.5155,
+      "step": 13945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109002308284176e-05,
+      "loss": 0.5551,
+      "step": 13946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108938189279303e-05,
+      "loss": 0.4977,
+      "step": 13947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1088740702744294e-05,
+      "loss": 0.5159,
+      "step": 13948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1088099512695566e-05,
+      "loss": 0.55,
+      "step": 13949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108745832264683e-05,
+      "loss": 0.5862,
+      "step": 13950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10868171325981e-05,
+      "loss": 0.5592,
+      "step": 13951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108617594254937e-05,
+      "loss": 0.5773,
+      "step": 13952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1085534752500646e-05,
+      "loss": 0.4228,
+      "step": 13953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108489356245192e-05,
+      "loss": 0.5274,
+      "step": 13954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108425237240318e-05,
+      "loss": 0.454,
+      "step": 13955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1083611182354455e-05,
+      "loss": 0.5568,
+      "step": 13956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108296999230572e-05,
+      "loss": 0.6183,
+      "step": 13957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108232880225699e-05,
+      "loss": 0.602,
+      "step": 13958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1081687612208256e-05,
+      "loss": 0.5016,
+      "step": 13959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108104642215953e-05,
+      "loss": 0.603,
+      "step": 13960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10804052321108e-05,
+      "loss": 0.5553,
+      "step": 13961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107976404206207e-05,
+      "loss": 0.5405,
+      "step": 13962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107912285201334e-05,
+      "loss": 0.5146,
+      "step": 13963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107848166196461e-05,
+      "loss": 0.4993,
+      "step": 13964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107784047191588e-05,
+      "loss": 0.4142,
+      "step": 13965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1077199281867145e-05,
+      "loss": 0.4899,
+      "step": 13966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1076558091818416e-05,
+      "loss": 0.5542,
+      "step": 13967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107591690176968e-05,
+      "loss": 0.5466,
+      "step": 13968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107527571172095e-05,
+      "loss": 0.6022,
+      "step": 13969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107463452167223e-05,
+      "loss": 0.6144,
+      "step": 13970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1073993331623496e-05,
+      "loss": 0.3967,
+      "step": 13971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107335214157477e-05,
+      "loss": 0.6373,
+      "step": 13972
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107271095152603e-05,
+      "loss": 0.5343,
+      "step": 13973
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1072069761477305e-05,
+      "loss": 0.5051,
+      "step": 13974
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.534,
+      "step": 13975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107078738137984e-05,
+      "loss": 0.5004,
+      "step": 13976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107014619133111e-05,
+      "loss": 0.4749,
+      "step": 13977
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1069505001282385e-05,
+      "loss": 0.5535,
+      "step": 13978
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106886381123366e-05,
+      "loss": 0.6237,
+      "step": 13979
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106822262118492e-05,
+      "loss": 0.6648,
+      "step": 13980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106758143113619e-05,
+      "loss": 0.6196,
+      "step": 13981
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106694024108746e-05,
+      "loss": 0.4818,
+      "step": 13982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106629905103873e-05,
+      "loss": 0.5809,
+      "step": 13983
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1065657860989995e-05,
+      "loss": 0.47,
+      "step": 13984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1065016670941267e-05,
+      "loss": 0.4843,
+      "step": 13985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106437548089254e-05,
+      "loss": 0.4517,
+      "step": 13986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106373429084381e-05,
+      "loss": 0.6959,
+      "step": 13987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106309310079508e-05,
+      "loss": 0.5468,
+      "step": 13988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106245191074635e-05,
+      "loss": 0.5664,
+      "step": 13989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106181072069762e-05,
+      "loss": 0.5804,
+      "step": 13990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106116953064888e-05,
+      "loss": 0.5771,
+      "step": 13991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1060528340600155e-05,
+      "loss": 0.5042,
+      "step": 13992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105988715055143e-05,
+      "loss": 0.5656,
+      "step": 13993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10592459605027e-05,
+      "loss": 0.7265,
+      "step": 13994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1058604770453963e-05,
+      "loss": 0.5166,
+      "step": 13995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1057963580405235e-05,
+      "loss": 0.614,
+      "step": 13996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105732239035651e-05,
+      "loss": 0.6223,
+      "step": 13997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105668120030777e-05,
+      "loss": 0.5713,
+      "step": 13998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1056040010259044e-05,
+      "loss": 0.5331,
+      "step": 13999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105539882021031e-05,
+      "loss": 0.465,
+      "step": 14000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105475763016158e-05,
+      "loss": 0.4936,
+      "step": 14001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105411644011285e-05,
+      "loss": 0.5762,
+      "step": 14002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1053475250064124e-05,
+      "loss": 0.4683,
+      "step": 14003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105283406001539e-05,
+      "loss": 0.53,
+      "step": 14004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105219286996666e-05,
+      "loss": 0.6495,
+      "step": 14005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105155167991793e-05,
+      "loss": 0.5149,
+      "step": 14006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10509104898692e-05,
+      "loss": 0.5549,
+      "step": 14007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105026929982047e-05,
+      "loss": 0.5856,
+      "step": 14008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104962810977174e-05,
+      "loss": 0.6919,
+      "step": 14009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104898691972301e-05,
+      "loss": 0.5424,
+      "step": 14010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104834572967428e-05,
+      "loss": 0.6047,
+      "step": 14011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104770453962555e-05,
+      "loss": 0.5258,
+      "step": 14012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1047063349576814e-05,
+      "loss": 0.467,
+      "step": 14013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1046422159528085e-05,
+      "loss": 0.5157,
+      "step": 14014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104578096947936e-05,
+      "loss": 0.6223,
+      "step": 14015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104513977943062e-05,
+      "loss": 0.5475,
+      "step": 14016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1044498589381894e-05,
+      "loss": 0.4586,
+      "step": 14017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1043857399333165e-05,
+      "loss": 0.5621,
+      "step": 14018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104321620928444e-05,
+      "loss": 0.4851,
+      "step": 14019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10425750192357e-05,
+      "loss": 0.5383,
+      "step": 14020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1041933829186974e-05,
+      "loss": 0.482,
+      "step": 14021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104129263913824e-05,
+      "loss": 0.6434,
+      "step": 14022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104065144908951e-05,
+      "loss": 0.5546,
+      "step": 14023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104001025904078e-05,
+      "loss": 0.5299,
+      "step": 14024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1039369068992054e-05,
+      "loss": 0.5208,
+      "step": 14025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1038727878943326e-05,
+      "loss": 0.5104,
+      "step": 14026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103808668889459e-05,
+      "loss": 0.5635,
+      "step": 14027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103744549884586e-05,
+      "loss": 0.5324,
+      "step": 14028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103680430879713e-05,
+      "loss": 0.5206,
+      "step": 14029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10361631187484e-05,
+      "loss": 0.5008,
+      "step": 14030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1035521928699664e-05,
+      "loss": 0.5309,
+      "step": 14031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1034880738650936e-05,
+      "loss": 0.5257,
+      "step": 14032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103423954860221e-05,
+      "loss": 0.5596,
+      "step": 14033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103359835855348e-05,
+      "loss": 0.5191,
+      "step": 14034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103295716850475e-05,
+      "loss": 0.5533,
+      "step": 14035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1032315978456016e-05,
+      "loss": 0.5754,
+      "step": 14036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103167478840729e-05,
+      "loss": 0.5331,
+      "step": 14037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103103359835855e-05,
+      "loss": 0.5135,
+      "step": 14038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1030392408309824e-05,
+      "loss": 0.5453,
+      "step": 14039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102975121826109e-05,
+      "loss": 0.4828,
+      "step": 14040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102911002821237e-05,
+      "loss": 0.4584,
+      "step": 14041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102846883816364e-05,
+      "loss": 0.5823,
+      "step": 14042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1027827648114904e-05,
+      "loss": 0.4882,
+      "step": 14043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1027186458066176e-05,
+      "loss": 0.5075,
+      "step": 14044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102654526801744e-05,
+      "loss": 0.5483,
+      "step": 14045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102590407796871e-05,
+      "loss": 0.507,
+      "step": 14046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102526288791998e-05,
+      "loss": 0.4965,
+      "step": 14047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102462169787125e-05,
+      "loss": 0.6613,
+      "step": 14048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102398050782252e-05,
+      "loss": 0.5669,
+      "step": 14049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102333931777379e-05,
+      "loss": 0.5908,
+      "step": 14050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1022698127725064e-05,
+      "loss": 0.5007,
+      "step": 14051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102205693767633e-05,
+      "loss": 0.5386,
+      "step": 14052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10214157476276e-05,
+      "loss": 0.4926,
+      "step": 14053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1020774557578866e-05,
+      "loss": 0.5347,
+      "step": 14054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102013336753014e-05,
+      "loss": 0.5935,
+      "step": 14055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10194921774814e-05,
+      "loss": 0.4859,
+      "step": 14056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1018850987432674e-05,
+      "loss": 0.4757,
+      "step": 14057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1018209797383946e-05,
+      "loss": 0.5124,
+      "step": 14058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101756860733522e-05,
+      "loss": 0.4623,
+      "step": 14059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101692741728649e-05,
+      "loss": 0.5298,
+      "step": 14060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1016286227237754e-05,
+      "loss": 0.5324,
+      "step": 14061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1015645037189026e-05,
+      "loss": 0.4994,
+      "step": 14062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101500384714029e-05,
+      "loss": 0.4957,
+      "step": 14063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101436265709156e-05,
+      "loss": 0.5638,
+      "step": 14064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1013721467042835e-05,
+      "loss": 0.5643,
+      "step": 14065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1013080276994106e-05,
+      "loss": 0.5696,
+      "step": 14066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101243908694537e-05,
+      "loss": 0.5432,
+      "step": 14067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101179789689664e-05,
+      "loss": 0.4498,
+      "step": 14068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1011156706847915e-05,
+      "loss": 0.4781,
+      "step": 14069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101051551679918e-05,
+      "loss": 0.6087,
+      "step": 14070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100987432675045e-05,
+      "loss": 0.6423,
+      "step": 14071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1009233136701716e-05,
+      "loss": 0.578,
+      "step": 14072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100859194665299e-05,
+      "loss": 0.4654,
+      "step": 14073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100795075660426e-05,
+      "loss": 0.474,
+      "step": 14074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100730956655553e-05,
+      "loss": 0.4542,
+      "step": 14075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1006668376506796e-05,
+      "loss": 0.6003,
+      "step": 14076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100602718645807e-05,
+      "loss": 0.4692,
+      "step": 14077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100538599640934e-05,
+      "loss": 0.5752,
+      "step": 14078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1004744806360605e-05,
+      "loss": 0.4578,
+      "step": 14079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1004103616311876e-05,
+      "loss": 0.5026,
+      "step": 14080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100346242626315e-05,
+      "loss": 0.5639,
+      "step": 14081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100282123621442e-05,
+      "loss": 0.4288,
+      "step": 14082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1002180046165685e-05,
+      "loss": 0.5267,
+      "step": 14083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1001538856116956e-05,
+      "loss": 0.4717,
+      "step": 14084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100089766606822e-05,
+      "loss": 0.6007,
+      "step": 14085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.100025647601949e-05,
+      "loss": 0.4214,
+      "step": 14086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0999615285970765e-05,
+      "loss": 0.5682,
+      "step": 14087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099897409592203e-05,
+      "loss": 0.5762,
+      "step": 14088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.09983329058733e-05,
+      "loss": 0.5822,
+      "step": 14089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099769171582457e-05,
+      "loss": 0.4803,
+      "step": 14090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0997050525775845e-05,
+      "loss": 0.428,
+      "step": 14091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099640933572711e-05,
+      "loss": 0.5344,
+      "step": 14092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099576814567838e-05,
+      "loss": 0.5008,
+      "step": 14093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0995126955629647e-05,
+      "loss": 0.5279,
+      "step": 14094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099448576558092e-05,
+      "loss": 0.5468,
+      "step": 14095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099384457553219e-05,
+      "loss": 0.4723,
+      "step": 14096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099320338548346e-05,
+      "loss": 0.4875,
+      "step": 14097
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0992562195434733e-05,
+      "loss": 0.5259,
+      "step": 14098
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0991921005386e-05,
+      "loss": 0.5736,
+      "step": 14099
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099127981533727e-05,
+      "loss": 0.6499,
+      "step": 14100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0990638625288535e-05,
+      "loss": 0.5321,
+      "step": 14101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098999743523981e-05,
+      "loss": 0.6262,
+      "step": 14102
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098935624519107e-05,
+      "loss": 0.4818,
+      "step": 14103
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098871505514234e-05,
+      "loss": 0.4597,
+      "step": 14104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0988073865093615e-05,
+      "loss": 0.5556,
+      "step": 14105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098743267504489e-05,
+      "loss": 0.5561,
+      "step": 14106
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098679148499616e-05,
+      "loss": 0.5221,
+      "step": 14107
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0986150294947423e-05,
+      "loss": 0.5297,
+      "step": 14108
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0985509104898695e-05,
+      "loss": 0.4998,
+      "step": 14109
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098486791484996e-05,
+      "loss": 0.5332,
+      "step": 14110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098422672480123e-05,
+      "loss": 0.6082,
+      "step": 14111
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.09835855347525e-05,
+      "loss": 0.5087,
+      "step": 14112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0982944344703775e-05,
+      "loss": 0.4222,
+      "step": 14113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098230315465505e-05,
+      "loss": 0.5246,
+      "step": 14114
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098166196460631e-05,
+      "loss": 0.4982,
+      "step": 14115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0981020774557584e-05,
+      "loss": 0.4938,
+      "step": 14116
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098037958450885e-05,
+      "loss": 0.6251,
+      "step": 14117
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097973839446012e-05,
+      "loss": 0.4222,
+      "step": 14118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0979097204411385e-05,
+      "loss": 0.5066,
+      "step": 14119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097845601436266e-05,
+      "loss": 0.5038,
+      "step": 14120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097781482431393e-05,
+      "loss": 0.7409,
+      "step": 14121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.09771736342652e-05,
+      "loss": 0.4409,
+      "step": 14122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097653244421647e-05,
+      "loss": 0.632,
+      "step": 14123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097589125416774e-05,
+      "loss": 0.5573,
+      "step": 14124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097525006411901e-05,
+      "loss": 0.544,
+      "step": 14125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0974608874070274e-05,
+      "loss": 0.4945,
+      "step": 14126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0973967684021545e-05,
+      "loss": 0.5735,
+      "step": 14127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097332649397281e-05,
+      "loss": 0.4519,
+      "step": 14128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097268530392408e-05,
+      "loss": 0.5377,
+      "step": 14129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0972044113875354e-05,
+      "loss": 0.5123,
+      "step": 14130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0971402923826626e-05,
+      "loss": 0.4551,
+      "step": 14131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.09707617337779e-05,
+      "loss": 0.5546,
+      "step": 14132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.097012054372916e-05,
+      "loss": 0.5251,
+      "step": 14133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0969479353680434e-05,
+      "loss": 0.5434,
+      "step": 14134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.09688381636317e-05,
+      "loss": 0.582,
+      "step": 14135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096819697358297e-05,
+      "loss": 0.6954,
+      "step": 14136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096755578353424e-05,
+      "loss": 0.4262,
+      "step": 14137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0966914593485514e-05,
+      "loss": 0.5299,
+      "step": 14138
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096627340343678e-05,
+      "loss": 0.4695,
+      "step": 14139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096563221338805e-05,
+      "loss": 0.4818,
+      "step": 14140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096499102333932e-05,
+      "loss": 0.5513,
+      "step": 14141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096434983329059e-05,
+      "loss": 0.6223,
+      "step": 14142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096370864324186e-05,
+      "loss": 0.5691,
+      "step": 14143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0963067453193124e-05,
+      "loss": 0.6137,
+      "step": 14144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0962426263144396e-05,
+      "loss": 0.4789,
+      "step": 14145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096178507309567e-05,
+      "loss": 0.484,
+      "step": 14146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096114388304694e-05,
+      "loss": 0.6214,
+      "step": 14147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0960502692998204e-05,
+      "loss": 0.5864,
+      "step": 14148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0959861502949476e-05,
+      "loss": 0.5445,
+      "step": 14149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095922031290075e-05,
+      "loss": 0.4708,
+      "step": 14150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095857912285201e-05,
+      "loss": 0.5788,
+      "step": 14151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0957937932803284e-05,
+      "loss": 0.5147,
+      "step": 14152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0957296742754556e-05,
+      "loss": 0.58,
+      "step": 14153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095665555270583e-05,
+      "loss": 0.5639,
+      "step": 14154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095601436265709e-05,
+      "loss": 0.6084,
+      "step": 14155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0955373172608364e-05,
+      "loss": 0.505,
+      "step": 14156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0954731982559636e-05,
+      "loss": 0.5178,
+      "step": 14157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.09540907925109e-05,
+      "loss": 0.4192,
+      "step": 14158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095344960246217e-05,
+      "loss": 0.5764,
+      "step": 14159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095280841241344e-05,
+      "loss": 0.4743,
+      "step": 14160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095216722236471e-05,
+      "loss": 0.5278,
+      "step": 14161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095152603231598e-05,
+      "loss": 0.5189,
+      "step": 14162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095088484226725e-05,
+      "loss": 0.5626,
+      "step": 14163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095024365221852e-05,
+      "loss": 0.4603,
+      "step": 14164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.094960246216979e-05,
+      "loss": 0.5929,
+      "step": 14165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.094896127212106e-05,
+      "loss": 0.5344,
+      "step": 14166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0948320082072326e-05,
+      "loss": 0.626,
+      "step": 14167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.09476788920236e-05,
+      "loss": 0.5213,
+      "step": 14168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.094703770197487e-05,
+      "loss": 0.4979,
+      "step": 14169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.094639651192614e-05,
+      "loss": 0.5081,
+      "step": 14170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0945755321877406e-05,
+      "loss": 0.4937,
+      "step": 14171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.094511413182868e-05,
+      "loss": 0.6152,
+      "step": 14172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.094447294177994e-05,
+      "loss": 0.6186,
+      "step": 14173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0943831751731214e-05,
+      "loss": 0.4472,
+      "step": 14174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0943190561682486e-05,
+      "loss": 0.5652,
+      "step": 14175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.094254937163375e-05,
+      "loss": 0.5233,
+      "step": 14176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.094190818158502e-05,
+      "loss": 0.4323,
+      "step": 14177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0941266991536295e-05,
+      "loss": 0.5027,
+      "step": 14178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0940625801487566e-05,
+      "loss": 0.6144,
+      "step": 14179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093998461143883e-05,
+      "loss": 0.467,
+      "step": 14180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.09393434213901e-05,
+      "loss": 0.605,
+      "step": 14181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093870223134137e-05,
+      "loss": 0.6161,
+      "step": 14182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093806104129264e-05,
+      "loss": 0.5625,
+      "step": 14183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093741985124391e-05,
+      "loss": 0.4345,
+      "step": 14184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093677866119518e-05,
+      "loss": 0.5291,
+      "step": 14185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0936137471146455e-05,
+      "loss": 0.6276,
+      "step": 14186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093549628109772e-05,
+      "loss": 0.5342,
+      "step": 14187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093485509104899e-05,
+      "loss": 0.4546,
+      "step": 14188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0934213901000256e-05,
+      "loss": 0.5999,
+      "step": 14189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093357271095153e-05,
+      "loss": 0.5444,
+      "step": 14190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093293152090279e-05,
+      "loss": 0.5089,
+      "step": 14191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0932290330854065e-05,
+      "loss": 0.5517,
+      "step": 14192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0931649140805336e-05,
+      "loss": 0.5345,
+      "step": 14193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093100795075661e-05,
+      "loss": 0.4314,
+      "step": 14194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093036676070788e-05,
+      "loss": 0.5836,
+      "step": 14195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0929725570659145e-05,
+      "loss": 0.4837,
+      "step": 14196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0929084380610417e-05,
+      "loss": 0.4445,
+      "step": 14197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092844319056168e-05,
+      "loss": 0.529,
+      "step": 14198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092780200051295e-05,
+      "loss": 0.5075,
+      "step": 14199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092716081046422e-05,
+      "loss": 0.6732,
+      "step": 14200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0926519620415497e-05,
+      "loss": 0.4532,
+      "step": 14201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092587843036677e-05,
+      "loss": 0.537,
+      "step": 14202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092523724031803e-05,
+      "loss": 0.6632,
+      "step": 14203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0924596050269305e-05,
+      "loss": 0.5266,
+      "step": 14204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092395486022057e-05,
+      "loss": 0.5994,
+      "step": 14205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092331367017184e-05,
+      "loss": 0.5289,
+      "step": 14206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0922672480123107e-05,
+      "loss": 0.578,
+      "step": 14207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092203129007438e-05,
+      "loss": 0.5346,
+      "step": 14208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092139010002565e-05,
+      "loss": 0.4908,
+      "step": 14209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092074890997692e-05,
+      "loss": 0.5836,
+      "step": 14210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0920107719928193e-05,
+      "loss": 0.5341,
+      "step": 14211
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091946652987946e-05,
+      "loss": 0.5853,
+      "step": 14212
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091882533983073e-05,
+      "loss": 0.6157,
+      "step": 14213
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0918184149781995e-05,
+      "loss": 0.5317,
+      "step": 14214
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091754295973327e-05,
+      "loss": 0.5172,
+      "step": 14215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091690176968453e-05,
+      "loss": 0.6029,
+      "step": 14216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0916260579635803e-05,
+      "loss": 0.5912,
+      "step": 14217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0915619389587075e-05,
+      "loss": 0.5993,
+      "step": 14218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091497819953835e-05,
+      "loss": 0.4845,
+      "step": 14219
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091433700948962e-05,
+      "loss": 0.5302,
+      "step": 14220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0913695819440884e-05,
+      "loss": 0.6127,
+      "step": 14221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0913054629392155e-05,
+      "loss": 0.5214,
+      "step": 14222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091241343934342e-05,
+      "loss": 0.501,
+      "step": 14223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091177224929469e-05,
+      "loss": 0.5511,
+      "step": 14224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0911131059245964e-05,
+      "loss": 0.6608,
+      "step": 14225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0910489869197235e-05,
+      "loss": 0.5422,
+      "step": 14226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.09098486791485e-05,
+      "loss": 0.5563,
+      "step": 14227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090920748909977e-05,
+      "loss": 0.5314,
+      "step": 14228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0908566299051044e-05,
+      "loss": 0.4605,
+      "step": 14229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090792510900231e-05,
+      "loss": 0.62,
+      "step": 14230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090728391895358e-05,
+      "loss": 0.5636,
+      "step": 14231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0906642728904845e-05,
+      "loss": 0.518,
+      "step": 14232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090600153885612e-05,
+      "loss": 0.4912,
+      "step": 14233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090536034880739e-05,
+      "loss": 0.6567,
+      "step": 14234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090471915875866e-05,
+      "loss": 0.4572,
+      "step": 14235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0904077968709925e-05,
+      "loss": 0.4181,
+      "step": 14236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.09034367786612e-05,
+      "loss": 0.4595,
+      "step": 14237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090279558861247e-05,
+      "loss": 0.5569,
+      "step": 14238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0902154398563734e-05,
+      "loss": 0.5089,
+      "step": 14239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0901513208515005e-05,
+      "loss": 0.4972,
+      "step": 14240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090087201846628e-05,
+      "loss": 0.5762,
+      "step": 14241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090023082841755e-05,
+      "loss": 0.4415,
+      "step": 14242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0899589638368814e-05,
+      "loss": 0.6411,
+      "step": 14243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0898948448320086e-05,
+      "loss": 0.5297,
+      "step": 14244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089830725827135e-05,
+      "loss": 0.6593,
+      "step": 14245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089766606822262e-05,
+      "loss": 0.5077,
+      "step": 14246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0897024878173894e-05,
+      "loss": 0.5272,
+      "step": 14247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089638368812516e-05,
+      "loss": 0.6239,
+      "step": 14248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089574249807643e-05,
+      "loss": 0.4577,
+      "step": 14249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08951013080277e-05,
+      "loss": 0.5301,
+      "step": 14250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0894460117978974e-05,
+      "loss": 0.5242,
+      "step": 14251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089381892793024e-05,
+      "loss": 0.4371,
+      "step": 14252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089317773788151e-05,
+      "loss": 0.522,
+      "step": 14253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0892536547832776e-05,
+      "loss": 0.5243,
+      "step": 14254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089189535778405e-05,
+      "loss": 0.5798,
+      "step": 14255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089125416773532e-05,
+      "loss": 0.5475,
+      "step": 14256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089061297768659e-05,
+      "loss": 0.5446,
+      "step": 14257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088997178763786e-05,
+      "loss": 0.522,
+      "step": 14258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088933059758913e-05,
+      "loss": 0.5658,
+      "step": 14259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08886894075404e-05,
+      "loss": 0.5396,
+      "step": 14260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0888048217491664e-05,
+      "loss": 0.5241,
+      "step": 14261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0887407027442936e-05,
+      "loss": 0.5436,
+      "step": 14262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08867658373942e-05,
+      "loss": 0.5698,
+      "step": 14263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088612464734547e-05,
+      "loss": 0.5917,
+      "step": 14264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0885483457296744e-05,
+      "loss": 0.4872,
+      "step": 14265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0884842267248016e-05,
+      "loss": 0.4923,
+      "step": 14266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088420107719929e-05,
+      "loss": 0.5409,
+      "step": 14267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088355988715055e-05,
+      "loss": 0.55,
+      "step": 14268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0882918697101824e-05,
+      "loss": 0.6478,
+      "step": 14269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088227750705309e-05,
+      "loss": 0.6159,
+      "step": 14270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088163631700436e-05,
+      "loss": 0.6188,
+      "step": 14271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0880995126955626e-05,
+      "loss": 0.556,
+      "step": 14272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0880353936906904e-05,
+      "loss": 0.5574,
+      "step": 14273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0879712746858176e-05,
+      "loss": 0.4942,
+      "step": 14274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087907155680944e-05,
+      "loss": 0.4944,
+      "step": 14275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087843036676071e-05,
+      "loss": 0.5107,
+      "step": 14276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087778917671198e-05,
+      "loss": 0.5195,
+      "step": 14277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087714798666325e-05,
+      "loss": 0.5608,
+      "step": 14278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0876506796614514e-05,
+      "loss": 0.5195,
+      "step": 14279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0875865606565786e-05,
+      "loss": 0.5292,
+      "step": 14280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087522441651706e-05,
+      "loss": 0.6467,
+      "step": 14281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087458322646833e-05,
+      "loss": 0.5306,
+      "step": 14282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08739420364196e-05,
+      "loss": 0.5809,
+      "step": 14283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0873300846370866e-05,
+      "loss": 0.5476,
+      "step": 14284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087265965632214e-05,
+      "loss": 0.5287,
+      "step": 14285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08720184662734e-05,
+      "loss": 0.5684,
+      "step": 14286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0871377276224675e-05,
+      "loss": 0.5967,
+      "step": 14287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087073608617594e-05,
+      "loss": 0.5497,
+      "step": 14288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087009489612721e-05,
+      "loss": 0.4805,
+      "step": 14289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086945370607848e-05,
+      "loss": 0.4638,
+      "step": 14290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0868812516029755e-05,
+      "loss": 0.6153,
+      "step": 14291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0868171325981026e-05,
+      "loss": 0.5545,
+      "step": 14292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086753013593229e-05,
+      "loss": 0.4707,
+      "step": 14293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086688894588356e-05,
+      "loss": 0.5711,
+      "step": 14294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086624775583483e-05,
+      "loss": 0.5104,
+      "step": 14295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08656065657861e-05,
+      "loss": 0.6746,
+      "step": 14296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086496537573737e-05,
+      "loss": 0.5156,
+      "step": 14297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086432418568864e-05,
+      "loss": 0.5729,
+      "step": 14298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086368299563991e-05,
+      "loss": 0.6436,
+      "step": 14299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086304180559118e-05,
+      "loss": 0.6031,
+      "step": 14300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086240061554245e-05,
+      "loss": 0.4949,
+      "step": 14301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0861759425493716e-05,
+      "loss": 0.4351,
+      "step": 14302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086111823544499e-05,
+      "loss": 0.4751,
+      "step": 14303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086047704539625e-05,
+      "loss": 0.5025,
+      "step": 14304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0859835855347525e-05,
+      "loss": 0.6102,
+      "step": 14305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0859194665298796e-05,
+      "loss": 0.5642,
+      "step": 14306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085855347525007e-05,
+      "loss": 0.594,
+      "step": 14307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085791228520133e-05,
+      "loss": 0.4807,
+      "step": 14308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0857271095152605e-05,
+      "loss": 0.468,
+      "step": 14309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0856629905103877e-05,
+      "loss": 0.4695,
+      "step": 14310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085598871505514e-05,
+      "loss": 0.6433,
+      "step": 14311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085534752500641e-05,
+      "loss": 0.5529,
+      "step": 14312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0854706334957685e-05,
+      "loss": 0.6177,
+      "step": 14313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085406514490896e-05,
+      "loss": 0.6003,
+      "step": 14314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085342395486022e-05,
+      "loss": 0.6018,
+      "step": 14315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085278276481149e-05,
+      "loss": 0.5567,
+      "step": 14316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085214157476276e-05,
+      "loss": 0.4337,
+      "step": 14317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085150038471403e-05,
+      "loss": 0.5752,
+      "step": 14318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08508591946653e-05,
+      "loss": 0.5413,
+      "step": 14319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085021800461657e-05,
+      "loss": 0.5151,
+      "step": 14320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084957681456784e-05,
+      "loss": 0.5879,
+      "step": 14321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084893562451911e-05,
+      "loss": 0.5274,
+      "step": 14322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084829443447038e-05,
+      "loss": 0.5031,
+      "step": 14323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084765324442165e-05,
+      "loss": 0.6269,
+      "step": 14324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084701205437292e-05,
+      "loss": 0.617,
+      "step": 14325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084637086432418e-05,
+      "loss": 0.5779,
+      "step": 14326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0845729674275455e-05,
+      "loss": 0.5454,
+      "step": 14327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084508848422673e-05,
+      "loss": 0.5194,
+      "step": 14328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0844447294178e-05,
+      "loss": 0.5454,
+      "step": 14329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084380610412927e-05,
+      "loss": 0.5453,
+      "step": 14330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0843164914080535e-05,
+      "loss": 0.5185,
+      "step": 14331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084252372403181e-05,
+      "loss": 0.5371,
+      "step": 14332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084188253398307e-05,
+      "loss": 0.6287,
+      "step": 14333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0841241343934344e-05,
+      "loss": 0.5595,
+      "step": 14334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.084060015388561e-05,
+      "loss": 0.5608,
+      "step": 14335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083995896383688e-05,
+      "loss": 0.5189,
+      "step": 14336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083931777378815e-05,
+      "loss": 0.4781,
+      "step": 14337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0838676583739424e-05,
+      "loss": 0.558,
+      "step": 14338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0838035393690695e-05,
+      "loss": 0.4753,
+      "step": 14339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083739420364196e-05,
+      "loss": 0.6213,
+      "step": 14340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083675301359323e-05,
+      "loss": 0.5062,
+      "step": 14341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08361118235445e-05,
+      "loss": 0.6571,
+      "step": 14342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083547063349577e-05,
+      "loss": 0.4743,
+      "step": 14343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083482944344704e-05,
+      "loss": 0.445,
+      "step": 14344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083418825339831e-05,
+      "loss": 0.488,
+      "step": 14345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0833547063349584e-05,
+      "loss": 0.5566,
+      "step": 14346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083290587330085e-05,
+      "loss": 0.5127,
+      "step": 14347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083226468325212e-05,
+      "loss": 0.6329,
+      "step": 14348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0831623493203385e-05,
+      "loss": 0.4916,
+      "step": 14349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083098230315466e-05,
+      "loss": 0.5674,
+      "step": 14350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083034111310592e-05,
+      "loss": 0.6095,
+      "step": 14351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0829699923057194e-05,
+      "loss": 0.5547,
+      "step": 14352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0829058733008465e-05,
+      "loss": 0.5397,
+      "step": 14353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082841754295974e-05,
+      "loss": 0.6095,
+      "step": 14354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082777635291101e-05,
+      "loss": 0.6202,
+      "step": 14355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0827135162862274e-05,
+      "loss": 0.545,
+      "step": 14356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0826493972813546e-05,
+      "loss": 0.5369,
+      "step": 14357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082585278276481e-05,
+      "loss": 0.4205,
+      "step": 14358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082521159271608e-05,
+      "loss": 0.5137,
+      "step": 14359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082457040266735e-05,
+      "loss": 0.5519,
+      "step": 14360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0823929212618626e-05,
+      "loss": 0.5996,
+      "step": 14361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08232880225699e-05,
+      "loss": 0.5084,
+      "step": 14362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082264683252116e-05,
+      "loss": 0.503,
+      "step": 14363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0822005642472434e-05,
+      "loss": 0.5964,
+      "step": 14364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08213644524237e-05,
+      "loss": 0.6975,
+      "step": 14365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082072326237497e-05,
+      "loss": 0.6685,
+      "step": 14366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0820082072326236e-05,
+      "loss": 0.5221,
+      "step": 14367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081944088227751e-05,
+      "loss": 0.5567,
+      "step": 14368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081879969222878e-05,
+      "loss": 0.6816,
+      "step": 14369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081815850218005e-05,
+      "loss": 0.5196,
+      "step": 14370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081751731213132e-05,
+      "loss": 0.6195,
+      "step": 14371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081687612208259e-05,
+      "loss": 0.5651,
+      "step": 14372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081623493203386e-05,
+      "loss": 0.5858,
+      "step": 14373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0815593741985124e-05,
+      "loss": 0.5253,
+      "step": 14374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0814952551936396e-05,
+      "loss": 0.6152,
+      "step": 14375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081431136188766e-05,
+      "loss": 0.621,
+      "step": 14376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081367017183893e-05,
+      "loss": 0.4604,
+      "step": 14377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0813028981790204e-05,
+      "loss": 0.5405,
+      "step": 14378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0812387791741476e-05,
+      "loss": 0.5574,
+      "step": 14379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081174660169275e-05,
+      "loss": 0.7167,
+      "step": 14380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081110541164401e-05,
+      "loss": 0.5668,
+      "step": 14381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0810464221595284e-05,
+      "loss": 0.5325,
+      "step": 14382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080982303154655e-05,
+      "loss": 0.5467,
+      "step": 14383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080918184149782e-05,
+      "loss": 0.494,
+      "step": 14384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080854065144909e-05,
+      "loss": 0.6176,
+      "step": 14385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0807899461400364e-05,
+      "loss": 0.474,
+      "step": 14386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080725827135163e-05,
+      "loss": 0.6368,
+      "step": 14387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08066170813029e-05,
+      "loss": 0.4229,
+      "step": 14388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080597589125417e-05,
+      "loss": 0.518,
+      "step": 14389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080533470120544e-05,
+      "loss": 0.5807,
+      "step": 14390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080469351115671e-05,
+      "loss": 0.5228,
+      "step": 14391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0804052321107974e-05,
+      "loss": 0.4584,
+      "step": 14392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0803411131059246e-05,
+      "loss": 0.5571,
+      "step": 14393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080276994101052e-05,
+      "loss": 0.5857,
+      "step": 14394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080212875096179e-05,
+      "loss": 0.5485,
+      "step": 14395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0801487560913054e-05,
+      "loss": 0.4969,
+      "step": 14396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0800846370864326e-05,
+      "loss": 0.4782,
+      "step": 14397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08002051808156e-05,
+      "loss": 0.5293,
+      "step": 14398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079956399076686e-05,
+      "loss": 0.4851,
+      "step": 14399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0798922800718135e-05,
+      "loss": 0.5896,
+      "step": 14400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0798281610669406e-05,
+      "loss": 0.5702,
+      "step": 14401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079764042062068e-05,
+      "loss": 0.5843,
+      "step": 14402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079699923057194e-05,
+      "loss": 0.5411,
+      "step": 14403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0796358040523215e-05,
+      "loss": 0.6047,
+      "step": 14404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079571685047448e-05,
+      "loss": 0.5083,
+      "step": 14405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079507566042575e-05,
+      "loss": 0.4831,
+      "step": 14406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079443447037702e-05,
+      "loss": 0.4656,
+      "step": 14407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079379328032829e-05,
+      "loss": 0.5098,
+      "step": 14408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079315209027956e-05,
+      "loss": 0.4834,
+      "step": 14409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079251090023083e-05,
+      "loss": 0.5121,
+      "step": 14410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.07918697101821e-05,
+      "loss": 0.5747,
+      "step": 14411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079122852013337e-05,
+      "loss": 0.6002,
+      "step": 14412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.079058733008464e-05,
+      "loss": 0.5601,
+      "step": 14413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0789946140035905e-05,
+      "loss": 0.4875,
+      "step": 14414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0789304949987176e-05,
+      "loss": 0.6266,
+      "step": 14415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078866375993845e-05,
+      "loss": 0.5695,
+      "step": 14416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078802256988972e-05,
+      "loss": 0.5022,
+      "step": 14417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078738137984099e-05,
+      "loss": 0.5004,
+      "step": 14418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0786740189792256e-05,
+      "loss": 0.5025,
+      "step": 14419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078609899974353e-05,
+      "loss": 0.6067,
+      "step": 14420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078545780969479e-05,
+      "loss": 0.51,
+      "step": 14421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0784816619646065e-05,
+      "loss": 0.5346,
+      "step": 14422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078417542959733e-05,
+      "loss": 0.6039,
+      "step": 14423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.07835342395486e-05,
+      "loss": 0.6061,
+      "step": 14424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078289304949987e-05,
+      "loss": 0.5539,
+      "step": 14425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0782251859451145e-05,
+      "loss": 0.6363,
+      "step": 14426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078161066940242e-05,
+      "loss": 0.509,
+      "step": 14427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078096947935368e-05,
+      "loss": 0.5882,
+      "step": 14428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.078032828930495e-05,
+      "loss": 0.5551,
+      "step": 14429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.077968709925622e-05,
+      "loss": 0.5127,
+      "step": 14430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.077904590920749e-05,
+      "loss": 0.5397,
+      "step": 14431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0778404719158755e-05,
+      "loss": 0.6084,
+      "step": 14432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0777763529110033e-05,
+      "loss": 0.4062,
+      "step": 14433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0777122339061305e-05,
+      "loss": 0.5383,
+      "step": 14434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.077648114901257e-05,
+      "loss": 0.5117,
+      "step": 14435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077583995896384e-05,
+      "loss": 0.5043,
+      "step": 14436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077519876891511e-05,
+      "loss": 0.5216,
+      "step": 14437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077455757886638e-05,
+      "loss": 0.455,
+      "step": 14438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0773916388817643e-05,
+      "loss": 0.6572,
+      "step": 14439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0773275198768915e-05,
+      "loss": 0.4757,
+      "step": 14440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077263400872019e-05,
+      "loss": 0.4636,
+      "step": 14441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077199281867146e-05,
+      "loss": 0.4513,
+      "step": 14442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077135162862273e-05,
+      "loss": 0.5815,
+      "step": 14443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0770710438573995e-05,
+      "loss": 0.4999,
+      "step": 14444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077006924852527e-05,
+      "loss": 0.5605,
+      "step": 14445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076942805847653e-05,
+      "loss": 0.5516,
+      "step": 14446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0768786868427804e-05,
+      "loss": 0.5677,
+      "step": 14447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076814567837907e-05,
+      "loss": 0.5899,
+      "step": 14448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076750448833035e-05,
+      "loss": 0.5813,
+      "step": 14449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076686329828161e-05,
+      "loss": 0.6369,
+      "step": 14450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0766222108232884e-05,
+      "loss": 0.5631,
+      "step": 14451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0765580918184155e-05,
+      "loss": 0.4261,
+      "step": 14452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076493972813542e-05,
+      "loss": 0.4991,
+      "step": 14453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076429853808669e-05,
+      "loss": 0.6553,
+      "step": 14454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076365734803796e-05,
+      "loss": 0.5791,
+      "step": 14455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076301615798923e-05,
+      "loss": 0.5405,
+      "step": 14456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07623749679405e-05,
+      "loss": 0.5013,
+      "step": 14457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076173377789177e-05,
+      "loss": 0.5622,
+      "step": 14458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076109258784304e-05,
+      "loss": 0.6419,
+      "step": 14459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076045139779431e-05,
+      "loss": 0.4766,
+      "step": 14460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075981020774558e-05,
+      "loss": 0.5476,
+      "step": 14461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0759169017696845e-05,
+      "loss": 0.4714,
+      "step": 14462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075852782764812e-05,
+      "loss": 0.6295,
+      "step": 14463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075788663759938e-05,
+      "loss": 0.4742,
+      "step": 14464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0757245447550654e-05,
+      "loss": 0.5698,
+      "step": 14465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0756604257501926e-05,
+      "loss": 0.5806,
+      "step": 14466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07559630674532e-05,
+      "loss": 0.5358,
+      "step": 14467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075532187740446e-05,
+      "loss": 0.5254,
+      "step": 14468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0754680687355734e-05,
+      "loss": 0.5118,
+      "step": 14469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0754039497307006e-05,
+      "loss": 0.6049,
+      "step": 14470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075339830725827e-05,
+      "loss": 0.4956,
+      "step": 14471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075275711720954e-05,
+      "loss": 0.6545,
+      "step": 14472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0752115927160814e-05,
+      "loss": 0.5347,
+      "step": 14473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0751474737112086e-05,
+      "loss": 0.4639,
+      "step": 14474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075083354706335e-05,
+      "loss": 0.6256,
+      "step": 14475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075019235701462e-05,
+      "loss": 0.438,
+      "step": 14476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074955116696589e-05,
+      "loss": 0.5374,
+      "step": 14477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074890997691716e-05,
+      "loss": 0.4977,
+      "step": 14478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074826878686843e-05,
+      "loss": 0.347,
+      "step": 14479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0747627596819696e-05,
+      "loss": 0.5677,
+      "step": 14480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074698640677097e-05,
+      "loss": 0.5461,
+      "step": 14481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074634521672224e-05,
+      "loss": 0.4717,
+      "step": 14482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074570402667351e-05,
+      "loss": 0.6237,
+      "step": 14483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0745062836624776e-05,
+      "loss": 0.5611,
+      "step": 14484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074442164657605e-05,
+      "loss": 0.4797,
+      "step": 14485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074378045652731e-05,
+      "loss": 0.4858,
+      "step": 14486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0743139266478584e-05,
+      "loss": 0.5038,
+      "step": 14487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0742498076429856e-05,
+      "loss": 0.5512,
+      "step": 14488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074185688638113e-05,
+      "loss": 0.3927,
+      "step": 14489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07412156963324e-05,
+      "loss": 0.701,
+      "step": 14490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0740574506283664e-05,
+      "loss": 0.4788,
+      "step": 14491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0739933316234936e-05,
+      "loss": 0.5605,
+      "step": 14492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07392921261862e-05,
+      "loss": 0.5666,
+      "step": 14493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073865093613747e-05,
+      "loss": 0.5506,
+      "step": 14494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073800974608874e-05,
+      "loss": 0.5033,
+      "step": 14495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073736855604001e-05,
+      "loss": 0.5447,
+      "step": 14496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073672736599128e-05,
+      "loss": 0.7116,
+      "step": 14497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073608617594255e-05,
+      "loss": 0.5943,
+      "step": 14498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0735444985893824e-05,
+      "loss": 0.4911,
+      "step": 14499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073480379584509e-05,
+      "loss": 0.5651,
+      "step": 14500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073416260579636e-05,
+      "loss": 0.5934,
+      "step": 14501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0733521415747626e-05,
+      "loss": 0.7392,
+      "step": 14502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07328802256989e-05,
+      "loss": 0.5336,
+      "step": 14503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073223903565016e-05,
+      "loss": 0.5289,
+      "step": 14504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073159784560144e-05,
+      "loss": 0.572,
+      "step": 14505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073095665555271e-05,
+      "loss": 0.4989,
+      "step": 14506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073031546550398e-05,
+      "loss": 0.5647,
+      "step": 14507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072967427545525e-05,
+      "loss": 0.5379,
+      "step": 14508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0729033085406514e-05,
+      "loss": 0.6117,
+      "step": 14509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0728391895357786e-05,
+      "loss": 0.5658,
+      "step": 14510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072775070530905e-05,
+      "loss": 0.4704,
+      "step": 14511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072710951526032e-05,
+      "loss": 0.5218,
+      "step": 14512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0726468325211595e-05,
+      "loss": 0.4864,
+      "step": 14513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0725827135162866e-05,
+      "loss": 0.4698,
+      "step": 14514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072518594511414e-05,
+      "loss": 0.4843,
+      "step": 14515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07245447550654e-05,
+      "loss": 0.5615,
+      "step": 14516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0723903565016675e-05,
+      "loss": 0.4837,
+      "step": 14517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072326237496794e-05,
+      "loss": 0.5851,
+      "step": 14518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072262118491921e-05,
+      "loss": 0.5631,
+      "step": 14519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0721979994870476e-05,
+      "loss": 0.5542,
+      "step": 14520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0721338804821755e-05,
+      "loss": 0.4085,
+      "step": 14521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072069761477302e-05,
+      "loss": 0.5203,
+      "step": 14522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.072005642472429e-05,
+      "loss": 0.5437,
+      "step": 14523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071941523467556e-05,
+      "loss": 0.4763,
+      "step": 14524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071877404462683e-05,
+      "loss": 0.494,
+      "step": 14525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07181328545781e-05,
+      "loss": 0.5031,
+      "step": 14526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0717491664529365e-05,
+      "loss": 0.5479,
+      "step": 14527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0716850474480636e-05,
+      "loss": 0.4336,
+      "step": 14528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071620928443191e-05,
+      "loss": 0.4905,
+      "step": 14529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071556809438318e-05,
+      "loss": 0.4915,
+      "step": 14530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071492690433445e-05,
+      "loss": 0.5426,
+      "step": 14531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0714285714285717e-05,
+      "loss": 0.6177,
+      "step": 14532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071364452423699e-05,
+      "loss": 0.5909,
+      "step": 14533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071300333418825e-05,
+      "loss": 0.5013,
+      "step": 14534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0712362144139525e-05,
+      "loss": 0.5791,
+      "step": 14535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071172095409079e-05,
+      "loss": 0.5072,
+      "step": 14536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071107976404206e-05,
+      "loss": 0.4902,
+      "step": 14537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071043857399333e-05,
+      "loss": 0.5476,
+      "step": 14538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0709797383944605e-05,
+      "loss": 0.4983,
+      "step": 14539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070915619389588e-05,
+      "loss": 0.4155,
+      "step": 14540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070851500384714e-05,
+      "loss": 0.5144,
+      "step": 14541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070787381379841e-05,
+      "loss": 0.5895,
+      "step": 14542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070723262374968e-05,
+      "loss": 0.5402,
+      "step": 14543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070659143370095e-05,
+      "loss": 0.4175,
+      "step": 14544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070595024365222e-05,
+      "loss": 0.6113,
+      "step": 14545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0705309053603493e-05,
+      "loss": 0.4417,
+      "step": 14546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070466786355476e-05,
+      "loss": 0.6375,
+      "step": 14547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070402667350603e-05,
+      "loss": 0.5133,
+      "step": 14548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07033854834573e-05,
+      "loss": 0.5677,
+      "step": 14549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070274429340857e-05,
+      "loss": 0.5355,
+      "step": 14550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070210310335984e-05,
+      "loss": 0.6027,
+      "step": 14551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0701461913311103e-05,
+      "loss": 0.6435,
+      "step": 14552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0700820723262375e-05,
+      "loss": 0.7756,
+      "step": 14553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070017953321365e-05,
+      "loss": 0.5034,
+      "step": 14554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069953834316492e-05,
+      "loss": 0.5319,
+      "step": 14555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0698897153116184e-05,
+      "loss": 0.5093,
+      "step": 14556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0698255963067455e-05,
+      "loss": 0.5255,
+      "step": 14557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069761477301873e-05,
+      "loss": 0.4688,
+      "step": 14558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069697358296999e-05,
+      "loss": 0.5527,
+      "step": 14559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0696332392921264e-05,
+      "loss": 0.4921,
+      "step": 14560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0695691202872535e-05,
+      "loss": 0.5144,
+      "step": 14561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069505001282381e-05,
+      "loss": 0.5032,
+      "step": 14562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069440882277507e-05,
+      "loss": 0.6189,
+      "step": 14563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0693767632726344e-05,
+      "loss": 0.4907,
+      "step": 14564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069312644267761e-05,
+      "loss": 0.5472,
+      "step": 14565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069248525262888e-05,
+      "loss": 0.5099,
+      "step": 14566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069184406258015e-05,
+      "loss": 0.5141,
+      "step": 14567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069120287253142e-05,
+      "loss": 0.466,
+      "step": 14568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069056168248269e-05,
+      "loss": 0.5046,
+      "step": 14569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068992049243396e-05,
+      "loss": 0.5409,
+      "step": 14570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068927930238523e-05,
+      "loss": 0.457,
+      "step": 14571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.06886381123365e-05,
+      "loss": 0.5365,
+      "step": 14572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068799692228777e-05,
+      "loss": 0.5646,
+      "step": 14573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0687355732239034e-05,
+      "loss": 0.4474,
+      "step": 14574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0686714542190305e-05,
+      "loss": 0.4835,
+      "step": 14575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068607335214158e-05,
+      "loss": 0.5193,
+      "step": 14576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068543216209285e-05,
+      "loss": 0.6077,
+      "step": 14577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068479097204412e-05,
+      "loss": 0.5116,
+      "step": 14578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0684149781995386e-05,
+      "loss": 0.555,
+      "step": 14579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068350859194666e-05,
+      "loss": 0.4665,
+      "step": 14580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068286740189792e-05,
+      "loss": 0.4356,
+      "step": 14581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0682226211849194e-05,
+      "loss": 0.69,
+      "step": 14582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068158502180046e-05,
+      "loss": 0.5718,
+      "step": 14583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068094383175173e-05,
+      "loss": 0.5575,
+      "step": 14584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0680302641703e-05,
+      "loss": 0.5773,
+      "step": 14585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0679661451654274e-05,
+      "loss": 0.4445,
+      "step": 14586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0679020261605546e-05,
+      "loss": 0.5177,
+      "step": 14587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067837907155681e-05,
+      "loss": 0.5304,
+      "step": 14588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067773788150808e-05,
+      "loss": 0.408,
+      "step": 14589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067709669145935e-05,
+      "loss": 0.6015,
+      "step": 14590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067645550141062e-05,
+      "loss": 0.4551,
+      "step": 14591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0675814311361884e-05,
+      "loss": 0.5016,
+      "step": 14592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067517312131316e-05,
+      "loss": 0.5263,
+      "step": 14593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0674531931264434e-05,
+      "loss": 0.4819,
+      "step": 14594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.06738907412157e-05,
+      "loss": 0.606,
+      "step": 14595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067324955116697e-05,
+      "loss": 0.6641,
+      "step": 14596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0672608361118236e-05,
+      "loss": 0.4715,
+      "step": 14597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067196717106951e-05,
+      "loss": 0.5096,
+      "step": 14598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.067132598102077e-05,
+      "loss": 0.5534,
+      "step": 14599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0670684790972044e-05,
+      "loss": 0.4836,
+      "step": 14600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0670043600923316e-05,
+      "loss": 0.5738,
+      "step": 14601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066940241087459e-05,
+      "loss": 0.5,
+      "step": 14602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066876122082586e-05,
+      "loss": 0.4379,
+      "step": 14603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0668120030777124e-05,
+      "loss": 0.4781,
+      "step": 14604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0667478840728396e-05,
+      "loss": 0.5116,
+      "step": 14605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066683765067966e-05,
+      "loss": 0.5725,
+      "step": 14606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066619646063093e-05,
+      "loss": 0.5066,
+      "step": 14607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.06655552705822e-05,
+      "loss": 0.4825,
+      "step": 14608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0664914080533476e-05,
+      "loss": 0.464,
+      "step": 14609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066427289048474e-05,
+      "loss": 0.6205,
+      "step": 14610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066363170043601e-05,
+      "loss": 0.5732,
+      "step": 14611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0662990510387284e-05,
+      "loss": 0.4832,
+      "step": 14612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066234932033855e-05,
+      "loss": 0.5584,
+      "step": 14613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066170813028982e-05,
+      "loss": 0.5284,
+      "step": 14614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0661066940241086e-05,
+      "loss": 0.5898,
+      "step": 14615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066042575019236e-05,
+      "loss": 0.4908,
+      "step": 14616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065978456014363e-05,
+      "loss": 0.496,
+      "step": 14617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.06591433700949e-05,
+      "loss": 0.4733,
+      "step": 14618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0658502180046166e-05,
+      "loss": 0.5899,
+      "step": 14619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065786098999744e-05,
+      "loss": 0.5847,
+      "step": 14620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065721979994871e-05,
+      "loss": 0.5247,
+      "step": 14621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0656578609899975e-05,
+      "loss": 0.5771,
+      "step": 14622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0655937419851246e-05,
+      "loss": 0.5038,
+      "step": 14623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065529622980251e-05,
+      "loss": 0.5681,
+      "step": 14624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065465503975378e-05,
+      "loss": 0.5986,
+      "step": 14625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0654013849705055e-05,
+      "loss": 0.4701,
+      "step": 14626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0653372659656326e-05,
+      "loss": 0.3904,
+      "step": 14627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065273146960759e-05,
+      "loss": 0.5965,
+      "step": 14628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065209027955886e-05,
+      "loss": 0.7199,
+      "step": 14629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0651449089510135e-05,
+      "loss": 0.5869,
+      "step": 14630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.06508078994614e-05,
+      "loss": 0.5275,
+      "step": 14631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065016670941267e-05,
+      "loss": 0.462,
+      "step": 14632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064952551936394e-05,
+      "loss": 0.4875,
+      "step": 14633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0648884329315215e-05,
+      "loss": 0.4158,
+      "step": 14634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064824313926648e-05,
+      "loss": 0.6053,
+      "step": 14635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064760194921775e-05,
+      "loss": 0.565,
+      "step": 14636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0646960759169016e-05,
+      "loss": 0.5015,
+      "step": 14637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064631956912029e-05,
+      "loss": 0.57,
+      "step": 14638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064567837907156e-05,
+      "loss": 0.5038,
+      "step": 14639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0645037189022825e-05,
+      "loss": 0.442,
+      "step": 14640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0644395998974096e-05,
+      "loss": 0.5875,
+      "step": 14641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064375480892537e-05,
+      "loss": 0.5076,
+      "step": 14642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064311361887664e-05,
+      "loss": 0.5352,
+      "step": 14643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0642472428827905e-05,
+      "loss": 0.5008,
+      "step": 14644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0641831238779177e-05,
+      "loss": 0.4247,
+      "step": 14645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064119004873044e-05,
+      "loss": 0.4926,
+      "step": 14646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064054885868171e-05,
+      "loss": 0.6171,
+      "step": 14647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0639907668632985e-05,
+      "loss": 0.6929,
+      "step": 14648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063926647858426e-05,
+      "loss": 0.6001,
+      "step": 14649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063862528853553e-05,
+      "loss": 0.4807,
+      "step": 14650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063798409848679e-05,
+      "loss": 0.4698,
+      "step": 14651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0637342908438065e-05,
+      "loss": 0.5524,
+      "step": 14652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063670171838933e-05,
+      "loss": 0.5183,
+      "step": 14653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.06360605283406e-05,
+      "loss": 0.5955,
+      "step": 14654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063541933829187e-05,
+      "loss": 0.446,
+      "step": 14655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063477814824314e-05,
+      "loss": 0.5835,
+      "step": 14656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063413695819441e-05,
+      "loss": 0.4392,
+      "step": 14657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063349576814568e-05,
+      "loss": 0.5626,
+      "step": 14658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0632854578096953e-05,
+      "loss": 0.5783,
+      "step": 14659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063221338804822e-05,
+      "loss": 0.5189,
+      "step": 14660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063157219799949e-05,
+      "loss": 0.5486,
+      "step": 14661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0630931007950755e-05,
+      "loss": 0.489,
+      "step": 14662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063028981790203e-05,
+      "loss": 0.4993,
+      "step": 14663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062964862785329e-05,
+      "loss": 0.428,
+      "step": 14664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062900743780457e-05,
+      "loss": 0.4376,
+      "step": 14665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062836624775584e-05,
+      "loss": 0.673,
+      "step": 14666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062772505770711e-05,
+      "loss": 0.4366,
+      "step": 14667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062708386765838e-05,
+      "loss": 0.5594,
+      "step": 14668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0626442677609644e-05,
+      "loss": 0.6635,
+      "step": 14669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0625801487560915e-05,
+      "loss": 0.635,
+      "step": 14670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062516029751218e-05,
+      "loss": 0.3875,
+      "step": 14671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062451910746345e-05,
+      "loss": 0.6273,
+      "step": 14672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0623877917414724e-05,
+      "loss": 0.5673,
+      "step": 14673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0623236727365995e-05,
+      "loss": 0.6271,
+      "step": 14674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062259553731727e-05,
+      "loss": 0.4352,
+      "step": 14675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062195434726853e-05,
+      "loss": 0.4935,
+      "step": 14676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0621313157219804e-05,
+      "loss": 0.5974,
+      "step": 14677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062067196717107e-05,
+      "loss": 0.5302,
+      "step": 14678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.062003077712234e-05,
+      "loss": 0.4958,
+      "step": 14679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0619389587073605e-05,
+      "loss": 0.4447,
+      "step": 14680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0618748397024884e-05,
+      "loss": 0.4852,
+      "step": 14681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061810720697615e-05,
+      "loss": 0.4864,
+      "step": 14682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061746601692742e-05,
+      "loss": 0.5178,
+      "step": 14683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061682482687869e-05,
+      "loss": 0.4525,
+      "step": 14684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061618363682996e-05,
+      "loss": 0.5899,
+      "step": 14685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061554244678123e-05,
+      "loss": 0.6074,
+      "step": 14686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0614901256732494e-05,
+      "loss": 0.5746,
+      "step": 14687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0614260066683766e-05,
+      "loss": 0.6107,
+      "step": 14688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061361887663504e-05,
+      "loss": 0.5928,
+      "step": 14689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061297768658631e-05,
+      "loss": 0.4859,
+      "step": 14690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0612336496537574e-05,
+      "loss": 0.5333,
+      "step": 14691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0611695306488846e-05,
+      "loss": 0.4674,
+      "step": 14692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061105411644012e-05,
+      "loss": 0.5518,
+      "step": 14693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.061041292639138e-05,
+      "loss": 0.5052,
+      "step": 14694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0609771736342654e-05,
+      "loss": 0.5084,
+      "step": 14695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060913054629392e-05,
+      "loss": 0.6132,
+      "step": 14696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060848935624519e-05,
+      "loss": 0.511,
+      "step": 14697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060784816619646e-05,
+      "loss": 0.4383,
+      "step": 14698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0607206976147734e-05,
+      "loss": 0.4774,
+      "step": 14699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0606565786099e-05,
+      "loss": 0.5314,
+      "step": 14700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060592459605027e-05,
+      "loss": 0.5966,
+      "step": 14701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060528340600154e-05,
+      "loss": 0.5434,
+      "step": 14702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060464221595281e-05,
+      "loss": 0.692,
+      "step": 14703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060400102590408e-05,
+      "loss": 0.5304,
+      "step": 14704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060335983585535e-05,
+      "loss": 0.4189,
+      "step": 14705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060271864580662e-05,
+      "loss": 0.4908,
+      "step": 14706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060207745575789e-05,
+      "loss": 0.4834,
+      "step": 14707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060143626570916e-05,
+      "loss": 0.4767,
+      "step": 14708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0600795075660424e-05,
+      "loss": 0.5159,
+      "step": 14709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0600153885611696e-05,
+      "loss": 0.3904,
+      "step": 14710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059951269556297e-05,
+      "loss": 0.4474,
+      "step": 14711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059887150551423e-05,
+      "loss": 0.6198,
+      "step": 14712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0598230315465504e-05,
+      "loss": 0.4335,
+      "step": 14713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0597589125416776e-05,
+      "loss": 0.4905,
+      "step": 14714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059694793536805e-05,
+      "loss": 0.5585,
+      "step": 14715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059630674531931e-05,
+      "loss": 0.5595,
+      "step": 14716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0595665555270584e-05,
+      "loss": 0.388,
+      "step": 14717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059502436522185e-05,
+      "loss": 0.6083,
+      "step": 14718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059438317517312e-05,
+      "loss": 0.5177,
+      "step": 14719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059374198512439e-05,
+      "loss": 0.5566,
+      "step": 14720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0593100795075664e-05,
+      "loss": 0.5623,
+      "step": 14721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0592459605026936e-05,
+      "loss": 0.5487,
+      "step": 14722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05918184149782e-05,
+      "loss": 0.4545,
+      "step": 14723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059117722492947e-05,
+      "loss": 0.6777,
+      "step": 14724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059053603488074e-05,
+      "loss": 0.489,
+      "step": 14725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058989484483201e-05,
+      "loss": 0.4691,
+      "step": 14726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058925365478328e-05,
+      "loss": 0.6134,
+      "step": 14727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0588612464734546e-05,
+      "loss": 0.4739,
+      "step": 14728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058797127468582e-05,
+      "loss": 0.4799,
+      "step": 14729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058733008463709e-05,
+      "loss": 0.5663,
+      "step": 14730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058668889458836e-05,
+      "loss": 0.4367,
+      "step": 14731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0586047704539626e-05,
+      "loss": 0.5346,
+      "step": 14732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05854065144909e-05,
+      "loss": 0.5645,
+      "step": 14733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058476532444216e-05,
+      "loss": 0.5393,
+      "step": 14734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0584124134393435e-05,
+      "loss": 0.5397,
+      "step": 14735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0583482944344706e-05,
+      "loss": 0.4105,
+      "step": 14736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058284175429598e-05,
+      "loss": 0.5781,
+      "step": 14737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058220056424725e-05,
+      "loss": 0.4057,
+      "step": 14738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0581559374198515e-05,
+      "loss": 0.4119,
+      "step": 14739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0580918184149786e-05,
+      "loss": 0.5762,
+      "step": 14740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058027699410105e-05,
+      "loss": 0.4834,
+      "step": 14741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057963580405232e-05,
+      "loss": 0.4354,
+      "step": 14742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057899461400359e-05,
+      "loss": 0.433,
+      "step": 14743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057835342395486e-05,
+      "loss": 0.5745,
+      "step": 14744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057771223390613e-05,
+      "loss": 0.5303,
+      "step": 14745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05770710438574e-05,
+      "loss": 0.5415,
+      "step": 14746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0576429853808675e-05,
+      "loss": 0.5984,
+      "step": 14747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057578866375994e-05,
+      "loss": 0.5516,
+      "step": 14748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057514747371121e-05,
+      "loss": 0.6037,
+      "step": 14749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0574506283662476e-05,
+      "loss": 0.4659,
+      "step": 14750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057386509361375e-05,
+      "loss": 0.5408,
+      "step": 14751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057322390356501e-05,
+      "loss": 0.5744,
+      "step": 14752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057258271351629e-05,
+      "loss": 0.469,
+      "step": 14753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057194152346756e-05,
+      "loss": 0.6256,
+      "step": 14754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057130033341883e-05,
+      "loss": 0.4978,
+      "step": 14755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05706591433701e-05,
+      "loss": 0.5007,
+      "step": 14756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0570017953321365e-05,
+      "loss": 0.432,
+      "step": 14757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0569376763272637e-05,
+      "loss": 0.5198,
+      "step": 14758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05687355732239e-05,
+      "loss": 0.5101,
+      "step": 14759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056809438317517e-05,
+      "loss": 0.5989,
+      "step": 14760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0567453193126445e-05,
+      "loss": 0.4709,
+      "step": 14761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056681200307772e-05,
+      "loss": 0.5365,
+      "step": 14762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056617081302899e-05,
+      "loss": 0.5638,
+      "step": 14763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056552962298025e-05,
+      "loss": 0.5681,
+      "step": 14764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0564888432931525e-05,
+      "loss": 0.5109,
+      "step": 14765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056424724288279e-05,
+      "loss": 0.4822,
+      "step": 14766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056360605283406e-05,
+      "loss": 0.5907,
+      "step": 14767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056296486278533e-05,
+      "loss": 0.5234,
+      "step": 14768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0562323672736605e-05,
+      "loss": 0.5383,
+      "step": 14769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056168248268787e-05,
+      "loss": 0.491,
+      "step": 14770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056104129263914e-05,
+      "loss": 0.5904,
+      "step": 14771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0560400102590414e-05,
+      "loss": 0.563,
+      "step": 14772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055975891254168e-05,
+      "loss": 0.5404,
+      "step": 14773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055911772249295e-05,
+      "loss": 0.5034,
+      "step": 14774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0558476532444215e-05,
+      "loss": 0.4859,
+      "step": 14775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055783534239549e-05,
+      "loss": 0.6769,
+      "step": 14776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055719415234676e-05,
+      "loss": 0.5275,
+      "step": 14777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055655296229803e-05,
+      "loss": 0.4635,
+      "step": 14778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0555911772249295e-05,
+      "loss": 0.555,
+      "step": 14779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055527058220057e-05,
+      "loss": 0.6537,
+      "step": 14780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055462939215184e-05,
+      "loss": 0.4616,
+      "step": 14781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0553988202103104e-05,
+      "loss": 0.5025,
+      "step": 14782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0553347012054375e-05,
+      "loss": 0.5569,
+      "step": 14783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055270582200564e-05,
+      "loss": 0.5937,
+      "step": 14784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055206463195691e-05,
+      "loss": 0.5195,
+      "step": 14785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0551423441908184e-05,
+      "loss": 0.5144,
+      "step": 14786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0550782251859455e-05,
+      "loss": 0.6228,
+      "step": 14787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055014106181072e-05,
+      "loss": 0.467,
+      "step": 14788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054949987176199e-05,
+      "loss": 0.544,
+      "step": 14789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0548858681713264e-05,
+      "loss": 0.4795,
+      "step": 14790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054821749166453e-05,
+      "loss": 0.5158,
+      "step": 14791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05475763016158e-05,
+      "loss": 0.5023,
+      "step": 14792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054693511156707e-05,
+      "loss": 0.4217,
+      "step": 14793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0546293921518344e-05,
+      "loss": 0.5723,
+      "step": 14794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054565273146961e-05,
+      "loss": 0.5331,
+      "step": 14795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054501154142088e-05,
+      "loss": 0.4895,
+      "step": 14796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0544370351372145e-05,
+      "loss": 0.5283,
+      "step": 14797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054372916132342e-05,
+      "loss": 0.5408,
+      "step": 14798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054308797127469e-05,
+      "loss": 0.5817,
+      "step": 14799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0542446781225954e-05,
+      "loss": 0.5226,
+      "step": 14800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0541805591177226e-05,
+      "loss": 0.5793,
+      "step": 14801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05411644011285e-05,
+      "loss": 0.6567,
+      "step": 14802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054052321107977e-05,
+      "loss": 0.6667,
+      "step": 14803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0539882021031034e-05,
+      "loss": 0.5882,
+      "step": 14804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0539240830982306e-05,
+      "loss": 0.5225,
+      "step": 14805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053859964093357e-05,
+      "loss": 0.5732,
+      "step": 14806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053795845088484e-05,
+      "loss": 0.5031,
+      "step": 14807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0537317260836114e-05,
+      "loss": 0.5368,
+      "step": 14808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0536676070787386e-05,
+      "loss": 0.5288,
+      "step": 14809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053603488073866e-05,
+      "loss": 0.4312,
+      "step": 14810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053539369068992e-05,
+      "loss": 0.5146,
+      "step": 14811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0534752500641194e-05,
+      "loss": 0.5032,
+      "step": 14812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053411131059246e-05,
+      "loss": 0.5203,
+      "step": 14813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053347012054373e-05,
+      "loss": 0.5901,
+      "step": 14814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0532828930494996e-05,
+      "loss": 0.5682,
+      "step": 14815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053218774044627e-05,
+      "loss": 0.4755,
+      "step": 14816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053154655039754e-05,
+      "loss": 0.5596,
+      "step": 14817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053090536034881e-05,
+      "loss": 0.5102,
+      "step": 14818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053026417030008e-05,
+      "loss": 0.5188,
+      "step": 14819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052962298025135e-05,
+      "loss": 0.6224,
+      "step": 14820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052898179020262e-05,
+      "loss": 0.524,
+      "step": 14821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0528340600153884e-05,
+      "loss": 0.6225,
+      "step": 14822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0527699410105156e-05,
+      "loss": 0.5727,
+      "step": 14823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052705822005642e-05,
+      "loss": 0.5273,
+      "step": 14824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05264170300077e-05,
+      "loss": 0.6563,
+      "step": 14825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052577583995897e-05,
+      "loss": 0.4824,
+      "step": 14826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0525134649910236e-05,
+      "loss": 0.5197,
+      "step": 14827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052449345986151e-05,
+      "loss": 0.5169,
+      "step": 14828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052385226981277e-05,
+      "loss": 0.4969,
+      "step": 14829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0523211079764044e-05,
+      "loss": 0.5233,
+      "step": 14830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052256988971531e-05,
+      "loss": 0.4509,
+      "step": 14831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052192869966658e-05,
+      "loss": 0.5939,
+      "step": 14832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052128750961785e-05,
+      "loss": 0.4774,
+      "step": 14833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0520646319569124e-05,
+      "loss": 0.6082,
+      "step": 14834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0520005129520396e-05,
+      "loss": 0.4735,
+      "step": 14835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051936393947166e-05,
+      "loss": 0.4869,
+      "step": 14836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051872274942293e-05,
+      "loss": 0.604,
+      "step": 14837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05180815593742e-05,
+      "loss": 0.5774,
+      "step": 14838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051744036932547e-05,
+      "loss": 0.5874,
+      "step": 14839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0516799179276734e-05,
+      "loss": 0.5056,
+      "step": 14840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051615798922801e-05,
+      "loss": 0.3865,
+      "step": 14841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051551679917928e-05,
+      "loss": 0.5192,
+      "step": 14842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051487560913055e-05,
+      "loss": 0.504,
+      "step": 14843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051423441908182e-05,
+      "loss": 0.456,
+      "step": 14844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0513593229033086e-05,
+      "loss": 0.6041,
+      "step": 14845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051295203898436e-05,
+      "loss": 0.4224,
+      "step": 14846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051231084893562e-05,
+      "loss": 0.4988,
+      "step": 14847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0511669658886895e-05,
+      "loss": 0.5219,
+      "step": 14848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0511028468838166e-05,
+      "loss": 0.4235,
+      "step": 14849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051038727878944e-05,
+      "loss": 0.4629,
+      "step": 14850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05097460887407e-05,
+      "loss": 0.6152,
+      "step": 14851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0509104898691975e-05,
+      "loss": 0.4912,
+      "step": 14852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0508463708643246e-05,
+      "loss": 0.6376,
+      "step": 14853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050782251859451e-05,
+      "loss": 0.4105,
+      "step": 14854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050718132854578e-05,
+      "loss": 0.5854,
+      "step": 14855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050654013849705e-05,
+      "loss": 0.5578,
+      "step": 14856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0505898948448326e-05,
+      "loss": 0.5442,
+      "step": 14857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050525775839959e-05,
+      "loss": 0.6673,
+      "step": 14858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050461656835086e-05,
+      "loss": 0.5369,
+      "step": 14859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050397537830213e-05,
+      "loss": 0.4821,
+      "step": 14860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05033341882534e-05,
+      "loss": 0.4559,
+      "step": 14861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050269299820467e-05,
+      "loss": 0.6437,
+      "step": 14862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0502051808155936e-05,
+      "loss": 0.6153,
+      "step": 14863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050141061810721e-05,
+      "loss": 0.5897,
+      "step": 14864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050076942805848e-05,
+      "loss": 0.4603,
+      "step": 14865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050012823800975e-05,
+      "loss": 0.5825,
+      "step": 14866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0499487047961017e-05,
+      "loss": 0.5936,
+      "step": 14867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049884585791229e-05,
+      "loss": 0.5242,
+      "step": 14868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049820466786355e-05,
+      "loss": 0.582,
+      "step": 14869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0497563477814825e-05,
+      "loss": 0.5189,
+      "step": 14870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04969222877661e-05,
+      "loss": 0.5916,
+      "step": 14871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049628109771736e-05,
+      "loss": 0.5043,
+      "step": 14872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049563990766863e-05,
+      "loss": 0.5037,
+      "step": 14873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0494998717619905e-05,
+      "loss": 0.6189,
+      "step": 14874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049435752757118e-05,
+      "loss": 0.588,
+      "step": 14875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049371633752244e-05,
+      "loss": 0.5111,
+      "step": 14876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049307514747371e-05,
+      "loss": 0.5661,
+      "step": 14877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049243395742498e-05,
+      "loss": 0.511,
+      "step": 14878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049179276737625e-05,
+      "loss": 0.5431,
+      "step": 14879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049115157732752e-05,
+      "loss": 0.4778,
+      "step": 14880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0490510387278793e-05,
+      "loss": 0.5178,
+      "step": 14881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0489869197230065e-05,
+      "loss": 0.645,
+      "step": 14882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048922800718133e-05,
+      "loss": 0.5356,
+      "step": 14883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04885868171326e-05,
+      "loss": 0.5322,
+      "step": 14884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048794562708387e-05,
+      "loss": 0.5832,
+      "step": 14885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048730443703514e-05,
+      "loss": 0.5819,
+      "step": 14886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0486663246986403e-05,
+      "loss": 0.5743,
+      "step": 14887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0486022056937675e-05,
+      "loss": 0.5545,
+      "step": 14888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048538086688895e-05,
+      "loss": 0.5423,
+      "step": 14889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048473967684022e-05,
+      "loss": 0.5418,
+      "step": 14890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048409848679149e-05,
+      "loss": 0.4664,
+      "step": 14891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0483457296742755e-05,
+      "loss": 0.5358,
+      "step": 14892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048281610669403e-05,
+      "loss": 0.4539,
+      "step": 14893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048217491664529e-05,
+      "loss": 0.4742,
+      "step": 14894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0481533726596564e-05,
+      "loss": 0.4546,
+      "step": 14895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0480892536547835e-05,
+      "loss": 0.6352,
+      "step": 14896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048025134649911e-05,
+      "loss": 0.444,
+      "step": 14897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047961015645038e-05,
+      "loss": 0.4971,
+      "step": 14898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0478968966401644e-05,
+      "loss": 0.5284,
+      "step": 14899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0478327776352915e-05,
+      "loss": 0.5505,
+      "step": 14900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047768658630418e-05,
+      "loss": 0.5126,
+      "step": 14901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047704539625545e-05,
+      "loss": 0.4721,
+      "step": 14902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047640420620672e-05,
+      "loss": 0.6113,
+      "step": 14903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047576301615799e-05,
+      "loss": 0.5906,
+      "step": 14904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047512182610926e-05,
+      "loss": 0.5385,
+      "step": 14905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047448063606053e-05,
+      "loss": 0.619,
+      "step": 14906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0473839446011804e-05,
+      "loss": 0.5557,
+      "step": 14907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047319825596307e-05,
+      "loss": 0.4663,
+      "step": 14908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047255706591434e-05,
+      "loss": 0.4605,
+      "step": 14909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0471915875865605e-05,
+      "loss": 0.4545,
+      "step": 14910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047127468581688e-05,
+      "loss": 0.5268,
+      "step": 14911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047063349576814e-05,
+      "loss": 0.5279,
+      "step": 14912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046999230571942e-05,
+      "loss": 0.4244,
+      "step": 14913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046935111567069e-05,
+      "loss": 0.503,
+      "step": 14914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046870992562196e-05,
+      "loss": 0.5039,
+      "step": 14915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046806873557323e-05,
+      "loss": 0.5302,
+      "step": 14916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0467427545524494e-05,
+      "loss": 0.5224,
+      "step": 14917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0466786355475766e-05,
+      "loss": 0.5091,
+      "step": 14918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046614516542703e-05,
+      "loss": 0.5228,
+      "step": 14919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04655039753783e-05,
+      "loss": 0.4241,
+      "step": 14920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0464862785329574e-05,
+      "loss": 0.5867,
+      "step": 14921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0464221595280846e-05,
+      "loss": 0.6403,
+      "step": 14922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046358040523212e-05,
+      "loss": 0.6416,
+      "step": 14923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046293921518338e-05,
+      "loss": 0.4014,
+      "step": 14924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0462298025134654e-05,
+      "loss": 0.6851,
+      "step": 14925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046165683508592e-05,
+      "loss": 0.4269,
+      "step": 14926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046101564503719e-05,
+      "loss": 0.446,
+      "step": 14927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0460374454988456e-05,
+      "loss": 0.4816,
+      "step": 14928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0459733264939734e-05,
+      "loss": 0.6044,
+      "step": 14929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0459092074891e-05,
+      "loss": 0.6373,
+      "step": 14930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045845088484227e-05,
+      "loss": 0.5654,
+      "step": 14931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045780969479354e-05,
+      "loss": 0.5287,
+      "step": 14932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045716850474481e-05,
+      "loss": 0.5878,
+      "step": 14933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045652731469608e-05,
+      "loss": 0.483,
+      "step": 14934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0455886124647344e-05,
+      "loss": 0.5526,
+      "step": 14935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0455244934598616e-05,
+      "loss": 0.462,
+      "step": 14936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045460374454989e-05,
+      "loss": 0.5045,
+      "step": 14937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045396255450116e-05,
+      "loss": 0.4549,
+      "step": 14938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0453321364452424e-05,
+      "loss": 0.5373,
+      "step": 14939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0452680174403696e-05,
+      "loss": 0.5385,
+      "step": 14940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045203898435497e-05,
+      "loss": 0.5145,
+      "step": 14941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045139779430623e-05,
+      "loss": 0.5556,
+      "step": 14942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0450756604257504e-05,
+      "loss": 0.4603,
+      "step": 14943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045011541420877e-05,
+      "loss": 0.6672,
+      "step": 14944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044947422416004e-05,
+      "loss": 0.5125,
+      "step": 14945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044883303411131e-05,
+      "loss": 0.6032,
+      "step": 14946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0448191844062584e-05,
+      "loss": 0.6427,
+      "step": 14947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044755065401385e-05,
+      "loss": 0.5063,
+      "step": 14948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044690946396512e-05,
+      "loss": 0.4801,
+      "step": 14949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044626827391639e-05,
+      "loss": 0.5243,
+      "step": 14950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044562708386766e-05,
+      "loss": 0.5011,
+      "step": 14951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044498589381893e-05,
+      "loss": 0.4673,
+      "step": 14952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04443447037702e-05,
+      "loss": 0.6014,
+      "step": 14953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044370351372147e-05,
+      "loss": 0.5287,
+      "step": 14954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044306232367274e-05,
+      "loss": 0.5732,
+      "step": 14955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.044242113362401e-05,
+      "loss": 0.5166,
+      "step": 14956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0441779943575275e-05,
+      "loss": 0.6071,
+      "step": 14957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0441138753526546e-05,
+      "loss": 0.5964,
+      "step": 14958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.044049756347782e-05,
+      "loss": 0.5099,
+      "step": 14959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043985637342908e-05,
+      "loss": 0.5304,
+      "step": 14960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0439215183380355e-05,
+      "loss": 0.4615,
+      "step": 14961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0438573993331626e-05,
+      "loss": 0.5725,
+      "step": 14962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04379328032829e-05,
+      "loss": 0.6344,
+      "step": 14963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043729161323416e-05,
+      "loss": 0.6011,
+      "step": 14964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0436650423185435e-05,
+      "loss": 0.4335,
+      "step": 14965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04360092331367e-05,
+      "loss": 0.5228,
+      "step": 14966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043536804308797e-05,
+      "loss": 0.5483,
+      "step": 14967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043472685303924e-05,
+      "loss": 0.4389,
+      "step": 14968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0434085662990515e-05,
+      "loss": 0.5645,
+      "step": 14969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0433444472941787e-05,
+      "loss": 0.4414,
+      "step": 14970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043280328289305e-05,
+      "loss": 0.5303,
+      "step": 14971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043216209284432e-05,
+      "loss": 0.4918,
+      "step": 14972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043152090279559e-05,
+      "loss": 0.6301,
+      "step": 14973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043087971274686e-05,
+      "loss": 0.5172,
+      "step": 14974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0430238522698125e-05,
+      "loss": 0.5737,
+      "step": 14975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0429597332649396e-05,
+      "loss": 0.4533,
+      "step": 14976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042895614260067e-05,
+      "loss": 0.5379,
+      "step": 14977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042831495255194e-05,
+      "loss": 0.6217,
+      "step": 14978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042767376250321e-05,
+      "loss": 0.5998,
+      "step": 14979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0427032572454477e-05,
+      "loss": 0.4695,
+      "step": 14980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042639138240575e-05,
+      "loss": 0.5757,
+      "step": 14981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042575019235701e-05,
+      "loss": 0.544,
+      "step": 14982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0425109002308285e-05,
+      "loss": 0.5496,
+      "step": 14983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042446781225955e-05,
+      "loss": 0.4888,
+      "step": 14984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042382662221083e-05,
+      "loss": 0.5857,
+      "step": 14985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04231854321621e-05,
+      "loss": 0.4849,
+      "step": 14986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0422544242113365e-05,
+      "loss": 0.7051,
+      "step": 14987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042190305206464e-05,
+      "loss": 0.5775,
+      "step": 14988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04212618620159e-05,
+      "loss": 0.6401,
+      "step": 14989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0420620671967173e-05,
+      "loss": 0.6221,
+      "step": 14990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041997948191844e-05,
+      "loss": 0.5254,
+      "step": 14991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041933829186971e-05,
+      "loss": 0.3985,
+      "step": 14992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041869710182098e-05,
+      "loss": 0.5938,
+      "step": 14993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0418055911772253e-05,
+      "loss": 0.5437,
+      "step": 14994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0417414721723525e-05,
+      "loss": 0.5063,
+      "step": 14995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041677353167479e-05,
+      "loss": 0.568,
+      "step": 14996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041613234162606e-05,
+      "loss": 0.5752,
+      "step": 14997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041549115157733e-05,
+      "loss": 0.4758,
+      "step": 14998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04148499615286e-05,
+      "loss": 0.5717,
+      "step": 14999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0414208771479863e-05,
+      "loss": 0.528,
+      "step": 15000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041356758143114e-05,
+      "loss": 0.5857,
+      "step": 15001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041292639138241e-05,
+      "loss": 0.5069,
+      "step": 15002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041228520133368e-05,
+      "loss": 0.5461,
+      "step": 15003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041164401128495e-05,
+      "loss": 0.6366,
+      "step": 15004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0411002821236215e-05,
+      "loss": 0.4952,
+      "step": 15005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041036163118749e-05,
+      "loss": 0.4906,
+      "step": 15006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040972044113875e-05,
+      "loss": 0.5209,
+      "step": 15007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0409079251090024e-05,
+      "loss": 0.4873,
+      "step": 15008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0408438061041295e-05,
+      "loss": 0.5341,
+      "step": 15009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040779687099257e-05,
+      "loss": 0.524,
+      "step": 15010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040715568094383e-05,
+      "loss": 0.5268,
+      "step": 15011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0406514490895104e-05,
+      "loss": 0.4905,
+      "step": 15012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0405873300846375e-05,
+      "loss": 0.471,
+      "step": 15013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040523211079764e-05,
+      "loss": 0.5432,
+      "step": 15014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040459092074891e-05,
+      "loss": 0.505,
+      "step": 15015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040394973070018e-05,
+      "loss": 0.467,
+      "step": 15016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0403308540651456e-05,
+      "loss": 0.5419,
+      "step": 15017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040266735060272e-05,
+      "loss": 0.5864,
+      "step": 15018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040202616055399e-05,
+      "loss": 0.5855,
+      "step": 15019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040138497050526e-05,
+      "loss": 0.67,
+      "step": 15020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040074378045653e-05,
+      "loss": 0.5526,
+      "step": 15021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04001025904078e-05,
+      "loss": 0.4781,
+      "step": 15022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0399461400359066e-05,
+      "loss": 0.5174,
+      "step": 15023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039882021031034e-05,
+      "loss": 0.5133,
+      "step": 15024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039817902026161e-05,
+      "loss": 0.5853,
+      "step": 15025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039753783021288e-05,
+      "loss": 0.4469,
+      "step": 15026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0396896640164146e-05,
+      "loss": 0.3994,
+      "step": 15027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039625545011542e-05,
+      "loss": 0.6058,
+      "step": 15028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039561426006668e-05,
+      "loss": 0.5839,
+      "step": 15029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0394973070017954e-05,
+      "loss": 0.6385,
+      "step": 15030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0394331879969226e-05,
+      "loss": 0.5155,
+      "step": 15031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039369068992049e-05,
+      "loss": 0.5881,
+      "step": 15032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039304949987176e-05,
+      "loss": 0.423,
+      "step": 15033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0392408309823034e-05,
+      "loss": 0.5827,
+      "step": 15034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0391767119774306e-05,
+      "loss": 0.4713,
+      "step": 15035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039112592972557e-05,
+      "loss": 0.5202,
+      "step": 15036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039048473967684e-05,
+      "loss": 0.52,
+      "step": 15037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038984354962811e-05,
+      "loss": 0.5844,
+      "step": 15038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038920235957938e-05,
+      "loss": 0.5017,
+      "step": 15039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038856116953065e-05,
+      "loss": 0.545,
+      "step": 15040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038791997948192e-05,
+      "loss": 0.3681,
+      "step": 15041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0387278789433194e-05,
+      "loss": 0.5867,
+      "step": 15042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038663759938446e-05,
+      "loss": 0.5421,
+      "step": 15043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038599640933573e-05,
+      "loss": 0.5455,
+      "step": 15044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0385355219286996e-05,
+      "loss": 0.5487,
+      "step": 15045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038471402923827e-05,
+      "loss": 0.4978,
+      "step": 15046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038407283918953e-05,
+      "loss": 0.5453,
+      "step": 15047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0383431649140804e-05,
+      "loss": 0.6163,
+      "step": 15048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0382790459092076e-05,
+      "loss": 0.5631,
+      "step": 15049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038214926904335e-05,
+      "loss": 0.5171,
+      "step": 15050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038150807899462e-05,
+      "loss": 0.5955,
+      "step": 15051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0380866888945884e-05,
+      "loss": 0.6181,
+      "step": 15052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0380225698897156e-05,
+      "loss": 0.6281,
+      "step": 15053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037958450884842e-05,
+      "loss": 0.5669,
+      "step": 15054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037894331879969e-05,
+      "loss": 0.524,
+      "step": 15055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0378302128750964e-05,
+      "loss": 0.6353,
+      "step": 15056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0377660938702236e-05,
+      "loss": 0.4906,
+      "step": 15057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037701974865351e-05,
+      "loss": 0.4875,
+      "step": 15058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037637855860477e-05,
+      "loss": 0.558,
+      "step": 15059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0375737368556044e-05,
+      "loss": 0.473,
+      "step": 15060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037509617850731e-05,
+      "loss": 0.547,
+      "step": 15061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037445498845858e-05,
+      "loss": 0.4652,
+      "step": 15062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0373813798409846e-05,
+      "loss": 0.5505,
+      "step": 15063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037317260836112e-05,
+      "loss": 0.4824,
+      "step": 15064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037253141831239e-05,
+      "loss": 0.4884,
+      "step": 15065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037189022826366e-05,
+      "loss": 0.4724,
+      "step": 15066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037124903821493e-05,
+      "loss": 0.5969,
+      "step": 15067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03706078481662e-05,
+      "loss": 0.5169,
+      "step": 15068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036996665811747e-05,
+      "loss": 0.4678,
+      "step": 15069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0369325468068735e-05,
+      "loss": 0.4685,
+      "step": 15070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0368684278020006e-05,
+      "loss": 0.4477,
+      "step": 15071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036804308797127e-05,
+      "loss": 0.5061,
+      "step": 15072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036740189792255e-05,
+      "loss": 0.5545,
+      "step": 15073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0366760707873815e-05,
+      "loss": 0.5998,
+      "step": 15074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0366119517825086e-05,
+      "loss": 0.5851,
+      "step": 15075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036547832777636e-05,
+      "loss": 0.582,
+      "step": 15076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036483713772762e-05,
+      "loss": 0.4656,
+      "step": 15077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0364195947678895e-05,
+      "loss": 0.4433,
+      "step": 15078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036355475763016e-05,
+      "loss": 0.5813,
+      "step": 15079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036291356758143e-05,
+      "loss": 0.4537,
+      "step": 15080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03622723775327e-05,
+      "loss": 0.605,
+      "step": 15081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0361631187483975e-05,
+      "loss": 0.4299,
+      "step": 15082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036098999743524e-05,
+      "loss": 0.4743,
+      "step": 15083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036034880738651e-05,
+      "loss": 0.543,
+      "step": 15084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035970761733778e-05,
+      "loss": 0.5992,
+      "step": 15085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035906642728905e-05,
+      "loss": 0.6081,
+      "step": 15086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035842523724032e-05,
+      "loss": 0.4255,
+      "step": 15087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0357784047191585e-05,
+      "loss": 0.4331,
+      "step": 15088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035714285714286e-05,
+      "loss": 0.5346,
+      "step": 15089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035650166709413e-05,
+      "loss": 0.5123,
+      "step": 15090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03558604770454e-05,
+      "loss": 0.47,
+      "step": 15091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0355219286996665e-05,
+      "loss": 0.5116,
+      "step": 15092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0354578096947937e-05,
+      "loss": 0.5914,
+      "step": 15093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035393690689921e-05,
+      "loss": 0.6227,
+      "step": 15094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035329571685047e-05,
+      "loss": 0.5836,
+      "step": 15095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0352654526801745e-05,
+      "loss": 0.5425,
+      "step": 15096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035201333675302e-05,
+      "loss": 0.7024,
+      "step": 15097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035137214670429e-05,
+      "loss": 0.4472,
+      "step": 15098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035073095665555e-05,
+      "loss": 0.5328,
+      "step": 15099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0350089766606825e-05,
+      "loss": 0.5396,
+      "step": 15100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03494485765581e-05,
+      "loss": 0.6033,
+      "step": 15101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034880738650936e-05,
+      "loss": 0.5929,
+      "step": 15102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0348166196460633e-05,
+      "loss": 0.5337,
+      "step": 15103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03475250064119e-05,
+      "loss": 0.5408,
+      "step": 15104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034688381636318e-05,
+      "loss": 0.4386,
+      "step": 15105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034624262631444e-05,
+      "loss": 0.5685,
+      "step": 15106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0345601436265714e-05,
+      "loss": 0.5525,
+      "step": 15107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034496024621698e-05,
+      "loss": 0.6618,
+      "step": 15108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034431905616825e-05,
+      "loss": 0.56,
+      "step": 15109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034367786611952e-05,
+      "loss": 0.5028,
+      "step": 15110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034303667607079e-05,
+      "loss": 0.5099,
+      "step": 15111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034239548602206e-05,
+      "loss": 0.7088,
+      "step": 15112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034175429597333e-05,
+      "loss": 0.4516,
+      "step": 15113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03411131059246e-05,
+      "loss": 0.5383,
+      "step": 15114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034047191587587e-05,
+      "loss": 0.5312,
+      "step": 15115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033983072582714e-05,
+      "loss": 0.4267,
+      "step": 15116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0339189535778404e-05,
+      "loss": 0.4855,
+      "step": 15117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0338548345729675e-05,
+      "loss": 0.6019,
+      "step": 15118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033790715568095e-05,
+      "loss": 0.5859,
+      "step": 15119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033726596563221e-05,
+      "loss": 0.553,
+      "step": 15120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0336624775583484e-05,
+      "loss": 0.5371,
+      "step": 15121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0335983585534755e-05,
+      "loss": 0.5603,
+      "step": 15122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033534239548603e-05,
+      "loss": 0.6013,
+      "step": 15123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033470120543729e-05,
+      "loss": 0.4336,
+      "step": 15124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0334060015388564e-05,
+      "loss": 0.5348,
+      "step": 15125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033341882533983e-05,
+      "loss": 0.5983,
+      "step": 15126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03327776352911e-05,
+      "loss": 0.5592,
+      "step": 15127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033213644524237e-05,
+      "loss": 0.584,
+      "step": 15128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0331495255193644e-05,
+      "loss": 0.4974,
+      "step": 15129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0330854065144916e-05,
+      "loss": 0.5709,
+      "step": 15130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033021287509618e-05,
+      "loss": 0.5683,
+      "step": 15131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032957168504745e-05,
+      "loss": 0.4814,
+      "step": 15132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032893049499872e-05,
+      "loss": 0.543,
+      "step": 15133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032828930494999e-05,
+      "loss": 0.4835,
+      "step": 15134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0327648114901254e-05,
+      "loss": 0.6577,
+      "step": 15135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0327006924852526e-05,
+      "loss": 0.5646,
+      "step": 15136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03263657348038e-05,
+      "loss": 0.5286,
+      "step": 15137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032572454475507e-05,
+      "loss": 0.577,
+      "step": 15138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032508335470634e-05,
+      "loss": 0.5047,
+      "step": 15139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0324442164657606e-05,
+      "loss": 0.495,
+      "step": 15140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032380097460888e-05,
+      "loss": 0.5734,
+      "step": 15141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032315978456014e-05,
+      "loss": 0.5073,
+      "step": 15142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0322518594511414e-05,
+      "loss": 0.571,
+      "step": 15143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0321877404462686e-05,
+      "loss": 0.5892,
+      "step": 15144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032123621441396e-05,
+      "loss": 0.5293,
+      "step": 15145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032059502436523e-05,
+      "loss": 0.5489,
+      "step": 15146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0319953834316494e-05,
+      "loss": 0.5406,
+      "step": 15147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0319312644267766e-05,
+      "loss": 0.5684,
+      "step": 15148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031867145421903e-05,
+      "loss": 0.4756,
+      "step": 15149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03180302641703e-05,
+      "loss": 0.6251,
+      "step": 15150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031738907412157e-05,
+      "loss": 0.3709,
+      "step": 15151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031674788407284e-05,
+      "loss": 0.6123,
+      "step": 15152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031610669402411e-05,
+      "loss": 0.4754,
+      "step": 15153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031546550397538e-05,
+      "loss": 0.4371,
+      "step": 15154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0314824313926654e-05,
+      "loss": 0.4788,
+      "step": 15155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031418312387792e-05,
+      "loss": 0.6595,
+      "step": 15156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031354193382919e-05,
+      "loss": 0.5468,
+      "step": 15157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0312900743780456e-05,
+      "loss": 0.4821,
+      "step": 15158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031225955373173e-05,
+      "loss": 0.4963,
+      "step": 15159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031161836368299e-05,
+      "loss": 0.522,
+      "step": 15160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031097717363427e-05,
+      "loss": 0.5188,
+      "step": 15161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0310335983585536e-05,
+      "loss": 0.4625,
+      "step": 15162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030969479353681e-05,
+      "loss": 0.5486,
+      "step": 15163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030905360348808e-05,
+      "loss": 0.4308,
+      "step": 15164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0308412413439344e-05,
+      "loss": 0.5038,
+      "step": 15165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0307771223390616e-05,
+      "loss": 0.5922,
+      "step": 15166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030713003334188e-05,
+      "loss": 0.5286,
+      "step": 15167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030648884329315e-05,
+      "loss": 0.5477,
+      "step": 15168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0305847653244424e-05,
+      "loss": 0.5268,
+      "step": 15169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0305206463195696e-05,
+      "loss": 0.4986,
+      "step": 15170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030456527314696e-05,
+      "loss": 0.6092,
+      "step": 15171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030392408309823e-05,
+      "loss": 0.5333,
+      "step": 15172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0303282893049505e-05,
+      "loss": 0.4687,
+      "step": 15173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030264170300077e-05,
+      "loss": 0.3616,
+      "step": 15174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030200051295204e-05,
+      "loss": 0.6105,
+      "step": 15175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0301359322903306e-05,
+      "loss": 0.5471,
+      "step": 15176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0300718132854585e-05,
+      "loss": 0.6116,
+      "step": 15177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030007694280585e-05,
+      "loss": 0.6426,
+      "step": 15178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029943575275712e-05,
+      "loss": 0.5856,
+      "step": 15179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0298794562708386e-05,
+      "loss": 0.4698,
+      "step": 15180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029815337265966e-05,
+      "loss": 0.5468,
+      "step": 15181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029751218261093e-05,
+      "loss": 0.5762,
+      "step": 15182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0296870992562195e-05,
+      "loss": 0.6226,
+      "step": 15183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0296229802513466e-05,
+      "loss": 0.5107,
+      "step": 15184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029558861246474e-05,
+      "loss": 0.5828,
+      "step": 15185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029494742241601e-05,
+      "loss": 0.6346,
+      "step": 15186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0294306232367275e-05,
+      "loss": 0.5282,
+      "step": 15187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0293665042318546e-05,
+      "loss": 0.4615,
+      "step": 15188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029302385226981e-05,
+      "loss": 0.5035,
+      "step": 15189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029238266222108e-05,
+      "loss": 0.5613,
+      "step": 15190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0291741472172355e-05,
+      "loss": 0.5416,
+      "step": 15191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029110028212362e-05,
+      "loss": 0.5464,
+      "step": 15192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029045909207489e-05,
+      "loss": 0.5538,
+      "step": 15193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028981790202616e-05,
+      "loss": 0.6291,
+      "step": 15194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0289176711977435e-05,
+      "loss": 0.5745,
+      "step": 15195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.02885355219287e-05,
+      "loss": 0.5809,
+      "step": 15196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028789433187997e-05,
+      "loss": 0.5511,
+      "step": 15197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0287253141831236e-05,
+      "loss": 0.5834,
+      "step": 15198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028661195178251e-05,
+      "loss": 0.5994,
+      "step": 15199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028597076173378e-05,
+      "loss": 0.4588,
+      "step": 15200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028532957168505e-05,
+      "loss": 0.6336,
+      "step": 15201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028468838163632e-05,
+      "loss": 0.7222,
+      "step": 15202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028404719158759e-05,
+      "loss": 0.6282,
+      "step": 15203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028340600153886e-05,
+      "loss": 0.5407,
+      "step": 15204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0282764811490125e-05,
+      "loss": 0.5423,
+      "step": 15205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.02821236214414e-05,
+      "loss": 0.5523,
+      "step": 15206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028148243139266e-05,
+      "loss": 0.5523,
+      "step": 15207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028084124134393e-05,
+      "loss": 0.5412,
+      "step": 15208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0280200051295205e-05,
+      "loss": 0.5603,
+      "step": 15209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027955886124648e-05,
+      "loss": 0.514,
+      "step": 15210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027891767119775e-05,
+      "loss": 0.5943,
+      "step": 15211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027827648114901e-05,
+      "loss": 0.4901,
+      "step": 15212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0277635291100285e-05,
+      "loss": 0.5545,
+      "step": 15213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027699410105155e-05,
+      "loss": 0.6017,
+      "step": 15214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027635291100282e-05,
+      "loss": 0.4986,
+      "step": 15215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0275711720954093e-05,
+      "loss": 0.5431,
+      "step": 15216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0275070530905365e-05,
+      "loss": 0.5546,
+      "step": 15217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027442934085664e-05,
+      "loss": 0.5826,
+      "step": 15218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02737881508079e-05,
+      "loss": 0.5443,
+      "step": 15219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0273146960759174e-05,
+      "loss": 0.519,
+      "step": 15220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027250577071044e-05,
+      "loss": 0.5199,
+      "step": 15221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027186458066171e-05,
+      "loss": 0.5685,
+      "step": 15222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0271223390612975e-05,
+      "loss": 0.4276,
+      "step": 15223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027058220056425e-05,
+      "loss": 0.4981,
+      "step": 15224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026994101051552e-05,
+      "loss": 0.4538,
+      "step": 15225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026929982046679e-05,
+      "loss": 0.5054,
+      "step": 15226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026865863041806e-05,
+      "loss": 0.4958,
+      "step": 15227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026801744036933e-05,
+      "loss": 0.5298,
+      "step": 15228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02673762503206e-05,
+      "loss": 0.6394,
+      "step": 15229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0266735060271864e-05,
+      "loss": 0.5835,
+      "step": 15230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0266093870223135e-05,
+      "loss": 0.5149,
+      "step": 15231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02654526801744e-05,
+      "loss": 0.4706,
+      "step": 15232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026481149012568e-05,
+      "loss": 0.5511,
+      "step": 15233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0264170300076944e-05,
+      "loss": 0.4604,
+      "step": 15234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0263529110028215e-05,
+      "loss": 0.5207,
+      "step": 15235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026288791997949e-05,
+      "loss": 0.5248,
+      "step": 15236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026224672993075e-05,
+      "loss": 0.4722,
+      "step": 15237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0261605539882024e-05,
+      "loss": 0.4883,
+      "step": 15238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026096434983329e-05,
+      "loss": 0.5937,
+      "step": 15239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026032315978456e-05,
+      "loss": 0.583,
+      "step": 15240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025968196973583e-05,
+      "loss": 0.4938,
+      "step": 15241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0259040779687104e-05,
+      "loss": 0.5881,
+      "step": 15242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025839958963837e-05,
+      "loss": 0.4984,
+      "step": 15243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025775839958964e-05,
+      "loss": 0.5798,
+      "step": 15244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025711720954091e-05,
+      "loss": 0.5776,
+      "step": 15245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025647601949218e-05,
+      "loss": 0.5983,
+      "step": 15246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025583482944345e-05,
+      "loss": 0.4855,
+      "step": 15247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0255193639394714e-05,
+      "loss": 0.4644,
+      "step": 15248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025455244934599e-05,
+      "loss": 0.5539,
+      "step": 15249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025391125929726e-05,
+      "loss": 0.5749,
+      "step": 15250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025327006924853e-05,
+      "loss": 0.4572,
+      "step": 15251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0252628879199794e-05,
+      "loss": 0.4652,
+      "step": 15252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0251987689151066e-05,
+      "loss": 0.5618,
+      "step": 15253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025134649910234e-05,
+      "loss": 0.6241,
+      "step": 15254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02507053090536e-05,
+      "loss": 0.5296,
+      "step": 15255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0250064119004874e-05,
+      "loss": 0.4542,
+      "step": 15256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0249422928956146e-05,
+      "loss": 0.5051,
+      "step": 15257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024878173890742e-05,
+      "loss": 0.4379,
+      "step": 15258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024814054885868e-05,
+      "loss": 0.8004,
+      "step": 15259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0247499358809954e-05,
+      "loss": 0.4657,
+      "step": 15260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024685816876122e-05,
+      "loss": 0.4685,
+      "step": 15261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024621697871249e-05,
+      "loss": 0.4982,
+      "step": 15262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024557578866376e-05,
+      "loss": 0.5792,
+      "step": 15263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024493459861503e-05,
+      "loss": 0.4834,
+      "step": 15264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0244293408566306e-05,
+      "loss": 0.4591,
+      "step": 15265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024365221851757e-05,
+      "loss": 0.4891,
+      "step": 15266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024301102846884e-05,
+      "loss": 0.498,
+      "step": 15267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024236983842011e-05,
+      "loss": 0.569,
+      "step": 15268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024172864837138e-05,
+      "loss": 0.4201,
+      "step": 15269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0241087458322644e-05,
+      "loss": 0.5411,
+      "step": 15270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0240446268273916e-05,
+      "loss": 0.5329,
+      "step": 15271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023980507822519e-05,
+      "loss": 0.624,
+      "step": 15272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023916388817646e-05,
+      "loss": 0.4337,
+      "step": 15273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023852269812773e-05,
+      "loss": 0.605,
+      "step": 15274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0237881508078996e-05,
+      "loss": 0.5058,
+      "step": 15275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023724031803027e-05,
+      "loss": 0.4915,
+      "step": 15276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023659912798153e-05,
+      "loss": 0.5657,
+      "step": 15277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0235957937932804e-05,
+      "loss": 0.5291,
+      "step": 15278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023531674788407e-05,
+      "loss": 0.4649,
+      "step": 15279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023467555783534e-05,
+      "loss": 0.4588,
+      "step": 15280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023403436778661e-05,
+      "loss": 0.4456,
+      "step": 15281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0233393177737884e-05,
+      "loss": 0.5986,
+      "step": 15282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0232751987689156e-05,
+      "loss": 0.5872,
+      "step": 15283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023211079764042e-05,
+      "loss": 0.5364,
+      "step": 15284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023146960759169e-05,
+      "loss": 0.4485,
+      "step": 15285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023082841754296e-05,
+      "loss": 0.5374,
+      "step": 15286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023018722749423e-05,
+      "loss": 0.5081,
+      "step": 15287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02295460374455e-05,
+      "loss": 0.5131,
+      "step": 15288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022890484739677e-05,
+      "loss": 0.5677,
+      "step": 15289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0228263657348045e-05,
+      "loss": 0.5394,
+      "step": 15290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022762246729931e-05,
+      "loss": 0.5596,
+      "step": 15291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022698127725058e-05,
+      "loss": 0.5249,
+      "step": 15292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0226340087201846e-05,
+      "loss": 0.5386,
+      "step": 15293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022569889715312e-05,
+      "loss": 0.4917,
+      "step": 15294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022505770710438e-05,
+      "loss": 0.5834,
+      "step": 15295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0224416517055655e-05,
+      "loss": 0.4297,
+      "step": 15296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0223775327006926e-05,
+      "loss": 0.5418,
+      "step": 15297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02231341369582e-05,
+      "loss": 0.6367,
+      "step": 15298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022249294690947e-05,
+      "loss": 0.4757,
+      "step": 15299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0221851756860735e-05,
+      "loss": 0.5697,
+      "step": 15300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0221210566812006e-05,
+      "loss": 0.4769,
+      "step": 15301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022056937676327e-05,
+      "loss": 0.4765,
+      "step": 15302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021992818671454e-05,
+      "loss": 0.4722,
+      "step": 15303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0219286996665815e-05,
+      "loss": 0.5792,
+      "step": 15304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0218645806617087e-05,
+      "loss": 0.5482,
+      "step": 15305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021800461656836e-05,
+      "loss": 0.6111,
+      "step": 15306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021736342651962e-05,
+      "loss": 0.6031,
+      "step": 15307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0216722236470895e-05,
+      "loss": 0.4317,
+      "step": 15308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021608104642216e-05,
+      "loss": 0.5843,
+      "step": 15309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021543985637343e-05,
+      "loss": 0.5601,
+      "step": 15310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0214798666324696e-05,
+      "loss": 0.5991,
+      "step": 15311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021415747627597e-05,
+      "loss": 0.6117,
+      "step": 15312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021351628622724e-05,
+      "loss": 0.5187,
+      "step": 15313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021287509617851e-05,
+      "loss": 0.646,
+      "step": 15314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021223390612978e-05,
+      "loss": 0.5559,
+      "step": 15315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021159271608105e-05,
+      "loss": 0.4742,
+      "step": 15316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021095152603232e-05,
+      "loss": 0.6448,
+      "step": 15317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0210310335983585e-05,
+      "loss": 0.5342,
+      "step": 15318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020966914593486e-05,
+      "loss": 0.6407,
+      "step": 15319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020902795588612e-05,
+      "loss": 0.534,
+      "step": 15320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02083867658374e-05,
+      "loss": 0.5663,
+      "step": 15321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0207745575788665e-05,
+      "loss": 0.61,
+      "step": 15322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020710438573994e-05,
+      "loss": 0.5929,
+      "step": 15323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020646319569121e-05,
+      "loss": 0.4973,
+      "step": 15324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0205822005642473e-05,
+      "loss": 0.5616,
+      "step": 15325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0205180815593745e-05,
+      "loss": 0.4811,
+      "step": 15326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020453962554501e-05,
+      "loss": 0.6051,
+      "step": 15327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020389843549628e-05,
+      "loss": 0.474,
+      "step": 15328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0203257245447554e-05,
+      "loss": 0.5755,
+      "step": 15329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0202616055398825e-05,
+      "loss": 0.5239,
+      "step": 15330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020197486535009e-05,
+      "loss": 0.5653,
+      "step": 15331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020133367530136e-05,
+      "loss": 0.5577,
+      "step": 15332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0200692485252634e-05,
+      "loss": 0.535,
+      "step": 15333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02000512952039e-05,
+      "loss": 0.5345,
+      "step": 15334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019941010515517e-05,
+      "loss": 0.4653,
+      "step": 15335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0198768915106435e-05,
+      "loss": 0.6142,
+      "step": 15336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0198127725057714e-05,
+      "loss": 0.5013,
+      "step": 15337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019748653500898e-05,
+      "loss": 0.4772,
+      "step": 15338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019684534496025e-05,
+      "loss": 0.4017,
+      "step": 15339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0196204154911515e-05,
+      "loss": 0.4529,
+      "step": 15340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019556296486279e-05,
+      "loss": 0.5794,
+      "step": 15341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019492177481406e-05,
+      "loss": 0.5586,
+      "step": 15342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0194280584765324e-05,
+      "loss": 0.4093,
+      "step": 15343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0193639394716595e-05,
+      "loss": 0.5253,
+      "step": 15344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019299820466787e-05,
+      "loss": 0.5117,
+      "step": 15345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019235701461914e-05,
+      "loss": 0.5296,
+      "step": 15346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0191715824570404e-05,
+      "loss": 0.47,
+      "step": 15347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0191074634521675e-05,
+      "loss": 0.5642,
+      "step": 15348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019043344447294e-05,
+      "loss": 0.4681,
+      "step": 15349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018979225442421e-05,
+      "loss": 0.5233,
+      "step": 15350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0189151064375484e-05,
+      "loss": 0.5789,
+      "step": 15351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018850987432675e-05,
+      "loss": 0.6446,
+      "step": 15352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018786868427802e-05,
+      "loss": 0.5145,
+      "step": 15353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018722749422929e-05,
+      "loss": 0.5597,
+      "step": 15354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0186586304180564e-05,
+      "loss": 0.4449,
+      "step": 15355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018594511413183e-05,
+      "loss": 0.5951,
+      "step": 15356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01853039240831e-05,
+      "loss": 0.558,
+      "step": 15357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0184662734034366e-05,
+      "loss": 0.4333,
+      "step": 15358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018402154398564e-05,
+      "loss": 0.47,
+      "step": 15359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018338035393691e-05,
+      "loss": 0.5366,
+      "step": 15360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018273916388818e-05,
+      "loss": 0.505,
+      "step": 15361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018209797383945e-05,
+      "loss": 0.5751,
+      "step": 15362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018145678379072e-05,
+      "loss": 0.7288,
+      "step": 15363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018081559374199e-05,
+      "loss": 0.6227,
+      "step": 15364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0180174403693254e-05,
+      "loss": 0.5561,
+      "step": 15365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0179533213644526e-05,
+      "loss": 0.5295,
+      "step": 15366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017889202359579e-05,
+      "loss": 0.4767,
+      "step": 15367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017825083354706e-05,
+      "loss": 0.7303,
+      "step": 15368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0177609643498334e-05,
+      "loss": 0.5007,
+      "step": 15369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0176968453449606e-05,
+      "loss": 0.4586,
+      "step": 15370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017632726340088e-05,
+      "loss": 0.5955,
+      "step": 15371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017568607335214e-05,
+      "loss": 0.609,
+      "step": 15372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0175044883303414e-05,
+      "loss": 0.5084,
+      "step": 15373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017440369325468e-05,
+      "loss": 0.462,
+      "step": 15374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017376250320595e-05,
+      "loss": 0.4979,
+      "step": 15375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017312131315722e-05,
+      "loss": 0.4594,
+      "step": 15376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0172480123108494e-05,
+      "loss": 0.6078,
+      "step": 15377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0171838933059766e-05,
+      "loss": 0.4602,
+      "step": 15378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017119774301103e-05,
+      "loss": 0.5134,
+      "step": 15379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01705565529623e-05,
+      "loss": 0.516,
+      "step": 15380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016991536291357e-05,
+      "loss": 0.606,
+      "step": 15381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016927417286484e-05,
+      "loss": 0.5767,
+      "step": 15382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0168632982816104e-05,
+      "loss": 0.5977,
+      "step": 15383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0167991792767376e-05,
+      "loss": 0.5044,
+      "step": 15384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016735060271865e-05,
+      "loss": 0.5235,
+      "step": 15385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016670941266992e-05,
+      "loss": 0.4507,
+      "step": 15386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016606822262119e-05,
+      "loss": 0.5405,
+      "step": 15387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0165427032572456e-05,
+      "loss": 0.5951,
+      "step": 15388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016478584252373e-05,
+      "loss": 0.5484,
+      "step": 15389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016414465247499e-05,
+      "loss": 0.5925,
+      "step": 15390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0163503462426264e-05,
+      "loss": 0.5283,
+      "step": 15391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0162862272377536e-05,
+      "loss": 0.5849,
+      "step": 15392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016222108232881e-05,
+      "loss": 0.4959,
+      "step": 15393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016157989228007e-05,
+      "loss": 0.5279,
+      "step": 15394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0160938702231345e-05,
+      "loss": 0.5009,
+      "step": 15395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0160297512182616e-05,
+      "loss": 0.5562,
+      "step": 15396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015965632213388e-05,
+      "loss": 0.602,
+      "step": 15397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015901513208515e-05,
+      "loss": 0.4103,
+      "step": 15398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015837394203642e-05,
+      "loss": 0.6087,
+      "step": 15399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015773275198769e-05,
+      "loss": 0.5564,
+      "step": 15400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015709156193896e-05,
+      "loss": 0.5444,
+      "step": 15401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015645037189023e-05,
+      "loss": 0.4839,
+      "step": 15402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01558091818415e-05,
+      "loss": 0.5403,
+      "step": 15403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015516799179277e-05,
+      "loss": 0.4745,
+      "step": 15404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015452680174404e-05,
+      "loss": 0.5585,
+      "step": 15405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0153885611695306e-05,
+      "loss": 0.5043,
+      "step": 15406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015324442164658e-05,
+      "loss": 0.5581,
+      "step": 15407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015260323159784e-05,
+      "loss": 0.5685,
+      "step": 15408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015196204154912e-05,
+      "loss": 0.5731,
+      "step": 15409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0151320851500386e-05,
+      "loss": 0.4854,
+      "step": 15410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015067966145166e-05,
+      "loss": 0.5049,
+      "step": 15411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015003847140292e-05,
+      "loss": 0.502,
+      "step": 15412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0149397281354195e-05,
+      "loss": 0.564,
+      "step": 15413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0148756091305466e-05,
+      "loss": 0.4996,
+      "step": 15414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014811490125673e-05,
+      "loss": 0.6287,
+      "step": 15415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0147473711208e-05,
+      "loss": 0.4341,
+      "step": 15416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0146832521159275e-05,
+      "loss": 0.5757,
+      "step": 15417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0146191331110547e-05,
+      "loss": 0.5765,
+      "step": 15418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014555014106181e-05,
+      "loss": 0.6131,
+      "step": 15419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014490895101308e-05,
+      "loss": 0.447,
+      "step": 15420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014426776096435e-05,
+      "loss": 0.5296,
+      "step": 15421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014362657091562e-05,
+      "loss": 0.5121,
+      "step": 15422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014298538086689e-05,
+      "loss": 0.5005,
+      "step": 15423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0142344190818157e-05,
+      "loss": 0.5483,
+      "step": 15424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0141703000769435e-05,
+      "loss": 0.5841,
+      "step": 15425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01410618107207e-05,
+      "loss": 0.5907,
+      "step": 15426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014042062067197e-05,
+      "loss": 0.6176,
+      "step": 15427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0139779430623237e-05,
+      "loss": 0.5223,
+      "step": 15428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013913824057451e-05,
+      "loss": 0.5133,
+      "step": 15429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013849705052577e-05,
+      "loss": 0.4723,
+      "step": 15430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0137855860477045e-05,
+      "loss": 0.4904,
+      "step": 15431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013721467042832e-05,
+      "loss": 0.4719,
+      "step": 15432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013657348037959e-05,
+      "loss": 0.49,
+      "step": 15433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013593229033086e-05,
+      "loss": 0.5043,
+      "step": 15434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0135291100282125e-05,
+      "loss": 0.4372,
+      "step": 15435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01346499102334e-05,
+      "loss": 0.4734,
+      "step": 15436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013400872018466e-05,
+      "loss": 0.5386,
+      "step": 15437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0133367530135933e-05,
+      "loss": 0.5631,
+      "step": 15438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01327263400872e-05,
+      "loss": 0.5111,
+      "step": 15439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013208515003847e-05,
+      "loss": 0.5358,
+      "step": 15440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013144395998974e-05,
+      "loss": 0.4679,
+      "step": 15441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0130802769941014e-05,
+      "loss": 0.5646,
+      "step": 15442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0130161579892285e-05,
+      "loss": 0.5142,
+      "step": 15443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012952038984355e-05,
+      "loss": 0.5075,
+      "step": 15444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012887919979482e-05,
+      "loss": 0.5432,
+      "step": 15445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012823800974609e-05,
+      "loss": 0.4232,
+      "step": 15446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012759681969736e-05,
+      "loss": 0.5453,
+      "step": 15447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012695562964863e-05,
+      "loss": 0.5338,
+      "step": 15448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01263144395999e-05,
+      "loss": 0.4857,
+      "step": 15449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0125673249551174e-05,
+      "loss": 0.5503,
+      "step": 15450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012503205950244e-05,
+      "loss": 0.4761,
+      "step": 15451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012439086945371e-05,
+      "loss": 0.5228,
+      "step": 15452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0123749679404975e-05,
+      "loss": 0.5265,
+      "step": 15453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012310848935625e-05,
+      "loss": 0.5829,
+      "step": 15454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012246729930751e-05,
+      "loss": 0.6186,
+      "step": 15455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0121826109258784e-05,
+      "loss": 0.45,
+      "step": 15456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0121184919210055e-05,
+      "loss": 0.5552,
+      "step": 15457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012054372916133e-05,
+      "loss": 0.6128,
+      "step": 15458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01199025391126e-05,
+      "loss": 0.4965,
+      "step": 15459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0119261349063864e-05,
+      "loss": 0.564,
+      "step": 15460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0118620159015135e-05,
+      "loss": 0.5282,
+      "step": 15461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01179789689664e-05,
+      "loss": 0.5669,
+      "step": 15462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011733777891767e-05,
+      "loss": 0.4468,
+      "step": 15463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0116696588868944e-05,
+      "loss": 0.5868,
+      "step": 15464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0116055398820216e-05,
+      "loss": 0.5264,
+      "step": 15465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011541420877148e-05,
+      "loss": 0.5813,
+      "step": 15466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011477301872275e-05,
+      "loss": 0.5616,
+      "step": 15467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0114131828674024e-05,
+      "loss": 0.5303,
+      "step": 15468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011349063862529e-05,
+      "loss": 0.5172,
+      "step": 15469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011284944857656e-05,
+      "loss": 0.5255,
+      "step": 15470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0112208258527826e-05,
+      "loss": 0.5503,
+      "step": 15471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01115670684791e-05,
+      "loss": 0.4278,
+      "step": 15472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011092587843037e-05,
+      "loss": 0.5645,
+      "step": 15473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011028468838164e-05,
+      "loss": 0.5021,
+      "step": 15474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0109643498332906e-05,
+      "loss": 0.4571,
+      "step": 15475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010900230828418e-05,
+      "loss": 0.6133,
+      "step": 15476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010836111823545e-05,
+      "loss": 0.6002,
+      "step": 15477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0107719928186714e-05,
+      "loss": 0.4991,
+      "step": 15478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0107078738137986e-05,
+      "loss": 0.5732,
+      "step": 15479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010643754808925e-05,
+      "loss": 0.4849,
+      "step": 15480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010579635804053e-05,
+      "loss": 0.5926,
+      "step": 15481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0105155167991794e-05,
+      "loss": 0.4876,
+      "step": 15482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0104513977943066e-05,
+      "loss": 0.5227,
+      "step": 15483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010387278789434e-05,
+      "loss": 0.5608,
+      "step": 15484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.01032315978456e-05,
+      "loss": 0.5406,
+      "step": 15485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0102590407796874e-05,
+      "loss": 0.4378,
+      "step": 15486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010194921774814e-05,
+      "loss": 0.519,
+      "step": 15487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010130802769941e-05,
+      "loss": 0.5761,
+      "step": 15488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010066683765068e-05,
+      "loss": 0.4522,
+      "step": 15489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0100025647601954e-05,
+      "loss": 0.4643,
+      "step": 15490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009938445755322e-05,
+      "loss": 0.5142,
+      "step": 15491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009874326750449e-05,
+      "loss": 0.4306,
+      "step": 15492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009810207745576e-05,
+      "loss": 0.5583,
+      "step": 15493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009746088740703e-05,
+      "loss": 0.4925,
+      "step": 15494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00968196973583e-05,
+      "loss": 0.5556,
+      "step": 15495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0096178507309564e-05,
+      "loss": 0.5262,
+      "step": 15496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009553731726084e-05,
+      "loss": 0.5538,
+      "step": 15497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009489612721211e-05,
+      "loss": 0.5271,
+      "step": 15498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009425493716338e-05,
+      "loss": 0.5235,
+      "step": 15499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0093613747114644e-05,
+      "loss": 0.5528,
+      "step": 15500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0092972557065916e-05,
+      "loss": 0.5566,
+      "step": 15501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009233136701719e-05,
+      "loss": 0.6002,
+      "step": 15502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009169017696845e-05,
+      "loss": 0.6167,
+      "step": 15503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0091048986919724e-05,
+      "loss": 0.5579,
+      "step": 15504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0090407796870996e-05,
+      "loss": 0.6685,
+      "step": 15505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008976660682227e-05,
+      "loss": 0.5359,
+      "step": 15506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008912541677353e-05,
+      "loss": 0.6249,
+      "step": 15507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0088484226724805e-05,
+      "loss": 0.4813,
+      "step": 15508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008784303667607e-05,
+      "loss": 0.5733,
+      "step": 15509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008720184662734e-05,
+      "loss": 0.5093,
+      "step": 15510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008656065657861e-05,
+      "loss": 0.4211,
+      "step": 15511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008591946652988e-05,
+      "loss": 0.5154,
+      "step": 15512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0085278276481156e-05,
+      "loss": 0.4915,
+      "step": 15513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008463708643242e-05,
+      "loss": 0.5233,
+      "step": 15514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008399589638369e-05,
+      "loss": 0.4467,
+      "step": 15515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008335470633496e-05,
+      "loss": 0.5281,
+      "step": 15516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008271351628623e-05,
+      "loss": 0.5528,
+      "step": 15517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0082072326237495e-05,
+      "loss": 0.6026,
+      "step": 15518
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0081431136188766e-05,
+      "loss": 0.6031,
+      "step": 15519
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008078994614004e-05,
+      "loss": 0.5929,
+      "step": 15520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008014875609131e-05,
+      "loss": 0.5312,
+      "step": 15521
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007950756604258e-05,
+      "loss": 0.5033,
+      "step": 15522
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0078866375993846e-05,
+      "loss": 0.4603,
+      "step": 15523
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007822518594512e-05,
+      "loss": 0.539,
+      "step": 15524
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007758399589638e-05,
+      "loss": 0.5877,
+      "step": 15525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0076942805847655e-05,
+      "loss": 0.5179,
+      "step": 15526
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007630161579892e-05,
+      "loss": 0.5624,
+      "step": 15527
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007566042575019e-05,
+      "loss": 0.637,
+      "step": 15528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007501923570146e-05,
+      "loss": 0.6183,
+      "step": 15529
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0074378045652735e-05,
+      "loss": 0.6338,
+      "step": 15530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0073736855604007e-05,
+      "loss": 0.6401,
+      "step": 15531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007309566555527e-05,
+      "loss": 0.512,
+      "step": 15532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007245447550654e-05,
+      "loss": 0.5342,
+      "step": 15533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007181328545781e-05,
+      "loss": 0.5381,
+      "step": 15534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007117209540908e-05,
+      "loss": 0.505,
+      "step": 15535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007053090536035e-05,
+      "loss": 0.514,
+      "step": 15536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006988971531162e-05,
+      "loss": 0.4916,
+      "step": 15537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0069248525262895e-05,
+      "loss": 0.5458,
+      "step": 15538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006860733521416e-05,
+      "loss": 0.664,
+      "step": 15539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006796614516543e-05,
+      "loss": 0.517,
+      "step": 15540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00673249551167e-05,
+      "loss": 0.4252,
+      "step": 15541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006668376506797e-05,
+      "loss": 0.5605,
+      "step": 15542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006604257501923e-05,
+      "loss": 0.5126,
+      "step": 15543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0065401384970505e-05,
+      "loss": 0.5047,
+      "step": 15544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006476019492178e-05,
+      "loss": 0.5391,
+      "step": 15545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006411900487305e-05,
+      "loss": 0.5995,
+      "step": 15546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006347781482432e-05,
+      "loss": 0.4841,
+      "step": 15547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0062836624775585e-05,
+      "loss": 0.5485,
+      "step": 15548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006219543472686e-05,
+      "loss": 0.5614,
+      "step": 15549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006155424467812e-05,
+      "loss": 0.5414,
+      "step": 15550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0060913054629393e-05,
+      "loss": 0.4519,
+      "step": 15551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0060271864580665e-05,
+      "loss": 0.5274,
+      "step": 15552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005963067453194e-05,
+      "loss": 0.5196,
+      "step": 15553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00589894844832e-05,
+      "loss": 0.4344,
+      "step": 15554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0058348294434474e-05,
+      "loss": 0.6002,
+      "step": 15555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0057707104385745e-05,
+      "loss": 0.5168,
+      "step": 15556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005706591433701e-05,
+      "loss": 0.6346,
+      "step": 15557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005642472428828e-05,
+      "loss": 0.5513,
+      "step": 15558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005578353423955e-05,
+      "loss": 0.461,
+      "step": 15559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005514234419082e-05,
+      "loss": 0.547,
+      "step": 15560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005450115414209e-05,
+      "loss": 0.6419,
+      "step": 15561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005385996409336e-05,
+      "loss": 0.4219,
+      "step": 15562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005321877404463e-05,
+      "loss": 0.5671,
+      "step": 15563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00525775839959e-05,
+      "loss": 0.5608,
+      "step": 15564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005193639394717e-05,
+      "loss": 0.5883,
+      "step": 15565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0051295203898435e-05,
+      "loss": 0.4817,
+      "step": 15566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005065401384971e-05,
+      "loss": 0.6181,
+      "step": 15567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005001282380097e-05,
+      "loss": 0.5053,
+      "step": 15568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004937163375225e-05,
+      "loss": 0.3918,
+      "step": 15569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0048730443703515e-05,
+      "loss": 0.6974,
+      "step": 15570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004808925365479e-05,
+      "loss": 0.5616,
+      "step": 15571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004744806360605e-05,
+      "loss": 0.548,
+      "step": 15572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0046806873557324e-05,
+      "loss": 0.5225,
+      "step": 15573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0046165683508596e-05,
+      "loss": 0.5101,
+      "step": 15574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004552449345986e-05,
+      "loss": 0.479,
+      "step": 15575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004488330341113e-05,
+      "loss": 0.61,
+      "step": 15576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0044242113362404e-05,
+      "loss": 0.5797,
+      "step": 15577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0043600923313676e-05,
+      "loss": 0.5553,
+      "step": 15578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004295973326494e-05,
+      "loss": 0.5374,
+      "step": 15579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004231854321621e-05,
+      "loss": 0.4984,
+      "step": 15580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004167735316748e-05,
+      "loss": 0.5997,
+      "step": 15581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004103616311875e-05,
+      "loss": 0.6198,
+      "step": 15582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004039497307002e-05,
+      "loss": 0.5014,
+      "step": 15583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0039753783021286e-05,
+      "loss": 0.4854,
+      "step": 15584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0039112592972564e-05,
+      "loss": 0.5129,
+      "step": 15585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003847140292383e-05,
+      "loss": 0.5749,
+      "step": 15586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00378302128751e-05,
+      "loss": 0.5111,
+      "step": 15587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0037189022826366e-05,
+      "loss": 0.562,
+      "step": 15588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003654783277764e-05,
+      "loss": 0.5441,
+      "step": 15589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00359066427289e-05,
+      "loss": 0.6088,
+      "step": 15590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0035265452680174e-05,
+      "loss": 0.5826,
+      "step": 15591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0034624262631446e-05,
+      "loss": 0.485,
+      "step": 15592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003398307258272e-05,
+      "loss": 0.5255,
+      "step": 15593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003334188253399e-05,
+      "loss": 0.5222,
+      "step": 15594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0032700692485254e-05,
+      "loss": 0.6138,
+      "step": 15595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0032059502436526e-05,
+      "loss": 0.5544,
+      "step": 15596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003141831238779e-05,
+      "loss": 0.5675,
+      "step": 15597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003077712233906e-05,
+      "loss": 0.5252,
+      "step": 15598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003013593229033e-05,
+      "loss": 0.5178,
+      "step": 15599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00294947422416e-05,
+      "loss": 0.4728,
+      "step": 15600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002885355219287e-05,
+      "loss": 0.5602,
+      "step": 15601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002821236214414e-05,
+      "loss": 0.5545,
+      "step": 15602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0027571172095414e-05,
+      "loss": 0.4982,
+      "step": 15603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002692998204668e-05,
+      "loss": 0.6366,
+      "step": 15604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002628879199795e-05,
+      "loss": 0.4863,
+      "step": 15605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0025647601949216e-05,
+      "loss": 0.4425,
+      "step": 15606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002500641190049e-05,
+      "loss": 0.4366,
+      "step": 15607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002436522185176e-05,
+      "loss": 0.5722,
+      "step": 15608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002372403180303e-05,
+      "loss": 0.7448,
+      "step": 15609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00230828417543e-05,
+      "loss": 0.5743,
+      "step": 15610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002244165170557e-05,
+      "loss": 0.6367,
+      "step": 15611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002180046165684e-05,
+      "loss": 0.5411,
+      "step": 15612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0021159271608104e-05,
+      "loss": 0.5364,
+      "step": 15613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0020518081559376e-05,
+      "loss": 0.5877,
+      "step": 15614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001987689151064e-05,
+      "loss": 0.4904,
+      "step": 15615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001923570146191e-05,
+      "loss": 0.525,
+      "step": 15616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0018594511413184e-05,
+      "loss": 0.5561,
+      "step": 15617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0017953321364456e-05,
+      "loss": 0.6503,
+      "step": 15618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001731213131573e-05,
+      "loss": 0.5426,
+      "step": 15619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001667094126699e-05,
+      "loss": 0.5777,
+      "step": 15620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0016029751218265e-05,
+      "loss": 0.5527,
+      "step": 15621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001538856116953e-05,
+      "loss": 0.6713,
+      "step": 15622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00147473711208e-05,
+      "loss": 0.5215,
+      "step": 15623
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001410618107207e-05,
+      "loss": 0.4548,
+      "step": 15624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0013464991023345e-05,
+      "loss": 0.4662,
+      "step": 15625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001282380097461e-05,
+      "loss": 0.6361,
+      "step": 15626
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001218261092588e-05,
+      "loss": 0.5589,
+      "step": 15627
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001154142087715e-05,
+      "loss": 0.4314,
+      "step": 15628
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001090023082842e-05,
+      "loss": 0.5475,
+      "step": 15629
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001025904077969e-05,
+      "loss": 0.595,
+      "step": 15630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0009617850730955e-05,
+      "loss": 0.5569,
+      "step": 15631
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0008976660682226e-05,
+      "loss": 0.4966,
+      "step": 15632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00083354706335e-05,
+      "loss": 0.5188,
+      "step": 15633
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000769428058477e-05,
+      "loss": 0.5159,
+      "step": 15634
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0007053090536035e-05,
+      "loss": 0.5211,
+      "step": 15635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0006411900487306e-05,
+      "loss": 0.5248,
+      "step": 15636
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000577071043858e-05,
+      "loss": 0.6201,
+      "step": 15637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000512952038984e-05,
+      "loss": 0.4359,
+      "step": 15638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0004488330341115e-05,
+      "loss": 0.5707,
+      "step": 15639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000384714029238e-05,
+      "loss": 0.5188,
+      "step": 15640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000320595024366e-05,
+      "loss": 0.5049,
+      "step": 15641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000256476019492e-05,
+      "loss": 0.557,
+      "step": 15642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0001923570146195e-05,
+      "loss": 0.5356,
+      "step": 15643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000128238009746e-05,
+      "loss": 0.4994,
+      "step": 15644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000064119004873e-05,
+      "loss": 0.5105,
+      "step": 15645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4e-05,
+      "loss": 0.5103,
+      "step": 15646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999935880995127e-05,
+      "loss": 0.482,
+      "step": 15647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999871761990254e-05,
+      "loss": 0.4868,
+      "step": 15648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999807642985381e-05,
+      "loss": 0.5805,
+      "step": 15649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999743523980508e-05,
+      "loss": 0.5592,
+      "step": 15650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999679404975635e-05,
+      "loss": 0.5865,
+      "step": 15651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999615285970762e-05,
+      "loss": 0.552,
+      "step": 15652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9995511669658885e-05,
+      "loss": 0.4742,
+      "step": 15653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999487047961016e-05,
+      "loss": 0.4759,
+      "step": 15654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999422928956143e-05,
+      "loss": 0.5934,
+      "step": 15655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999358809951269e-05,
+      "loss": 0.5444,
+      "step": 15656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999294690946397e-05,
+      "loss": 0.5772,
+      "step": 15657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999230571941524e-05,
+      "loss": 0.5811,
+      "step": 15658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999166452936651e-05,
+      "loss": 0.5992,
+      "step": 15659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9991023339317773e-05,
+      "loss": 0.5646,
+      "step": 15660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9990382149269045e-05,
+      "loss": 0.5729,
+      "step": 15661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998974095922031e-05,
+      "loss": 0.6078,
+      "step": 15662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998909976917158e-05,
+      "loss": 0.5875,
+      "step": 15663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9988458579122854e-05,
+      "loss": 0.5394,
+      "step": 15664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9987817389074125e-05,
+      "loss": 0.5237,
+      "step": 15665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99871761990254e-05,
+      "loss": 0.4206,
+      "step": 15666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998653500897666e-05,
+      "loss": 0.6087,
+      "step": 15667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9985893818927934e-05,
+      "loss": 0.5526,
+      "step": 15668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99852526288792e-05,
+      "loss": 0.5051,
+      "step": 15669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998461143883047e-05,
+      "loss": 0.5628,
+      "step": 15670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998397024878174e-05,
+      "loss": 0.5479,
+      "step": 15671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998332905873301e-05,
+      "loss": 0.5571,
+      "step": 15672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9982687868684285e-05,
+      "loss": 0.5293,
+      "step": 15673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998204667863555e-05,
+      "loss": 0.6564,
+      "step": 15674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998140548858682e-05,
+      "loss": 0.4948,
+      "step": 15675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998076429853809e-05,
+      "loss": 0.603,
+      "step": 15676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.998012310848936e-05,
+      "loss": 0.5278,
+      "step": 15677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9979481918440624e-05,
+      "loss": 0.5593,
+      "step": 15678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9978840728391895e-05,
+      "loss": 0.4909,
+      "step": 15679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997819953834317e-05,
+      "loss": 0.4827,
+      "step": 15680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997755834829444e-05,
+      "loss": 0.552,
+      "step": 15681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997691715824571e-05,
+      "loss": 0.5583,
+      "step": 15682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9976275968196975e-05,
+      "loss": 0.4819,
+      "step": 15683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997563477814825e-05,
+      "loss": 0.505,
+      "step": 15684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997499358809951e-05,
+      "loss": 0.5778,
+      "step": 15685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9974352398050784e-05,
+      "loss": 0.4921,
+      "step": 15686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997371120800205e-05,
+      "loss": 0.4543,
+      "step": 15687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997307001795332e-05,
+      "loss": 0.4985,
+      "step": 15688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997242882790459e-05,
+      "loss": 0.5316,
+      "step": 15689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9971787637855864e-05,
+      "loss": 0.5837,
+      "step": 15690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9971146447807136e-05,
+      "loss": 0.5986,
+      "step": 15691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99705052577584e-05,
+      "loss": 0.6676,
+      "step": 15692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996986406770967e-05,
+      "loss": 0.5808,
+      "step": 15693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996922287766094e-05,
+      "loss": 0.6726,
+      "step": 15694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996858168761221e-05,
+      "loss": 0.4712,
+      "step": 15695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996794049756348e-05,
+      "loss": 0.5104,
+      "step": 15696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996729930751475e-05,
+      "loss": 0.4976,
+      "step": 15697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9966658117466024e-05,
+      "loss": 0.4346,
+      "step": 15698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996601692741729e-05,
+      "loss": 0.4833,
+      "step": 15699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996537573736856e-05,
+      "loss": 0.6434,
+      "step": 15700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9964734547319826e-05,
+      "loss": 0.5027,
+      "step": 15701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99640933572711e-05,
+      "loss": 0.5719,
+      "step": 15702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996345216722236e-05,
+      "loss": 0.5875,
+      "step": 15703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9962810977173634e-05,
+      "loss": 0.5487,
+      "step": 15704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9962169787124906e-05,
+      "loss": 0.5008,
+      "step": 15705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996152859707618e-05,
+      "loss": 0.5956,
+      "step": 15706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996088740702745e-05,
+      "loss": 0.5387,
+      "step": 15707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9960246216978714e-05,
+      "loss": 0.5453,
+      "step": 15708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9959605026929986e-05,
+      "loss": 0.5758,
+      "step": 15709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995896383688125e-05,
+      "loss": 0.5301,
+      "step": 15710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995832264683252e-05,
+      "loss": 0.5613,
+      "step": 15711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9957681456783794e-05,
+      "loss": 0.5017,
+      "step": 15712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9957040266735066e-05,
+      "loss": 0.5683,
+      "step": 15713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995639907668633e-05,
+      "loss": 0.5616,
+      "step": 15714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99557578866376e-05,
+      "loss": 0.57,
+      "step": 15715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9955116696588874e-05,
+      "loss": 0.4999,
+      "step": 15716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995447550654014e-05,
+      "loss": 0.4768,
+      "step": 15717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995383431649141e-05,
+      "loss": 0.6215,
+      "step": 15718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9953193126442676e-05,
+      "loss": 0.586,
+      "step": 15719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995255193639395e-05,
+      "loss": 0.5892,
+      "step": 15720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995191074634522e-05,
+      "loss": 0.5566,
+      "step": 15721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.995126955629649e-05,
+      "loss": 0.5405,
+      "step": 15722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9950628366247756e-05,
+      "loss": 0.5961,
+      "step": 15723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994998717619903e-05,
+      "loss": 0.5172,
+      "step": 15724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99493459861503e-05,
+      "loss": 0.5809,
+      "step": 15725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9948704796101564e-05,
+      "loss": 0.5189,
+      "step": 15726
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9948063606052836e-05,
+      "loss": 0.4386,
+      "step": 15727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99474224160041e-05,
+      "loss": 0.5005,
+      "step": 15728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994678122595538e-05,
+      "loss": 0.4679,
+      "step": 15729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9946140035906645e-05,
+      "loss": 0.4746,
+      "step": 15730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9945498845857916e-05,
+      "loss": 0.5865,
+      "step": 15731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994485765580918e-05,
+      "loss": 0.5319,
+      "step": 15732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994421646576045e-05,
+      "loss": 0.4829,
+      "step": 15733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9943575275711725e-05,
+      "loss": 0.581,
+      "step": 15734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994293408566299e-05,
+      "loss": 0.4764,
+      "step": 15735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994229289561426e-05,
+      "loss": 0.4838,
+      "step": 15736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994165170556553e-05,
+      "loss": 0.4944,
+      "step": 15737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9941010515516805e-05,
+      "loss": 0.4268,
+      "step": 15738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994036932546807e-05,
+      "loss": 0.4839,
+      "step": 15739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993972813541934e-05,
+      "loss": 0.5994,
+      "step": 15740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9939086945370606e-05,
+      "loss": 0.6366,
+      "step": 15741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993844575532188e-05,
+      "loss": 0.5584,
+      "step": 15742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993780456527315e-05,
+      "loss": 0.5378,
+      "step": 15743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9937163375224415e-05,
+      "loss": 0.5845,
+      "step": 15744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993652218517569e-05,
+      "loss": 0.5427,
+      "step": 15745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993588099512696e-05,
+      "loss": 0.494,
+      "step": 15746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993523980507823e-05,
+      "loss": 0.5372,
+      "step": 15747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9934598615029495e-05,
+      "loss": 0.562,
+      "step": 15748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9933957424980766e-05,
+      "loss": 0.537,
+      "step": 15749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993331623493203e-05,
+      "loss": 0.6075,
+      "step": 15750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99326750448833e-05,
+      "loss": 0.6007,
+      "step": 15751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9932033854834575e-05,
+      "loss": 0.492,
+      "step": 15752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9931392664785847e-05,
+      "loss": 0.5396,
+      "step": 15753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993075147473712e-05,
+      "loss": 0.5647,
+      "step": 15754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993011028468838e-05,
+      "loss": 0.5203,
+      "step": 15755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9929469094639655e-05,
+      "loss": 0.5191,
+      "step": 15756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992882790459092e-05,
+      "loss": 0.5883,
+      "step": 15757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992818671454219e-05,
+      "loss": 0.481,
+      "step": 15758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9927545524493457e-05,
+      "loss": 0.5379,
+      "step": 15759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992690433444473e-05,
+      "loss": 0.4999,
+      "step": 15760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9926263144396e-05,
+      "loss": 0.4607,
+      "step": 15761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992562195434727e-05,
+      "loss": 0.4995,
+      "step": 15762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9924980764298543e-05,
+      "loss": 0.5553,
+      "step": 15763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992433957424981e-05,
+      "loss": 0.4747,
+      "step": 15764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992369838420108e-05,
+      "loss": 0.5859,
+      "step": 15765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9923057194152345e-05,
+      "loss": 0.4973,
+      "step": 15766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992241600410362e-05,
+      "loss": 0.4853,
+      "step": 15767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992177481405489e-05,
+      "loss": 0.4861,
+      "step": 15768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992113362400616e-05,
+      "loss": 0.5779,
+      "step": 15769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992049243395743e-05,
+      "loss": 0.4434,
+      "step": 15770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99198512439087e-05,
+      "loss": 0.5726,
+      "step": 15771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991921005385997e-05,
+      "loss": 0.5809,
+      "step": 15772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9918568863811233e-05,
+      "loss": 0.6307,
+      "step": 15773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9917927673762505e-05,
+      "loss": 0.4892,
+      "step": 15774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991728648371377e-05,
+      "loss": 0.5573,
+      "step": 15775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991664529366504e-05,
+      "loss": 0.5609,
+      "step": 15776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9916004103616314e-05,
+      "loss": 0.538,
+      "step": 15777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9915362913567585e-05,
+      "loss": 0.4906,
+      "step": 15778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991472172351886e-05,
+      "loss": 0.5829,
+      "step": 15779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991408053347012e-05,
+      "loss": 0.5967,
+      "step": 15780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9913439343421394e-05,
+      "loss": 0.533,
+      "step": 15781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991279815337266e-05,
+      "loss": 0.5251,
+      "step": 15782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991215696332393e-05,
+      "loss": 0.5219,
+      "step": 15783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99115157732752e-05,
+      "loss": 0.582,
+      "step": 15784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9910874583226474e-05,
+      "loss": 0.4553,
+      "step": 15785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991023339317774e-05,
+      "loss": 0.4588,
+      "step": 15786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990959220312901e-05,
+      "loss": 0.4342,
+      "step": 15787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990895101308028e-05,
+      "loss": 0.4874,
+      "step": 15788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990830982303155e-05,
+      "loss": 0.4693,
+      "step": 15789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990766863298282e-05,
+      "loss": 0.4482,
+      "step": 15790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9907027442934084e-05,
+      "loss": 0.5264,
+      "step": 15791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9906386252885355e-05,
+      "loss": 0.5413,
+      "step": 15792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990574506283663e-05,
+      "loss": 0.4613,
+      "step": 15793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99051038727879e-05,
+      "loss": 0.4602,
+      "step": 15794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9904462682739164e-05,
+      "loss": 0.5523,
+      "step": 15795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9903821492690436e-05,
+      "loss": 0.6029,
+      "step": 15796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990318030264171e-05,
+      "loss": 0.5122,
+      "step": 15797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990253911259297e-05,
+      "loss": 0.5968,
+      "step": 15798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9901897922544244e-05,
+      "loss": 0.4285,
+      "step": 15799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9901256732495516e-05,
+      "loss": 0.4804,
+      "step": 15800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990061554244679e-05,
+      "loss": 0.5248,
+      "step": 15801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989997435239805e-05,
+      "loss": 0.4707,
+      "step": 15802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9899333162349324e-05,
+      "loss": 0.6384,
+      "step": 15803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989869197230059e-05,
+      "loss": 0.4831,
+      "step": 15804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989805078225186e-05,
+      "loss": 0.6374,
+      "step": 15805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989740959220313e-05,
+      "loss": 0.472,
+      "step": 15806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98967684021544e-05,
+      "loss": 0.5168,
+      "step": 15807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989612721210567e-05,
+      "loss": 0.4017,
+      "step": 15808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989548602205694e-05,
+      "loss": 0.6168,
+      "step": 15809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989484483200821e-05,
+      "loss": 0.6222,
+      "step": 15810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989420364195948e-05,
+      "loss": 0.4975,
+      "step": 15811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989356245191075e-05,
+      "loss": 0.5057,
+      "step": 15812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9892921261862014e-05,
+      "loss": 0.4397,
+      "step": 15813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9892280071813286e-05,
+      "loss": 0.607,
+      "step": 15814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989163888176456e-05,
+      "loss": 0.4272,
+      "step": 15815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989099769171582e-05,
+      "loss": 0.6249,
+      "step": 15816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98903565016671e-05,
+      "loss": 0.5126,
+      "step": 15817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9889715311618366e-05,
+      "loss": 0.6032,
+      "step": 15818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988907412156964e-05,
+      "loss": 0.5324,
+      "step": 15819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98884329315209e-05,
+      "loss": 0.6293,
+      "step": 15820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9887791741472174e-05,
+      "loss": 0.5117,
+      "step": 15821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988715055142344e-05,
+      "loss": 0.4846,
+      "step": 15822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988650936137471e-05,
+      "loss": 0.5044,
+      "step": 15823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988586817132598e-05,
+      "loss": 0.5156,
+      "step": 15824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9885226981277254e-05,
+      "loss": 0.5213,
+      "step": 15825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9884585791228526e-05,
+      "loss": 0.5515,
+      "step": 15826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988394460117979e-05,
+      "loss": 0.5234,
+      "step": 15827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988330341113106e-05,
+      "loss": 0.5406,
+      "step": 15828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988266222108233e-05,
+      "loss": 0.4051,
+      "step": 15829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98820210310336e-05,
+      "loss": 0.5769,
+      "step": 15830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9881379840984864e-05,
+      "loss": 0.6034,
+      "step": 15831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9880738650936136e-05,
+      "loss": 0.4435,
+      "step": 15832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9880097460887414e-05,
+      "loss": 0.4504,
+      "step": 15833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987945627083868e-05,
+      "loss": 0.4555,
+      "step": 15834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987881508078995e-05,
+      "loss": 0.4728,
+      "step": 15835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9878173890741216e-05,
+      "loss": 0.5402,
+      "step": 15836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987753270069249e-05,
+      "loss": 0.6027,
+      "step": 15837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987689151064375e-05,
+      "loss": 0.5156,
+      "step": 15838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9876250320595024e-05,
+      "loss": 0.4694,
+      "step": 15839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9875609130546296e-05,
+      "loss": 0.5262,
+      "step": 15840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987496794049757e-05,
+      "loss": 0.5187,
+      "step": 15841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987432675044884e-05,
+      "loss": 0.4706,
+      "step": 15842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9873685560400105e-05,
+      "loss": 0.5403,
+      "step": 15843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9873044370351376e-05,
+      "loss": 0.6269,
+      "step": 15844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987240318030264e-05,
+      "loss": 0.4685,
+      "step": 15845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987176199025391e-05,
+      "loss": 0.5107,
+      "step": 15846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987112080020518e-05,
+      "loss": 0.6017,
+      "step": 15847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987047961015645e-05,
+      "loss": 0.4327,
+      "step": 15848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986983842010772e-05,
+      "loss": 0.4824,
+      "step": 15849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986919723005899e-05,
+      "loss": 0.5529,
+      "step": 15850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9868556040010265e-05,
+      "loss": 0.53,
+      "step": 15851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986791484996153e-05,
+      "loss": 0.4524,
+      "step": 15852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98672736599128e-05,
+      "loss": 0.5877,
+      "step": 15853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9866632469864066e-05,
+      "loss": 0.3788,
+      "step": 15854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986599127981534e-05,
+      "loss": 0.6086,
+      "step": 15855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986535008976661e-05,
+      "loss": 0.5322,
+      "step": 15856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986470889971788e-05,
+      "loss": 0.5424,
+      "step": 15857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986406770966915e-05,
+      "loss": 0.5234,
+      "step": 15858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986342651962042e-05,
+      "loss": 0.5429,
+      "step": 15859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986278532957169e-05,
+      "loss": 0.6678,
+      "step": 15860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9862144139522955e-05,
+      "loss": 0.5123,
+      "step": 15861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9861502949474227e-05,
+      "loss": 0.469,
+      "step": 15862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986086175942549e-05,
+      "loss": 0.5385,
+      "step": 15863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986022056937676e-05,
+      "loss": 0.5356,
+      "step": 15864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9859579379328035e-05,
+      "loss": 0.5674,
+      "step": 15865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9858938189279307e-05,
+      "loss": 0.4608,
+      "step": 15866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985829699923058e-05,
+      "loss": 0.5875,
+      "step": 15867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985765580918184e-05,
+      "loss": 0.5312,
+      "step": 15868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9857014619133115e-05,
+      "loss": 0.3816,
+      "step": 15869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985637342908438e-05,
+      "loss": 0.693,
+      "step": 15870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985573223903565e-05,
+      "loss": 0.5436,
+      "step": 15871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985509104898692e-05,
+      "loss": 0.4506,
+      "step": 15872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9854449858938195e-05,
+      "loss": 0.6469,
+      "step": 15873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985380866888946e-05,
+      "loss": 0.5158,
+      "step": 15874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985316747884073e-05,
+      "loss": 0.462,
+      "step": 15875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9852526288792003e-05,
+      "loss": 0.5771,
+      "step": 15876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985188509874327e-05,
+      "loss": 0.5101,
+      "step": 15877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985124390869454e-05,
+      "loss": 0.6255,
+      "step": 15878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9850602718645805e-05,
+      "loss": 0.7225,
+      "step": 15879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984996152859708e-05,
+      "loss": 0.5573,
+      "step": 15880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984932033854835e-05,
+      "loss": 0.41,
+      "step": 15881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984867914849962e-05,
+      "loss": 0.4673,
+      "step": 15882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9848037958450885e-05,
+      "loss": 0.5427,
+      "step": 15883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984739676840216e-05,
+      "loss": 0.5201,
+      "step": 15884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984675557835343e-05,
+      "loss": 0.6204,
+      "step": 15885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9846114388304693e-05,
+      "loss": 0.5514,
+      "step": 15886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9845473198255965e-05,
+      "loss": 0.5801,
+      "step": 15887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984483200820723e-05,
+      "loss": 0.4679,
+      "step": 15888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984419081815851e-05,
+      "loss": 0.557,
+      "step": 15889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9843549628109774e-05,
+      "loss": 0.507,
+      "step": 15890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9842908438061045e-05,
+      "loss": 0.5249,
+      "step": 15891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984226724801231e-05,
+      "loss": 0.6775,
+      "step": 15892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984162605796358e-05,
+      "loss": 0.5791,
+      "step": 15893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9840984867914854e-05,
+      "loss": 0.557,
+      "step": 15894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984034367786612e-05,
+      "loss": 0.4402,
+      "step": 15895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983970248781739e-05,
+      "loss": 0.6432,
+      "step": 15896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983906129776866e-05,
+      "loss": 0.6165,
+      "step": 15897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9838420107719934e-05,
+      "loss": 0.5449,
+      "step": 15898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98377789176712e-05,
+      "loss": 0.5013,
+      "step": 15899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983713772762247e-05,
+      "loss": 0.681,
+      "step": 15900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9836496537573735e-05,
+      "loss": 0.4646,
+      "step": 15901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983585534752501e-05,
+      "loss": 0.5276,
+      "step": 15902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983521415747628e-05,
+      "loss": 0.5787,
+      "step": 15903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9834572967427544e-05,
+      "loss": 0.5204,
+      "step": 15904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983393177737882e-05,
+      "loss": 0.5165,
+      "step": 15905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983329058733009e-05,
+      "loss": 0.6845,
+      "step": 15906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983264939728136e-05,
+      "loss": 0.6337,
+      "step": 15907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9832008207232624e-05,
+      "loss": 0.5379,
+      "step": 15908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9831367017183896e-05,
+      "loss": 0.4503,
+      "step": 15909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983072582713516e-05,
+      "loss": 0.5491,
+      "step": 15910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.983008463708643e-05,
+      "loss": 0.5291,
+      "step": 15911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9829443447037704e-05,
+      "loss": 0.4798,
+      "step": 15912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9828802256988976e-05,
+      "loss": 0.4994,
+      "step": 15913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982816106694025e-05,
+      "loss": 0.6105,
+      "step": 15914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982751987689151e-05,
+      "loss": 0.5165,
+      "step": 15915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9826878686842784e-05,
+      "loss": 0.639,
+      "step": 15916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982623749679405e-05,
+      "loss": 0.5743,
+      "step": 15917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982559630674532e-05,
+      "loss": 0.4928,
+      "step": 15918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9824955116696586e-05,
+      "loss": 0.52,
+      "step": 15919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982431392664786e-05,
+      "loss": 0.5951,
+      "step": 15920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9823672736599136e-05,
+      "loss": 0.5252,
+      "step": 15921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98230315465504e-05,
+      "loss": 0.4806,
+      "step": 15922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982239035650167e-05,
+      "loss": 0.4854,
+      "step": 15923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982174916645294e-05,
+      "loss": 0.5439,
+      "step": 15924
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982110797640421e-05,
+      "loss": 0.5408,
+      "step": 15925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9820466786355474e-05,
+      "loss": 0.6327,
+      "step": 15926
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9819825596306746e-05,
+      "loss": 0.5511,
+      "step": 15927
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981918440625802e-05,
+      "loss": 0.4828,
+      "step": 15928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981854321620929e-05,
+      "loss": 0.4881,
+      "step": 15929
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981790202616056e-05,
+      "loss": 0.5646,
+      "step": 15930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9817260836111826e-05,
+      "loss": 0.5229,
+      "step": 15931
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98166196460631e-05,
+      "loss": 0.5705,
+      "step": 15932
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981597845601436e-05,
+      "loss": 0.5217,
+      "step": 15933
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9815337265965634e-05,
+      "loss": 0.5197,
+      "step": 15934
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98146960759169e-05,
+      "loss": 0.5097,
+      "step": 15935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981405488586817e-05,
+      "loss": 0.5491,
+      "step": 15936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981341369581944e-05,
+      "loss": 0.6324,
+      "step": 15937
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9812772505770714e-05,
+      "loss": 0.4559,
+      "step": 15938
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9812131315721986e-05,
+      "loss": 0.4833,
+      "step": 15939
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981149012567325e-05,
+      "loss": 0.4315,
+      "step": 15940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981084893562452e-05,
+      "loss": 0.4886,
+      "step": 15941
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981020774557579e-05,
+      "loss": 0.6114,
+      "step": 15942
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980956655552706e-05,
+      "loss": 0.4777,
+      "step": 15943
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980892536547833e-05,
+      "loss": 0.5006,
+      "step": 15944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98082841754296e-05,
+      "loss": 0.429,
+      "step": 15945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980764298538087e-05,
+      "loss": 0.5118,
+      "step": 15946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980700179533214e-05,
+      "loss": 0.4851,
+      "step": 15947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980636060528341e-05,
+      "loss": 0.4554,
+      "step": 15948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9805719415234676e-05,
+      "loss": 0.6129,
+      "step": 15949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980507822518595e-05,
+      "loss": 0.5387,
+      "step": 15950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980443703513721e-05,
+      "loss": 0.7085,
+      "step": 15951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9803795845088484e-05,
+      "loss": 0.6067,
+      "step": 15952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9803154655039756e-05,
+      "loss": 0.6034,
+      "step": 15953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980251346499103e-05,
+      "loss": 0.6172,
+      "step": 15954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980187227494229e-05,
+      "loss": 0.5906,
+      "step": 15955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9801231084893565e-05,
+      "loss": 0.6358,
+      "step": 15956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9800589894844836e-05,
+      "loss": 0.4822,
+      "step": 15957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.97999487047961e-05,
+      "loss": 0.4608,
+      "step": 15958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979930751474737e-05,
+      "loss": 0.506,
+      "step": 15959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9798666324698645e-05,
+      "loss": 0.6087,
+      "step": 15960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9798025134649916e-05,
+      "loss": 0.4506,
+      "step": 15961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979738394460118e-05,
+      "loss": 0.4902,
+      "step": 15962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979674275455245e-05,
+      "loss": 0.476,
+      "step": 15963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979610156450372e-05,
+      "loss": 0.5031,
+      "step": 15964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979546037445499e-05,
+      "loss": 0.5322,
+      "step": 15965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979481918440626e-05,
+      "loss": 0.7429,
+      "step": 15966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9794177994357526e-05,
+      "loss": 0.4505,
+      "step": 15967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.97935368043088e-05,
+      "loss": 0.4884,
+      "step": 15968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979289561426007e-05,
+      "loss": 0.513,
+      "step": 15969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979225442421134e-05,
+      "loss": 0.5652,
+      "step": 15970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9791613234162606e-05,
+      "loss": 0.5284,
+      "step": 15971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979097204411388e-05,
+      "loss": 0.5692,
+      "step": 15972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979033085406514e-05,
+      "loss": 0.6365,
+      "step": 15973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9789689664016415e-05,
+      "loss": 0.5379,
+      "step": 15974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9789048473967687e-05,
+      "loss": 0.5597,
+      "step": 15975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978840728391895e-05,
+      "loss": 0.6069,
+      "step": 15976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978776609387023e-05,
+      "loss": 0.408,
+      "step": 15977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9787124903821495e-05,
+      "loss": 0.5063,
+      "step": 15978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978648371377277e-05,
+      "loss": 0.5806,
+      "step": 15979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978584252372403e-05,
+      "loss": 0.6489,
+      "step": 15980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.97852013336753e-05,
+      "loss": 0.5275,
+      "step": 15981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978456014362657e-05,
+      "loss": 0.545,
+      "step": 15982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978391895357784e-05,
+      "loss": 0.4088,
+      "step": 15983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978327776352911e-05,
+      "loss": 0.4152,
+      "step": 15984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978263657348038e-05,
+      "loss": 0.5748,
+      "step": 15985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9781995383431655e-05,
+      "loss": 0.4515,
+      "step": 15986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978135419338292e-05,
+      "loss": 0.5173,
+      "step": 15987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978071300333419e-05,
+      "loss": 0.4956,
+      "step": 15988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.978007181328546e-05,
+      "loss": 0.5911,
+      "step": 15989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.977943062323673e-05,
+      "loss": 0.5413,
+      "step": 15990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.977878943318799e-05,
+      "loss": 0.5522,
+      "step": 15991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9778148243139265e-05,
+      "loss": 0.5471,
+      "step": 15992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9777507053090544e-05,
+      "loss": 0.4346,
+      "step": 15993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.977686586304181e-05,
+      "loss": 0.6327,
+      "step": 15994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.977622467299308e-05,
+      "loss": 0.5385,
+      "step": 15995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9775583482944345e-05,
+      "loss": 0.5025,
+      "step": 15996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.977494229289562e-05,
+      "loss": 0.4732,
+      "step": 15997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.977430110284688e-05,
+      "loss": 0.6402,
+      "step": 15998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9773659912798154e-05,
+      "loss": 0.487,
+      "step": 15999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9773018722749425e-05,
+      "loss": 0.4848,
+      "step": 16000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97723775327007e-05,
+      "loss": 0.53,
+      "step": 16001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.977173634265197e-05,
+      "loss": 0.4512,
+      "step": 16002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9771095152603234e-05,
+      "loss": 0.6061,
+      "step": 16003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9770453962554505e-05,
+      "loss": 0.5148,
+      "step": 16004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976981277250577e-05,
+      "loss": 0.5629,
+      "step": 16005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976917158245704e-05,
+      "loss": 0.4506,
+      "step": 16006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976853039240831e-05,
+      "loss": 0.565,
+      "step": 16007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976788920235958e-05,
+      "loss": 0.439,
+      "step": 16008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976724801231085e-05,
+      "loss": 0.5912,
+      "step": 16009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976660682226212e-05,
+      "loss": 0.5843,
+      "step": 16010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9765965632213394e-05,
+      "loss": 0.5841,
+      "step": 16011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976532444216466e-05,
+      "loss": 0.6543,
+      "step": 16012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976468325211593e-05,
+      "loss": 0.5063,
+      "step": 16013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9764042062067195e-05,
+      "loss": 0.5878,
+      "step": 16014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976340087201847e-05,
+      "loss": 0.5982,
+      "step": 16015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976275968196974e-05,
+      "loss": 0.6039,
+      "step": 16016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976211849192101e-05,
+      "loss": 0.52,
+      "step": 16017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9761477301872275e-05,
+      "loss": 0.4933,
+      "step": 16018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976083611182355e-05,
+      "loss": 0.5256,
+      "step": 16019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976019492177482e-05,
+      "loss": 0.6582,
+      "step": 16020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9759553731726084e-05,
+      "loss": 0.5953,
+      "step": 16021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9758912541677356e-05,
+      "loss": 0.509,
+      "step": 16022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975827135162862e-05,
+      "loss": 0.4883,
+      "step": 16023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975763016157989e-05,
+      "loss": 0.5099,
+      "step": 16024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9756988971531164e-05,
+      "loss": 0.5443,
+      "step": 16025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9756347781482436e-05,
+      "loss": 0.5745,
+      "step": 16026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97557065914337e-05,
+      "loss": 0.4864,
+      "step": 16027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975506540138497e-05,
+      "loss": 0.6391,
+      "step": 16028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9754424211336244e-05,
+      "loss": 0.5605,
+      "step": 16029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975378302128751e-05,
+      "loss": 0.5423,
+      "step": 16030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975314183123878e-05,
+      "loss": 0.5788,
+      "step": 16031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975250064119005e-05,
+      "loss": 0.503,
+      "step": 16032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9751859451141324e-05,
+      "loss": 0.495,
+      "step": 16033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975121826109259e-05,
+      "loss": 0.536,
+      "step": 16034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975057707104386e-05,
+      "loss": 0.5478,
+      "step": 16035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9749935880995126e-05,
+      "loss": 0.4544,
+      "step": 16036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97492946909464e-05,
+      "loss": 0.4658,
+      "step": 16037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974865350089767e-05,
+      "loss": 0.4753,
+      "step": 16038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9748012310848934e-05,
+      "loss": 0.5405,
+      "step": 16039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9747371120800206e-05,
+      "loss": 0.573,
+      "step": 16040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974672993075148e-05,
+      "loss": 0.6207,
+      "step": 16041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974608874070275e-05,
+      "loss": 0.4811,
+      "step": 16042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9745447550654014e-05,
+      "loss": 0.5161,
+      "step": 16043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9744806360605286e-05,
+      "loss": 0.5079,
+      "step": 16044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974416517055656e-05,
+      "loss": 0.5223,
+      "step": 16045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974352398050782e-05,
+      "loss": 0.5717,
+      "step": 16046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9742882790459094e-05,
+      "loss": 0.4761,
+      "step": 16047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974224160041036e-05,
+      "loss": 0.5682,
+      "step": 16048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974160041036164e-05,
+      "loss": 0.5505,
+      "step": 16049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97409592203129e-05,
+      "loss": 0.5695,
+      "step": 16050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9740318030264174e-05,
+      "loss": 0.4142,
+      "step": 16051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973967684021544e-05,
+      "loss": 0.498,
+      "step": 16052
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973903565016671e-05,
+      "loss": 0.4284,
+      "step": 16053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973839446011798e-05,
+      "loss": 0.5764,
+      "step": 16054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973775327006925e-05,
+      "loss": 0.5198,
+      "step": 16055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973711208002052e-05,
+      "loss": 0.5456,
+      "step": 16056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973647088997179e-05,
+      "loss": 0.6311,
+      "step": 16057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973582969992306e-05,
+      "loss": 0.518,
+      "step": 16058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973518850987433e-05,
+      "loss": 0.5029,
+      "step": 16059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97345473198256e-05,
+      "loss": 0.5327,
+      "step": 16060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9733906129776864e-05,
+      "loss": 0.4838,
+      "step": 16061
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9733264939728136e-05,
+      "loss": 0.51,
+      "step": 16062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973262374967941e-05,
+      "loss": 0.5193,
+      "step": 16063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973198255963067e-05,
+      "loss": 0.6994,
+      "step": 16064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973134136958195e-05,
+      "loss": 0.4702,
+      "step": 16065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9730700179533216e-05,
+      "loss": 0.5765,
+      "step": 16066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973005898948449e-05,
+      "loss": 0.5451,
+      "step": 16067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972941779943575e-05,
+      "loss": 0.5168,
+      "step": 16068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9728776609387025e-05,
+      "loss": 0.595,
+      "step": 16069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972813541933829e-05,
+      "loss": 0.615,
+      "step": 16070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972749422928956e-05,
+      "loss": 0.5149,
+      "step": 16071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972685303924083e-05,
+      "loss": 0.4116,
+      "step": 16072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9726211849192105e-05,
+      "loss": 0.5609,
+      "step": 16073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9725570659143376e-05,
+      "loss": 0.5381,
+      "step": 16074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972492946909464e-05,
+      "loss": 0.4869,
+      "step": 16075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972428827904591e-05,
+      "loss": 0.4975,
+      "step": 16076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972364708899718e-05,
+      "loss": 0.5121,
+      "step": 16077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972300589894845e-05,
+      "loss": 0.561,
+      "step": 16078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9722364708899715e-05,
+      "loss": 0.5238,
+      "step": 16079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9721723518850986e-05,
+      "loss": 0.444,
+      "step": 16080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9721082328802265e-05,
+      "loss": 0.6023,
+      "step": 16081
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972044113875353e-05,
+      "loss": 0.5993,
+      "step": 16082
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97197999487048e-05,
+      "loss": 0.5386,
+      "step": 16083
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9719158758656066e-05,
+      "loss": 0.6026,
+      "step": 16084
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971851756860734e-05,
+      "loss": 0.4823,
+      "step": 16085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97178763785586e-05,
+      "loss": 0.4956,
+      "step": 16086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9717235188509875e-05,
+      "loss": 0.5257,
+      "step": 16087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9716593998461147e-05,
+      "loss": 0.4581,
+      "step": 16088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971595280841242e-05,
+      "loss": 0.5832,
+      "step": 16089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971531161836369e-05,
+      "loss": 0.5396,
+      "step": 16090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9714670428314955e-05,
+      "loss": 0.5147,
+      "step": 16091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971402923826623e-05,
+      "loss": 0.4994,
+      "step": 16092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971338804821749e-05,
+      "loss": 0.6058,
+      "step": 16093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971274685816876e-05,
+      "loss": 0.5656,
+      "step": 16094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971210566812003e-05,
+      "loss": 0.5331,
+      "step": 16095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97114644780713e-05,
+      "loss": 0.5019,
+      "step": 16096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971082328802257e-05,
+      "loss": 0.4978,
+      "step": 16097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9710182097973843e-05,
+      "loss": 0.5823,
+      "step": 16098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9709540907925115e-05,
+      "loss": 0.5049,
+      "step": 16099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970889971787638e-05,
+      "loss": 0.5344,
+      "step": 16100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970825852782765e-05,
+      "loss": 0.5089,
+      "step": 16101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970761733777892e-05,
+      "loss": 0.4722,
+      "step": 16102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970697614773019e-05,
+      "loss": 0.5211,
+      "step": 16103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970633495768146e-05,
+      "loss": 0.474,
+      "step": 16104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970569376763273e-05,
+      "loss": 0.5499,
+      "step": 16105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9705052577584e-05,
+      "loss": 0.6128,
+      "step": 16106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970441138753527e-05,
+      "loss": 0.568,
+      "step": 16107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970377019748654e-05,
+      "loss": 0.5059,
+      "step": 16108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9703129007437805e-05,
+      "loss": 0.5211,
+      "step": 16109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970248781738908e-05,
+      "loss": 0.519,
+      "step": 16110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970184662734034e-05,
+      "loss": 0.4889,
+      "step": 16111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9701205437291614e-05,
+      "loss": 0.5347,
+      "step": 16112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9700564247242885e-05,
+      "loss": 0.5618,
+      "step": 16113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969992305719416e-05,
+      "loss": 0.4683,
+      "step": 16114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969928186714542e-05,
+      "loss": 0.4814,
+      "step": 16115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9698640677096694e-05,
+      "loss": 0.4914,
+      "step": 16116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9697999487047965e-05,
+      "loss": 0.5074,
+      "step": 16117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969735829699923e-05,
+      "loss": 0.5601,
+      "step": 16118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96967171069505e-05,
+      "loss": 0.49,
+      "step": 16119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9696075916901774e-05,
+      "loss": 0.5955,
+      "step": 16120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9695434726853045e-05,
+      "loss": 0.5166,
+      "step": 16121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969479353680431e-05,
+      "loss": 0.5815,
+      "step": 16122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969415234675558e-05,
+      "loss": 0.5079,
+      "step": 16123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969351115670685e-05,
+      "loss": 0.475,
+      "step": 16124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969286996665812e-05,
+      "loss": 0.5082,
+      "step": 16125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969222877660939e-05,
+      "loss": 0.4389,
+      "step": 16126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9691587586560655e-05,
+      "loss": 0.5433,
+      "step": 16127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969094639651193e-05,
+      "loss": 0.466,
+      "step": 16128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96903052064632e-05,
+      "loss": 0.49,
+      "step": 16129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968966401641447e-05,
+      "loss": 0.5512,
+      "step": 16130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9689022826365736e-05,
+      "loss": 0.607,
+      "step": 16131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968838163631701e-05,
+      "loss": 0.5585,
+      "step": 16132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968774044626827e-05,
+      "loss": 0.3926,
+      "step": 16133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9687099256219544e-05,
+      "loss": 0.5663,
+      "step": 16134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9686458066170816e-05,
+      "loss": 0.6051,
+      "step": 16135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968581687612208e-05,
+      "loss": 0.5743,
+      "step": 16136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968517568607336e-05,
+      "loss": 0.7205,
+      "step": 16137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9684534496024624e-05,
+      "loss": 0.5354,
+      "step": 16138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9683893305975896e-05,
+      "loss": 0.5125,
+      "step": 16139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968325211592716e-05,
+      "loss": 0.4761,
+      "step": 16140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968261092587843e-05,
+      "loss": 0.5724,
+      "step": 16141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96819697358297e-05,
+      "loss": 0.52,
+      "step": 16142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968132854578097e-05,
+      "loss": 0.5157,
+      "step": 16143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968068735573224e-05,
+      "loss": 0.4734,
+      "step": 16144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968004616568351e-05,
+      "loss": 0.5699,
+      "step": 16145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9679404975634784e-05,
+      "loss": 0.4867,
+      "step": 16146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967876378558605e-05,
+      "loss": 0.5716,
+      "step": 16147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967812259553732e-05,
+      "loss": 0.5634,
+      "step": 16148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9677481405488586e-05,
+      "loss": 0.4994,
+      "step": 16149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967684021543986e-05,
+      "loss": 0.5211,
+      "step": 16150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967619902539112e-05,
+      "loss": 0.5184,
+      "step": 16151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9675557835342394e-05,
+      "loss": 0.5071,
+      "step": 16152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967491664529367e-05,
+      "loss": 0.5697,
+      "step": 16153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967427545524494e-05,
+      "loss": 0.5096,
+      "step": 16154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967363426519621e-05,
+      "loss": 0.4581,
+      "step": 16155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9672993075147474e-05,
+      "loss": 0.5007,
+      "step": 16156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9672351885098746e-05,
+      "loss": 0.4967,
+      "step": 16157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967171069505001e-05,
+      "loss": 0.6541,
+      "step": 16158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967106950500128e-05,
+      "loss": 0.5132,
+      "step": 16159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9670428314952554e-05,
+      "loss": 0.5792,
+      "step": 16160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9669787124903826e-05,
+      "loss": 0.4709,
+      "step": 16161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96691459348551e-05,
+      "loss": 0.4767,
+      "step": 16162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966850474480636e-05,
+      "loss": 0.4803,
+      "step": 16163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9667863554757634e-05,
+      "loss": 0.5797,
+      "step": 16164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96672223647089e-05,
+      "loss": 0.4932,
+      "step": 16165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966658117466017e-05,
+      "loss": 0.4856,
+      "step": 16166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9665939984611436e-05,
+      "loss": 0.5306,
+      "step": 16167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966529879456271e-05,
+      "loss": 0.6584,
+      "step": 16168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966465760451398e-05,
+      "loss": 0.5892,
+      "step": 16169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966401641446525e-05,
+      "loss": 0.5485,
+      "step": 16170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966337522441652e-05,
+      "loss": 0.4884,
+      "step": 16171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966273403436779e-05,
+      "loss": 0.5378,
+      "step": 16172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966209284431906e-05,
+      "loss": 0.6156,
+      "step": 16173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9661451654270324e-05,
+      "loss": 0.4414,
+      "step": 16174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9660810464221596e-05,
+      "loss": 0.5067,
+      "step": 16175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.966016927417287e-05,
+      "loss": 0.5214,
+      "step": 16176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965952808412414e-05,
+      "loss": 0.4874,
+      "step": 16177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9658886894075405e-05,
+      "loss": 0.4873,
+      "step": 16178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9658245704026676e-05,
+      "loss": 0.5247,
+      "step": 16179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965760451397795e-05,
+      "loss": 0.4103,
+      "step": 16180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965696332392921e-05,
+      "loss": 0.5596,
+      "step": 16181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9656322133880485e-05,
+      "loss": 0.5893,
+      "step": 16182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965568094383175e-05,
+      "loss": 0.53,
+      "step": 16183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965503975378302e-05,
+      "loss": 0.5566,
+      "step": 16184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965439856373429e-05,
+      "loss": 0.5218,
+      "step": 16185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9653757373685565e-05,
+      "loss": 0.3681,
+      "step": 16186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965311618363683e-05,
+      "loss": 0.514,
+      "step": 16187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96524749935881e-05,
+      "loss": 0.4082,
+      "step": 16188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965183380353937e-05,
+      "loss": 0.5452,
+      "step": 16189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965119261349064e-05,
+      "loss": 0.5848,
+      "step": 16190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965055142344191e-05,
+      "loss": 0.6026,
+      "step": 16191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964991023339318e-05,
+      "loss": 0.53,
+      "step": 16192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964926904334445e-05,
+      "loss": 0.4833,
+      "step": 16193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964862785329572e-05,
+      "loss": 0.5639,
+      "step": 16194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964798666324699e-05,
+      "loss": 0.4481,
+      "step": 16195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9647345473198255e-05,
+      "loss": 0.4972,
+      "step": 16196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9646704283149527e-05,
+      "loss": 0.5466,
+      "step": 16197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96460630931008e-05,
+      "loss": 0.4928,
+      "step": 16198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964542190305206e-05,
+      "loss": 0.5461,
+      "step": 16199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9644780713003335e-05,
+      "loss": 0.4576,
+      "step": 16200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9644139522954607e-05,
+      "loss": 0.4867,
+      "step": 16201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964349833290588e-05,
+      "loss": 0.5448,
+      "step": 16202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964285714285714e-05,
+      "loss": 0.6266,
+      "step": 16203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9642215952808415e-05,
+      "loss": 0.5152,
+      "step": 16204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964157476275968e-05,
+      "loss": 0.4875,
+      "step": 16205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964093357271095e-05,
+      "loss": 0.5553,
+      "step": 16206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964029238266222e-05,
+      "loss": 0.5801,
+      "step": 16207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9639651192613495e-05,
+      "loss": 0.5586,
+      "step": 16208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963901000256477e-05,
+      "loss": 0.5001,
+      "step": 16209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963836881251603e-05,
+      "loss": 0.6202,
+      "step": 16210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9637727622467303e-05,
+      "loss": 0.5356,
+      "step": 16211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963708643241857e-05,
+      "loss": 0.4533,
+      "step": 16212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963644524236984e-05,
+      "loss": 0.5428,
+      "step": 16213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9635804052321105e-05,
+      "loss": 0.5026,
+      "step": 16214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963516286227238e-05,
+      "loss": 0.5123,
+      "step": 16215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963452167222365e-05,
+      "loss": 0.5902,
+      "step": 16216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963388048217492e-05,
+      "loss": 0.6169,
+      "step": 16217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963323929212619e-05,
+      "loss": 0.5587,
+      "step": 16218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963259810207746e-05,
+      "loss": 0.4914,
+      "step": 16219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963195691202873e-05,
+      "loss": 0.4369,
+      "step": 16220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9631315721979993e-05,
+      "loss": 0.5832,
+      "step": 16221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9630674531931265e-05,
+      "loss": 0.5509,
+      "step": 16222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963003334188253e-05,
+      "loss": 0.4563,
+      "step": 16223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96293921518338e-05,
+      "loss": 0.5729,
+      "step": 16224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962875096178508e-05,
+      "loss": 0.522,
+      "step": 16225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9628109771736345e-05,
+      "loss": 0.5372,
+      "step": 16226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962746858168762e-05,
+      "loss": 0.4693,
+      "step": 16227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962682739163888e-05,
+      "loss": 0.4892,
+      "step": 16228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9626186201590154e-05,
+      "loss": 0.4887,
+      "step": 16229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962554501154142e-05,
+      "loss": 0.5463,
+      "step": 16230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962490382149269e-05,
+      "loss": 0.688,
+      "step": 16231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962426263144396e-05,
+      "loss": 0.4756,
+      "step": 16232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9623621441395234e-05,
+      "loss": 0.4304,
+      "step": 16233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9622980251346505e-05,
+      "loss": 0.558,
+      "step": 16234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962233906129777e-05,
+      "loss": 0.594,
+      "step": 16235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962169787124904e-05,
+      "loss": 0.4809,
+      "step": 16236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962105668120031e-05,
+      "loss": 0.5221,
+      "step": 16237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962041549115158e-05,
+      "loss": 0.533,
+      "step": 16238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9619774301102844e-05,
+      "loss": 0.4283,
+      "step": 16239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9619133111054115e-05,
+      "loss": 0.4121,
+      "step": 16240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9618491921005394e-05,
+      "loss": 0.5424,
+      "step": 16241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961785073095666e-05,
+      "loss": 0.7312,
+      "step": 16242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961720954090793e-05,
+      "loss": 0.5755,
+      "step": 16243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9616568350859196e-05,
+      "loss": 0.5814,
+      "step": 16244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961592716081047e-05,
+      "loss": 0.5373,
+      "step": 16245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961528597076173e-05,
+      "loss": 0.6484,
+      "step": 16246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9614644780713004e-05,
+      "loss": 0.6385,
+      "step": 16247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9614003590664276e-05,
+      "loss": 0.5182,
+      "step": 16248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961336240061555e-05,
+      "loss": 0.5169,
+      "step": 16249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961272121056682e-05,
+      "loss": 0.6185,
+      "step": 16250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9612080020518084e-05,
+      "loss": 0.4233,
+      "step": 16251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9611438830469356e-05,
+      "loss": 0.5882,
+      "step": 16252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961079764042062e-05,
+      "loss": 0.5115,
+      "step": 16253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.961015645037189e-05,
+      "loss": 0.4957,
+      "step": 16254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.960951526032316e-05,
+      "loss": 0.5681,
+      "step": 16255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.960887407027443e-05,
+      "loss": 0.5373,
+      "step": 16256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.96082328802257e-05,
+      "loss": 0.566,
+      "step": 16257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960759169017697e-05,
+      "loss": 0.5744,
+      "step": 16258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9606950500128244e-05,
+      "loss": 0.5961,
+      "step": 16259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960630931007951e-05,
+      "loss": 0.6061,
+      "step": 16260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960566812003078e-05,
+      "loss": 0.5601,
+      "step": 16261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9605026929982046e-05,
+      "loss": 0.5149,
+      "step": 16262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960438573993332e-05,
+      "loss": 0.4868,
+      "step": 16263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960374454988459e-05,
+      "loss": 0.4861,
+      "step": 16264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960310335983586e-05,
+      "loss": 0.492,
+      "step": 16265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9602462169787126e-05,
+      "loss": 0.4696,
+      "step": 16266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.96018209797384e-05,
+      "loss": 0.5021,
+      "step": 16267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960117978968967e-05,
+      "loss": 0.5557,
+      "step": 16268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9600538599640934e-05,
+      "loss": 0.4858,
+      "step": 16269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9599897409592206e-05,
+      "loss": 0.5231,
+      "step": 16270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959925621954347e-05,
+      "loss": 0.4836,
+      "step": 16271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959861502949474e-05,
+      "loss": 0.6224,
+      "step": 16272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9597973839446014e-05,
+      "loss": 0.4699,
+      "step": 16273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9597332649397286e-05,
+      "loss": 0.5433,
+      "step": 16274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959669145934855e-05,
+      "loss": 0.538,
+      "step": 16275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959605026929982e-05,
+      "loss": 0.5946,
+      "step": 16276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9595409079251094e-05,
+      "loss": 0.6279,
+      "step": 16277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959476788920236e-05,
+      "loss": 0.6023,
+      "step": 16278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959412669915363e-05,
+      "loss": 0.4322,
+      "step": 16279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95934855091049e-05,
+      "loss": 0.5053,
+      "step": 16280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9592844319056175e-05,
+      "loss": 0.5239,
+      "step": 16281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959220312900744e-05,
+      "loss": 0.4383,
+      "step": 16282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959156193895871e-05,
+      "loss": 0.5039,
+      "step": 16283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9590920748909976e-05,
+      "loss": 0.4821,
+      "step": 16284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959027955886125e-05,
+      "loss": 0.4504,
+      "step": 16285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958963836881252e-05,
+      "loss": 0.4754,
+      "step": 16286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9588997178763784e-05,
+      "loss": 0.5738,
+      "step": 16287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9588355988715056e-05,
+      "loss": 0.5617,
+      "step": 16288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958771479866633e-05,
+      "loss": 0.4045,
+      "step": 16289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95870736086176e-05,
+      "loss": 0.4566,
+      "step": 16290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9586432418568865e-05,
+      "loss": 0.5595,
+      "step": 16291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9585791228520136e-05,
+      "loss": 0.5356,
+      "step": 16292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95851500384714e-05,
+      "loss": 0.5721,
+      "step": 16293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958450884842267e-05,
+      "loss": 0.6553,
+      "step": 16294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9583867658373945e-05,
+      "loss": 0.4808,
+      "step": 16295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958322646832521e-05,
+      "loss": 0.5826,
+      "step": 16296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958258527827649e-05,
+      "loss": 0.6578,
+      "step": 16297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958194408822775e-05,
+      "loss": 0.6017,
+      "step": 16298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9581302898179025e-05,
+      "loss": 0.5482,
+      "step": 16299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958066170813029e-05,
+      "loss": 0.5084,
+      "step": 16300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958002051808156e-05,
+      "loss": 0.5206,
+      "step": 16301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9579379328032826e-05,
+      "loss": 0.5667,
+      "step": 16302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95787381379841e-05,
+      "loss": 0.5148,
+      "step": 16303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957809694793537e-05,
+      "loss": 0.5919,
+      "step": 16304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957745575788664e-05,
+      "loss": 0.677,
+      "step": 16305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957681456783791e-05,
+      "loss": 0.5799,
+      "step": 16306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957617337778918e-05,
+      "loss": 0.4668,
+      "step": 16307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957553218774045e-05,
+      "loss": 0.5865,
+      "step": 16308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9574890997691715e-05,
+      "loss": 0.5769,
+      "step": 16309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9574249807642987e-05,
+      "loss": 0.4675,
+      "step": 16310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957360861759425e-05,
+      "loss": 0.5002,
+      "step": 16311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957296742754552e-05,
+      "loss": 0.4514,
+      "step": 16312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95723262374968e-05,
+      "loss": 0.571,
+      "step": 16313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957168504744807e-05,
+      "loss": 0.6237,
+      "step": 16314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957104385739934e-05,
+      "loss": 0.5089,
+      "step": 16315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95704026673506e-05,
+      "loss": 0.5906,
+      "step": 16316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9569761477301875e-05,
+      "loss": 0.4464,
+      "step": 16317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956912028725314e-05,
+      "loss": 0.5435,
+      "step": 16318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956847909720441e-05,
+      "loss": 0.5024,
+      "step": 16319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956783790715568e-05,
+      "loss": 0.5407,
+      "step": 16320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9567196717106955e-05,
+      "loss": 0.5638,
+      "step": 16321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956655552705823e-05,
+      "loss": 0.4664,
+      "step": 16322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956591433700949e-05,
+      "loss": 0.5642,
+      "step": 16323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9565273146960763e-05,
+      "loss": 0.4338,
+      "step": 16324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956463195691203e-05,
+      "loss": 0.5845,
+      "step": 16325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95639907668633e-05,
+      "loss": 0.5359,
+      "step": 16326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9563349576814565e-05,
+      "loss": 0.4901,
+      "step": 16327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956270838676584e-05,
+      "loss": 0.5052,
+      "step": 16328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956206719671711e-05,
+      "loss": 0.5991,
+      "step": 16329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956142600666838e-05,
+      "loss": 0.4431,
+      "step": 16330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956078481661965e-05,
+      "loss": 0.5466,
+      "step": 16331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956014362657092e-05,
+      "loss": 0.5668,
+      "step": 16332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955950243652219e-05,
+      "loss": 0.5709,
+      "step": 16333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9558861246473454e-05,
+      "loss": 0.564,
+      "step": 16334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9558220056424725e-05,
+      "loss": 0.5982,
+      "step": 16335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9557578866376e-05,
+      "loss": 0.5979,
+      "step": 16336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955693767632727e-05,
+      "loss": 0.5433,
+      "step": 16337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9556296486278534e-05,
+      "loss": 0.5154,
+      "step": 16338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9555655296229805e-05,
+      "loss": 0.5192,
+      "step": 16339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955501410618108e-05,
+      "loss": 0.386,
+      "step": 16340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955437291613234e-05,
+      "loss": 0.5486,
+      "step": 16341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9553731726083614e-05,
+      "loss": 0.4647,
+      "step": 16342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955309053603488e-05,
+      "loss": 0.513,
+      "step": 16343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955244934598615e-05,
+      "loss": 0.5756,
+      "step": 16344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955180815593742e-05,
+      "loss": 0.5343,
+      "step": 16345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9551166965888694e-05,
+      "loss": 0.4309,
+      "step": 16346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955052577583996e-05,
+      "loss": 0.5576,
+      "step": 16347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954988458579123e-05,
+      "loss": 0.5685,
+      "step": 16348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95492433957425e-05,
+      "loss": 0.5484,
+      "step": 16349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954860220569377e-05,
+      "loss": 0.5671,
+      "step": 16350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954796101564504e-05,
+      "loss": 0.564,
+      "step": 16351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954731982559631e-05,
+      "loss": 0.5289,
+      "step": 16352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954667863554758e-05,
+      "loss": 0.4485,
+      "step": 16353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954603744549885e-05,
+      "loss": 0.5039,
+      "step": 16354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954539625545012e-05,
+      "loss": 0.5748,
+      "step": 16355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9544755065401384e-05,
+      "loss": 0.5348,
+      "step": 16356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9544113875352656e-05,
+      "loss": 0.5244,
+      "step": 16357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954347268530393e-05,
+      "loss": 0.6146,
+      "step": 16358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954283149525519e-05,
+      "loss": 0.5067,
+      "step": 16359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9542190305206464e-05,
+      "loss": 0.4949,
+      "step": 16360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9541549115157736e-05,
+      "loss": 0.5738,
+      "step": 16361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954090792510901e-05,
+      "loss": 0.5018,
+      "step": 16362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954026673506027e-05,
+      "loss": 0.5653,
+      "step": 16363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9539625545011544e-05,
+      "loss": 0.4609,
+      "step": 16364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953898435496281e-05,
+      "loss": 0.6116,
+      "step": 16365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953834316491408e-05,
+      "loss": 0.449,
+      "step": 16366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953770197486535e-05,
+      "loss": 0.6163,
+      "step": 16367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9537060784816624e-05,
+      "loss": 0.6128,
+      "step": 16368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9536419594767896e-05,
+      "loss": 0.4954,
+      "step": 16369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953577840471916e-05,
+      "loss": 0.4933,
+      "step": 16370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953513721467043e-05,
+      "loss": 0.5457,
+      "step": 16371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95344960246217e-05,
+      "loss": 0.4836,
+      "step": 16372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953385483457297e-05,
+      "loss": 0.5931,
+      "step": 16373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9533213644524234e-05,
+      "loss": 0.4345,
+      "step": 16374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9532572454475506e-05,
+      "loss": 0.3874,
+      "step": 16375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953193126442678e-05,
+      "loss": 0.5218,
+      "step": 16376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953129007437805e-05,
+      "loss": 0.5289,
+      "step": 16377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953064888432932e-05,
+      "loss": 0.5419,
+      "step": 16378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9530007694280586e-05,
+      "loss": 0.4262,
+      "step": 16379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952936650423186e-05,
+      "loss": 0.5133,
+      "step": 16380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952872531418312e-05,
+      "loss": 0.5087,
+      "step": 16381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9528084124134394e-05,
+      "loss": 0.5044,
+      "step": 16382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952744293408566e-05,
+      "loss": 0.4642,
+      "step": 16383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952680174403693e-05,
+      "loss": 0.4577,
+      "step": 16384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952616055398821e-05,
+      "loss": 0.6164,
+      "step": 16385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9525519363939474e-05,
+      "loss": 0.5518,
+      "step": 16386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9524878173890746e-05,
+      "loss": 0.5769,
+      "step": 16387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952423698384201e-05,
+      "loss": 0.5526,
+      "step": 16388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952359579379328e-05,
+      "loss": 0.5022,
+      "step": 16389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952295460374455e-05,
+      "loss": 0.5557,
+      "step": 16390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952231341369582e-05,
+      "loss": 0.5828,
+      "step": 16391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952167222364709e-05,
+      "loss": 0.569,
+      "step": 16392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952103103359836e-05,
+      "loss": 0.6437,
+      "step": 16393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9520389843549635e-05,
+      "loss": 0.534,
+      "step": 16394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95197486535009e-05,
+      "loss": 0.6118,
+      "step": 16395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951910746345217e-05,
+      "loss": 0.4944,
+      "step": 16396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9518466273403436e-05,
+      "loss": 0.3909,
+      "step": 16397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951782508335471e-05,
+      "loss": 0.6561,
+      "step": 16398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951718389330597e-05,
+      "loss": 0.5157,
+      "step": 16399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9516542703257245e-05,
+      "loss": 0.6377,
+      "step": 16400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9515901513208516e-05,
+      "loss": 0.4989,
+      "step": 16401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951526032315979e-05,
+      "loss": 0.4494,
+      "step": 16402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951461913311106e-05,
+      "loss": 0.5813,
+      "step": 16403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9513977943062325e-05,
+      "loss": 0.5151,
+      "step": 16404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9513336753013596e-05,
+      "loss": 0.5994,
+      "step": 16405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951269556296486e-05,
+      "loss": 0.4958,
+      "step": 16406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951205437291613e-05,
+      "loss": 0.5425,
+      "step": 16407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9511413182867405e-05,
+      "loss": 0.4877,
+      "step": 16408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9510771992818676e-05,
+      "loss": 0.6328,
+      "step": 16409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951013080276994e-05,
+      "loss": 0.5677,
+      "step": 16410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950948961272121e-05,
+      "loss": 0.432,
+      "step": 16411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9508848422672485e-05,
+      "loss": 0.4893,
+      "step": 16412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950820723262375e-05,
+      "loss": 0.6201,
+      "step": 16413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950756604257502e-05,
+      "loss": 0.5101,
+      "step": 16414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9506924852526286e-05,
+      "loss": 0.5067,
+      "step": 16415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950628366247756e-05,
+      "loss": 0.4717,
+      "step": 16416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950564247242883e-05,
+      "loss": 0.5195,
+      "step": 16417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95050012823801e-05,
+      "loss": 0.5065,
+      "step": 16418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9504360092331366e-05,
+      "loss": 0.458,
+      "step": 16419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950371890228264e-05,
+      "loss": 0.4763,
+      "step": 16420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950307771223391e-05,
+      "loss": 0.5909,
+      "step": 16421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9502436522185175e-05,
+      "loss": 0.5224,
+      "step": 16422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9501795332136447e-05,
+      "loss": 0.4391,
+      "step": 16423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950115414208772e-05,
+      "loss": 0.4962,
+      "step": 16424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.950051295203899e-05,
+      "loss": 0.4501,
+      "step": 16425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9499871761990255e-05,
+      "loss": 0.5861,
+      "step": 16426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949923057194153e-05,
+      "loss": 0.6139,
+      "step": 16427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.94985893818928e-05,
+      "loss": 0.7407,
+      "step": 16428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949794819184406e-05,
+      "loss": 0.5524,
+      "step": 16429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9497307001795335e-05,
+      "loss": 0.6109,
+      "step": 16430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.94966658117466e-05,
+      "loss": 0.5625,
+      "step": 16431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949602462169787e-05,
+      "loss": 0.5101,
+      "step": 16432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9495383431649143e-05,
+      "loss": 0.451,
+      "step": 16433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9494742241600415e-05,
+      "loss": 0.5168,
+      "step": 16434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949410105155168e-05,
+      "loss": 0.4635,
+      "step": 16435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949345986150295e-05,
+      "loss": 0.5224,
+      "step": 16436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9492818671454224e-05,
+      "loss": 0.4593,
+      "step": 16437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949217748140549e-05,
+      "loss": 0.6036,
+      "step": 16438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949153629135676e-05,
+      "loss": 0.6251,
+      "step": 16439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949089510130803e-05,
+      "loss": 0.5497,
+      "step": 16440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9490253911259304e-05,
+      "loss": 0.7035,
+      "step": 16441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.948961272121057e-05,
+      "loss": 0.5321,
+      "step": 16442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.948897153116184e-05,
+      "loss": 0.5447,
+      "step": 16443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9488330341113105e-05,
+      "loss": 0.5165,
+      "step": 16444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.948768915106438e-05,
+      "loss": 0.4981,
+      "step": 16445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.948704796101565e-05,
+      "loss": 0.5737,
+      "step": 16446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9486406770966914e-05,
+      "loss": 0.4785,
+      "step": 16447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9485765580918185e-05,
+      "loss": 0.6231,
+      "step": 16448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.948512439086946e-05,
+      "loss": 0.5577,
+      "step": 16449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.948448320082073e-05,
+      "loss": 0.53,
+      "step": 16450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9483842010771994e-05,
+      "loss": 0.4176,
+      "step": 16451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9483200820723265e-05,
+      "loss": 0.5201,
+      "step": 16452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.948255963067453e-05,
+      "loss": 0.5429,
+      "step": 16453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.94819184406258e-05,
+      "loss": 0.5851,
+      "step": 16454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9481277250577074e-05,
+      "loss": 0.525,
+      "step": 16455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9480636060528345e-05,
+      "loss": 0.5475,
+      "step": 16456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947999487047962e-05,
+      "loss": 0.433,
+      "step": 16457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947935368043088e-05,
+      "loss": 0.4966,
+      "step": 16458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9478712490382154e-05,
+      "loss": 0.4966,
+      "step": 16459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947807130033342e-05,
+      "loss": 0.6238,
+      "step": 16460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947743011028469e-05,
+      "loss": 0.3166,
+      "step": 16461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9476788920235955e-05,
+      "loss": 0.4247,
+      "step": 16462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947614773018723e-05,
+      "loss": 0.5269,
+      "step": 16463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.94755065401385e-05,
+      "loss": 0.4485,
+      "step": 16464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947486535008977e-05,
+      "loss": 0.4631,
+      "step": 16465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947422416004104e-05,
+      "loss": 0.5694,
+      "step": 16466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947358296999231e-05,
+      "loss": 0.6444,
+      "step": 16467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947294177994358e-05,
+      "loss": 0.5444,
+      "step": 16468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9472300589894844e-05,
+      "loss": 0.5908,
+      "step": 16469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9471659399846116e-05,
+      "loss": 0.4155,
+      "step": 16470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947101820979738e-05,
+      "loss": 0.5691,
+      "step": 16471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947037701974865e-05,
+      "loss": 0.4197,
+      "step": 16472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946973582969993e-05,
+      "loss": 0.5229,
+      "step": 16473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9469094639651196e-05,
+      "loss": 0.4782,
+      "step": 16474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946845344960247e-05,
+      "loss": 0.6443,
+      "step": 16475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946781225955373e-05,
+      "loss": 0.492,
+      "step": 16476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9467171069505004e-05,
+      "loss": 0.4828,
+      "step": 16477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946652987945627e-05,
+      "loss": 0.6264,
+      "step": 16478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946588868940754e-05,
+      "loss": 0.5339,
+      "step": 16479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946524749935881e-05,
+      "loss": 0.3521,
+      "step": 16480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9464606309310084e-05,
+      "loss": 0.4731,
+      "step": 16481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9463965119261356e-05,
+      "loss": 0.6522,
+      "step": 16482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946332392921262e-05,
+      "loss": 0.4844,
+      "step": 16483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946268273916389e-05,
+      "loss": 0.555,
+      "step": 16484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946204154911516e-05,
+      "loss": 0.5063,
+      "step": 16485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.946140035906643e-05,
+      "loss": 0.6291,
+      "step": 16486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9460759169017694e-05,
+      "loss": 0.4871,
+      "step": 16487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9460117978968966e-05,
+      "loss": 0.5745,
+      "step": 16488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945947678892024e-05,
+      "loss": 0.5199,
+      "step": 16489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945883559887151e-05,
+      "loss": 0.5862,
+      "step": 16490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945819440882278e-05,
+      "loss": 0.5183,
+      "step": 16491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9457553218774046e-05,
+      "loss": 0.5484,
+      "step": 16492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945691202872532e-05,
+      "loss": 0.5882,
+      "step": 16493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945627083867658e-05,
+      "loss": 0.515,
+      "step": 16494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9455629648627854e-05,
+      "loss": 0.4819,
+      "step": 16495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9454988458579126e-05,
+      "loss": 0.5943,
+      "step": 16496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.94543472685304e-05,
+      "loss": 0.5596,
+      "step": 16497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945370607848166e-05,
+      "loss": 0.5989,
+      "step": 16498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9453064888432934e-05,
+      "loss": 0.6612,
+      "step": 16499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9452423698384206e-05,
+      "loss": 0.4442,
+      "step": 16500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945178250833547e-05,
+      "loss": 0.4706,
+      "step": 16501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945114131828674e-05,
+      "loss": 0.5202,
+      "step": 16502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945050012823801e-05,
+      "loss": 0.5141,
+      "step": 16503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944985893818928e-05,
+      "loss": 0.5239,
+      "step": 16504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944921774814055e-05,
+      "loss": 0.6163,
+      "step": 16505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944857655809182e-05,
+      "loss": 0.5332,
+      "step": 16506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944793536804309e-05,
+      "loss": 0.6222,
+      "step": 16507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944729417799436e-05,
+      "loss": 0.5424,
+      "step": 16508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944665298794563e-05,
+      "loss": 0.5291,
+      "step": 16509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9446011797896896e-05,
+      "loss": 0.5171,
+      "step": 16510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944537060784817e-05,
+      "loss": 0.6157,
+      "step": 16511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944472941779944e-05,
+      "loss": 0.5981,
+      "step": 16512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944408822775071e-05,
+      "loss": 0.49,
+      "step": 16513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9443447037701976e-05,
+      "loss": 0.4901,
+      "step": 16514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944280584765325e-05,
+      "loss": 0.6013,
+      "step": 16515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.944216465760451e-05,
+      "loss": 0.5557,
+      "step": 16516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9441523467555785e-05,
+      "loss": 0.5645,
+      "step": 16517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9440882277507056e-05,
+      "loss": 0.5855,
+      "step": 16518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.944024108745832e-05,
+      "loss": 0.5773,
+      "step": 16519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943959989740959e-05,
+      "loss": 0.5126,
+      "step": 16520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9438958707360865e-05,
+      "loss": 0.4402,
+      "step": 16521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9438317517312136e-05,
+      "loss": 0.5316,
+      "step": 16522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94376763272634e-05,
+      "loss": 0.471,
+      "step": 16523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943703513721467e-05,
+      "loss": 0.5523,
+      "step": 16524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943639394716594e-05,
+      "loss": 0.4237,
+      "step": 16525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943575275711721e-05,
+      "loss": 0.5676,
+      "step": 16526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943511156706848e-05,
+      "loss": 0.6445,
+      "step": 16527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943447037701975e-05,
+      "loss": 0.4736,
+      "step": 16528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9433829186971025e-05,
+      "loss": 0.4448,
+      "step": 16529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943318799692229e-05,
+      "loss": 0.5654,
+      "step": 16530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943254680687356e-05,
+      "loss": 0.6123,
+      "step": 16531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9431905616824827e-05,
+      "loss": 0.6288,
+      "step": 16532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94312644267761e-05,
+      "loss": 0.5234,
+      "step": 16533
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943062323672736e-05,
+      "loss": 0.5734,
+      "step": 16534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9429982046678635e-05,
+      "loss": 0.5796,
+      "step": 16535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9429340856629907e-05,
+      "loss": 0.5582,
+      "step": 16536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942869966658118e-05,
+      "loss": 0.5447,
+      "step": 16537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942805847653245e-05,
+      "loss": 0.5367,
+      "step": 16538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9427417286483715e-05,
+      "loss": 0.5144,
+      "step": 16539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942677609643499e-05,
+      "loss": 0.6543,
+      "step": 16540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942613490638625e-05,
+      "loss": 0.437,
+      "step": 16541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942549371633752e-05,
+      "loss": 0.6325,
+      "step": 16542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942485252628879e-05,
+      "loss": 0.5004,
+      "step": 16543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942421133624006e-05,
+      "loss": 0.5227,
+      "step": 16544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942357014619134e-05,
+      "loss": 0.4431,
+      "step": 16545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9422928956142603e-05,
+      "loss": 0.4095,
+      "step": 16546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9422287766093875e-05,
+      "loss": 0.4851,
+      "step": 16547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942164657604514e-05,
+      "loss": 0.6195,
+      "step": 16548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942100538599641e-05,
+      "loss": 0.5993,
+      "step": 16549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942036419594768e-05,
+      "loss": 0.4589,
+      "step": 16550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941972300589895e-05,
+      "loss": 0.592,
+      "step": 16551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941908181585022e-05,
+      "loss": 0.6095,
+      "step": 16552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941844062580149e-05,
+      "loss": 0.5099,
+      "step": 16553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9417799435752764e-05,
+      "loss": 0.6451,
+      "step": 16554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941715824570403e-05,
+      "loss": 0.5402,
+      "step": 16555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94165170556553e-05,
+      "loss": 0.5049,
+      "step": 16556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9415875865606565e-05,
+      "loss": 0.5547,
+      "step": 16557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941523467555784e-05,
+      "loss": 0.4602,
+      "step": 16558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94145934855091e-05,
+      "loss": 0.5528,
+      "step": 16559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9413952295460374e-05,
+      "loss": 0.4882,
+      "step": 16560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9413311105411645e-05,
+      "loss": 0.5626,
+      "step": 16561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941266991536292e-05,
+      "loss": 0.5917,
+      "step": 16562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941202872531419e-05,
+      "loss": 0.4952,
+      "step": 16563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9411387535265454e-05,
+      "loss": 0.4943,
+      "step": 16564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9410746345216725e-05,
+      "loss": 0.4498,
+      "step": 16565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941010515516799e-05,
+      "loss": 0.5167,
+      "step": 16566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940946396511926e-05,
+      "loss": 0.4649,
+      "step": 16567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9408822775070534e-05,
+      "loss": 0.6277,
+      "step": 16568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9408181585021805e-05,
+      "loss": 0.5787,
+      "step": 16569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940754039497307e-05,
+      "loss": 0.4566,
+      "step": 16570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940689920492434e-05,
+      "loss": 0.4664,
+      "step": 16571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9406258014875614e-05,
+      "loss": 0.4307,
+      "step": 16572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940561682482688e-05,
+      "loss": 0.6261,
+      "step": 16573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940497563477815e-05,
+      "loss": 0.5564,
+      "step": 16574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9404334444729415e-05,
+      "loss": 0.4771,
+      "step": 16575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940369325468069e-05,
+      "loss": 0.416,
+      "step": 16576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940305206463196e-05,
+      "loss": 0.4927,
+      "step": 16577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940241087458323e-05,
+      "loss": 0.5925,
+      "step": 16578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9401769684534496e-05,
+      "loss": 0.4712,
+      "step": 16579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940112849448577e-05,
+      "loss": 0.5082,
+      "step": 16580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940048730443704e-05,
+      "loss": 0.493,
+      "step": 16581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9399846114388304e-05,
+      "loss": 0.5534,
+      "step": 16582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9399204924339576e-05,
+      "loss": 0.5088,
+      "step": 16583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939856373429085e-05,
+      "loss": 0.5574,
+      "step": 16584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939792254424212e-05,
+      "loss": 0.4111,
+      "step": 16585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9397281354193384e-05,
+      "loss": 0.5003,
+      "step": 16586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9396640164144656e-05,
+      "loss": 0.4524,
+      "step": 16587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939599897409592e-05,
+      "loss": 0.515,
+      "step": 16588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939535778404719e-05,
+      "loss": 0.5155,
+      "step": 16589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9394716593998464e-05,
+      "loss": 0.5867,
+      "step": 16590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939407540394973e-05,
+      "loss": 0.4961,
+      "step": 16591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9393434213901e-05,
+      "loss": 0.5865,
+      "step": 16592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939279302385227e-05,
+      "loss": 0.5116,
+      "step": 16593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9392151833803544e-05,
+      "loss": 0.5187,
+      "step": 16594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939151064375481e-05,
+      "loss": 0.569,
+      "step": 16595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939086945370608e-05,
+      "loss": 0.6125,
+      "step": 16596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9390228263657346e-05,
+      "loss": 0.5157,
+      "step": 16597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938958707360862e-05,
+      "loss": 0.4514,
+      "step": 16598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938894588355989e-05,
+      "loss": 0.5609,
+      "step": 16599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938830469351116e-05,
+      "loss": 0.517,
+      "step": 16600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938766350346243e-05,
+      "loss": 0.4667,
+      "step": 16601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93870223134137e-05,
+      "loss": 0.5336,
+      "step": 16602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938638112336497e-05,
+      "loss": 0.529,
+      "step": 16603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9385739933316234e-05,
+      "loss": 0.4986,
+      "step": 16604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9385098743267506e-05,
+      "loss": 0.438,
+      "step": 16605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938445755321877e-05,
+      "loss": 0.5506,
+      "step": 16606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938381636317004e-05,
+      "loss": 0.4483,
+      "step": 16607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9383175173121314e-05,
+      "loss": 0.5787,
+      "step": 16608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9382533983072586e-05,
+      "loss": 0.4792,
+      "step": 16609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938189279302386e-05,
+      "loss": 0.5116,
+      "step": 16610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.938125160297512e-05,
+      "loss": 0.5856,
+      "step": 16611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9380610412926394e-05,
+      "loss": 0.689,
+      "step": 16612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937996922287766e-05,
+      "loss": 0.4858,
+      "step": 16613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937932803282893e-05,
+      "loss": 0.505,
+      "step": 16614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93786868427802e-05,
+      "loss": 0.6311,
+      "step": 16615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9378045652731475e-05,
+      "loss": 0.5274,
+      "step": 16616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9377404462682746e-05,
+      "loss": 0.5994,
+      "step": 16617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937676327263401e-05,
+      "loss": 0.5364,
+      "step": 16618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937612208258528e-05,
+      "loss": 0.5762,
+      "step": 16619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937548089253655e-05,
+      "loss": 0.5408,
+      "step": 16620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937483970248782e-05,
+      "loss": 0.5444,
+      "step": 16621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9374198512439085e-05,
+      "loss": 0.4689,
+      "step": 16622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9373557322390356e-05,
+      "loss": 0.4639,
+      "step": 16623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937291613234163e-05,
+      "loss": 0.5444,
+      "step": 16624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93722749422929e-05,
+      "loss": 0.5582,
+      "step": 16625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937163375224417e-05,
+      "loss": 0.5582,
+      "step": 16626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9370992562195436e-05,
+      "loss": 0.4368,
+      "step": 16627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937035137214671e-05,
+      "loss": 0.5408,
+      "step": 16628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936971018209797e-05,
+      "loss": 0.5154,
+      "step": 16629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9369068992049245e-05,
+      "loss": 0.4541,
+      "step": 16630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936842780200051e-05,
+      "loss": 0.6084,
+      "step": 16631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936778661195178e-05,
+      "loss": 0.4026,
+      "step": 16632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936714542190306e-05,
+      "loss": 0.4245,
+      "step": 16633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9366504231854325e-05,
+      "loss": 0.5518,
+      "step": 16634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9365863041805596e-05,
+      "loss": 0.424,
+      "step": 16635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936522185175686e-05,
+      "loss": 0.5981,
+      "step": 16636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936458066170813e-05,
+      "loss": 0.6631,
+      "step": 16637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93639394716594e-05,
+      "loss": 0.5858,
+      "step": 16638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936329828161067e-05,
+      "loss": 0.5734,
+      "step": 16639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936265709156194e-05,
+      "loss": 0.5155,
+      "step": 16640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936201590151321e-05,
+      "loss": 0.52,
+      "step": 16641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9361374711464485e-05,
+      "loss": 0.4926,
+      "step": 16642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936073352141575e-05,
+      "loss": 0.5235,
+      "step": 16643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936009233136702e-05,
+      "loss": 0.4974,
+      "step": 16644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9359451141318287e-05,
+      "loss": 0.4881,
+      "step": 16645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935880995126956e-05,
+      "loss": 0.6063,
+      "step": 16646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935816876122082e-05,
+      "loss": 0.4382,
+      "step": 16647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9357527571172095e-05,
+      "loss": 0.5107,
+      "step": 16648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935688638112337e-05,
+      "loss": 0.5384,
+      "step": 16649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935624519107464e-05,
+      "loss": 0.5668,
+      "step": 16650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935560400102591e-05,
+      "loss": 0.4973,
+      "step": 16651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9354962810977175e-05,
+      "loss": 0.4784,
+      "step": 16652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935432162092845e-05,
+      "loss": 0.4774,
+      "step": 16653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935368043087971e-05,
+      "loss": 0.5528,
+      "step": 16654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9353039240830983e-05,
+      "loss": 0.4847,
+      "step": 16655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9352398050782255e-05,
+      "loss": 0.5845,
+      "step": 16656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935175686073353e-05,
+      "loss": 0.5521,
+      "step": 16657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935111567068479e-05,
+      "loss": 0.5119,
+      "step": 16658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9350474480636063e-05,
+      "loss": 0.4855,
+      "step": 16659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9349833290587335e-05,
+      "loss": 0.4962,
+      "step": 16660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93491921005386e-05,
+      "loss": 0.4944,
+      "step": 16661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934855091048987e-05,
+      "loss": 0.5119,
+      "step": 16662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934790972044114e-05,
+      "loss": 0.5448,
+      "step": 16663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934726853039241e-05,
+      "loss": 0.5398,
+      "step": 16664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934662734034368e-05,
+      "loss": 0.5341,
+      "step": 16665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934598615029495e-05,
+      "loss": 0.5241,
+      "step": 16666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934534496024622e-05,
+      "loss": 0.4612,
+      "step": 16667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934470377019749e-05,
+      "loss": 0.5125,
+      "step": 16668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934406258014876e-05,
+      "loss": 0.4056,
+      "step": 16669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9343421390100025e-05,
+      "loss": 0.5999,
+      "step": 16670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93427802000513e-05,
+      "loss": 0.484,
+      "step": 16671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934213901000257e-05,
+      "loss": 0.5045,
+      "step": 16672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934149781995384e-05,
+      "loss": 0.6612,
+      "step": 16673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9340856629905105e-05,
+      "loss": 0.4757,
+      "step": 16674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934021543985638e-05,
+      "loss": 0.5283,
+      "step": 16675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933957424980764e-05,
+      "loss": 0.5068,
+      "step": 16676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9338933059758914e-05,
+      "loss": 0.5029,
+      "step": 16677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9338291869710185e-05,
+      "loss": 0.6073,
+      "step": 16678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933765067966145e-05,
+      "loss": 0.5757,
+      "step": 16679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933700948961272e-05,
+      "loss": 0.5624,
+      "step": 16680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9336368299563994e-05,
+      "loss": 0.685,
+      "step": 16681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9335727109515266e-05,
+      "loss": 0.5548,
+      "step": 16682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933508591946653e-05,
+      "loss": 0.5274,
+      "step": 16683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93344447294178e-05,
+      "loss": 0.5466,
+      "step": 16684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933380353936907e-05,
+      "loss": 0.421,
+      "step": 16685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933316234932034e-05,
+      "loss": 0.645,
+      "step": 16686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933252115927161e-05,
+      "loss": 0.4927,
+      "step": 16687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933187996922288e-05,
+      "loss": 0.4892,
+      "step": 16688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9331238779174154e-05,
+      "loss": 0.5418,
+      "step": 16689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.933059758912542e-05,
+      "loss": 0.6399,
+      "step": 16690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932995639907669e-05,
+      "loss": 0.5261,
+      "step": 16691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9329315209027956e-05,
+      "loss": 0.6259,
+      "step": 16692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932867401897923e-05,
+      "loss": 0.5264,
+      "step": 16693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932803282893049e-05,
+      "loss": 0.5281,
+      "step": 16694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9327391638881764e-05,
+      "loss": 0.5507,
+      "step": 16695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9326750448833036e-05,
+      "loss": 0.5035,
+      "step": 16696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932610925878431e-05,
+      "loss": 0.5181,
+      "step": 16697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932546806873558e-05,
+      "loss": 0.4544,
+      "step": 16698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9324826878686844e-05,
+      "loss": 0.4701,
+      "step": 16699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9324185688638116e-05,
+      "loss": 0.5006,
+      "step": 16700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932354449858938e-05,
+      "loss": 0.4679,
+      "step": 16701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932290330854065e-05,
+      "loss": 0.5187,
+      "step": 16702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932226211849192e-05,
+      "loss": 0.5946,
+      "step": 16703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932162092844319e-05,
+      "loss": 0.6074,
+      "step": 16704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932097973839447e-05,
+      "loss": 0.5016,
+      "step": 16705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.932033854834573e-05,
+      "loss": 0.4696,
+      "step": 16706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9319697358297004e-05,
+      "loss": 0.5769,
+      "step": 16707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931905616824827e-05,
+      "loss": 0.4161,
+      "step": 16708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931841497819954e-05,
+      "loss": 0.4854,
+      "step": 16709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9317773788150806e-05,
+      "loss": 0.4792,
+      "step": 16710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931713259810208e-05,
+      "loss": 0.5376,
+      "step": 16711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931649140805335e-05,
+      "loss": 0.6037,
+      "step": 16712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931585021800462e-05,
+      "loss": 0.5094,
+      "step": 16713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931520902795589e-05,
+      "loss": 0.559,
+      "step": 16714
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931456783790716e-05,
+      "loss": 0.47,
+      "step": 16715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931392664785843e-05,
+      "loss": 0.5092,
+      "step": 16716
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9313285457809694e-05,
+      "loss": 0.5367,
+      "step": 16717
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9312644267760966e-05,
+      "loss": 0.5795,
+      "step": 16718
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931200307771223e-05,
+      "loss": 0.5909,
+      "step": 16719
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93113618876635e-05,
+      "loss": 0.5546,
+      "step": 16720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9310720697614774e-05,
+      "loss": 0.6774,
+      "step": 16721
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9310079507566046e-05,
+      "loss": 0.5758,
+      "step": 16722
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930943831751732e-05,
+      "loss": 0.32,
+      "step": 16723
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930879712746858e-05,
+      "loss": 0.4976,
+      "step": 16724
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9308155937419854e-05,
+      "loss": 0.4816,
+      "step": 16725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930751474737112e-05,
+      "loss": 0.5098,
+      "step": 16726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930687355732239e-05,
+      "loss": 0.4741,
+      "step": 16727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930623236727366e-05,
+      "loss": 0.535,
+      "step": 16728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9305591177224935e-05,
+      "loss": 0.5895,
+      "step": 16729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93049499871762e-05,
+      "loss": 0.4564,
+      "step": 16730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930430879712747e-05,
+      "loss": 0.5468,
+      "step": 16731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930366760707874e-05,
+      "loss": 0.5092,
+      "step": 16732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930302641703001e-05,
+      "loss": 0.4819,
+      "step": 16733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930238522698128e-05,
+      "loss": 0.5898,
+      "step": 16734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9301744036932545e-05,
+      "loss": 0.7092,
+      "step": 16735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9301102846883816e-05,
+      "loss": 0.5236,
+      "step": 16736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.930046165683509e-05,
+      "loss": 0.4927,
+      "step": 16737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929982046678636e-05,
+      "loss": 0.5247,
+      "step": 16738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9299179276737625e-05,
+      "loss": 0.6105,
+      "step": 16739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9298538086688896e-05,
+      "loss": 0.5805,
+      "step": 16740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929789689664017e-05,
+      "loss": 0.4892,
+      "step": 16741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929725570659143e-05,
+      "loss": 0.5344,
+      "step": 16742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9296614516542705e-05,
+      "loss": 0.578,
+      "step": 16743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9295973326493976e-05,
+      "loss": 0.6221,
+      "step": 16744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929533213644525e-05,
+      "loss": 0.6342,
+      "step": 16745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929469094639651e-05,
+      "loss": 0.49,
+      "step": 16746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9294049756347785e-05,
+      "loss": 0.5761,
+      "step": 16747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929340856629905e-05,
+      "loss": 0.6066,
+      "step": 16748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929276737625032e-05,
+      "loss": 0.4375,
+      "step": 16749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929212618620159e-05,
+      "loss": 0.6229,
+      "step": 16750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929148499615286e-05,
+      "loss": 0.482,
+      "step": 16751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.929084380610413e-05,
+      "loss": 0.5925,
+      "step": 16752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.92902026160554e-05,
+      "loss": 0.4467,
+      "step": 16753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928956142600667e-05,
+      "loss": 0.5649,
+      "step": 16754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928892023595794e-05,
+      "loss": 0.4101,
+      "step": 16755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928827904590921e-05,
+      "loss": 0.5833,
+      "step": 16756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9287637855860475e-05,
+      "loss": 0.6071,
+      "step": 16757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9286996665811747e-05,
+      "loss": 0.5784,
+      "step": 16758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928635547576302e-05,
+      "loss": 0.5506,
+      "step": 16759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.5466,
+      "step": 16760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928507309566556e-05,
+      "loss": 0.4904,
+      "step": 16761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928443190561683e-05,
+      "loss": 0.5459,
+      "step": 16762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.92837907155681e-05,
+      "loss": 0.4972,
+      "step": 16763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928314952551936e-05,
+      "loss": 0.5341,
+      "step": 16764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9282508335470635e-05,
+      "loss": 0.5071,
+      "step": 16765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.92818671454219e-05,
+      "loss": 0.5509,
+      "step": 16766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928122595537317e-05,
+      "loss": 0.5095,
+      "step": 16767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9280584765324443e-05,
+      "loss": 0.4965,
+      "step": 16768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9279943575275715e-05,
+      "loss": 0.539,
+      "step": 16769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.927930238522699e-05,
+      "loss": 0.5729,
+      "step": 16770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.927866119517825e-05,
+      "loss": 0.5493,
+      "step": 16771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9278020005129524e-05,
+      "loss": 0.4975,
+      "step": 16772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.927737881508079e-05,
+      "loss": 0.6071,
+      "step": 16773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.927673762503206e-05,
+      "loss": 0.5151,
+      "step": 16774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9276096434983325e-05,
+      "loss": 0.636,
+      "step": 16775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9275455244934604e-05,
+      "loss": 0.5363,
+      "step": 16776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9274814054885875e-05,
+      "loss": 0.5455,
+      "step": 16777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.927417286483714e-05,
+      "loss": 0.5732,
+      "step": 16778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.927353167478841e-05,
+      "loss": 0.4977,
+      "step": 16779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.927289048473968e-05,
+      "loss": 0.5365,
+      "step": 16780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.927224929469095e-05,
+      "loss": 0.4704,
+      "step": 16781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9271608104642214e-05,
+      "loss": 0.5189,
+      "step": 16782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9270966914593485e-05,
+      "loss": 0.5602,
+      "step": 16783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.927032572454476e-05,
+      "loss": 0.4928,
+      "step": 16784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926968453449603e-05,
+      "loss": 0.5534,
+      "step": 16785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92690433444473e-05,
+      "loss": 0.4099,
+      "step": 16786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9268402154398565e-05,
+      "loss": 0.5849,
+      "step": 16787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926776096434984e-05,
+      "loss": 0.4891,
+      "step": 16788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92671197743011e-05,
+      "loss": 0.5514,
+      "step": 16789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9266478584252374e-05,
+      "loss": 0.5636,
+      "step": 16790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926583739420364e-05,
+      "loss": 0.5583,
+      "step": 16791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926519620415491e-05,
+      "loss": 0.52,
+      "step": 16792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926455501410618e-05,
+      "loss": 0.5333,
+      "step": 16793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9263913824057454e-05,
+      "loss": 0.4931,
+      "step": 16794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9263272634008726e-05,
+      "loss": 0.5332,
+      "step": 16795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926263144395999e-05,
+      "loss": 0.5433,
+      "step": 16796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926199025391126e-05,
+      "loss": 0.5379,
+      "step": 16797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926134906386253e-05,
+      "loss": 0.56,
+      "step": 16798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92607078738138e-05,
+      "loss": 0.5419,
+      "step": 16799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926006668376507e-05,
+      "loss": 0.4915,
+      "step": 16800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925942549371634e-05,
+      "loss": 0.7326,
+      "step": 16801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9258784303667614e-05,
+      "loss": 0.5467,
+      "step": 16802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925814311361888e-05,
+      "loss": 0.4319,
+      "step": 16803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925750192357015e-05,
+      "loss": 0.6519,
+      "step": 16804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9256860733521416e-05,
+      "loss": 0.5518,
+      "step": 16805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925621954347269e-05,
+      "loss": 0.4997,
+      "step": 16806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925557835342395e-05,
+      "loss": 0.573,
+      "step": 16807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9254937163375224e-05,
+      "loss": 0.5731,
+      "step": 16808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9254295973326496e-05,
+      "loss": 0.5051,
+      "step": 16809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925365478327777e-05,
+      "loss": 0.4981,
+      "step": 16810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925301359322904e-05,
+      "loss": 0.4798,
+      "step": 16811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9252372403180304e-05,
+      "loss": 0.4704,
+      "step": 16812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9251731213131576e-05,
+      "loss": 0.5668,
+      "step": 16813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925109002308284e-05,
+      "loss": 0.439,
+      "step": 16814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925044883303411e-05,
+      "loss": 0.5833,
+      "step": 16815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9249807642985384e-05,
+      "loss": 0.6226,
+      "step": 16816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9249166452936656e-05,
+      "loss": 0.479,
+      "step": 16817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924852526288792e-05,
+      "loss": 0.5474,
+      "step": 16818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924788407283919e-05,
+      "loss": 0.5035,
+      "step": 16819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9247242882790464e-05,
+      "loss": 0.445,
+      "step": 16820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924660169274173e-05,
+      "loss": 0.4646,
+      "step": 16821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9245960502693e-05,
+      "loss": 0.4,
+      "step": 16822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9245319312644266e-05,
+      "loss": 0.5033,
+      "step": 16823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924467812259554e-05,
+      "loss": 0.6146,
+      "step": 16824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924403693254681e-05,
+      "loss": 0.5452,
+      "step": 16825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924339574249808e-05,
+      "loss": 0.5356,
+      "step": 16826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9242754552449346e-05,
+      "loss": 0.5449,
+      "step": 16827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924211336240062e-05,
+      "loss": 0.4982,
+      "step": 16828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924147217235189e-05,
+      "loss": 0.5464,
+      "step": 16829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9240830982303154e-05,
+      "loss": 0.6589,
+      "step": 16830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9240189792254426e-05,
+      "loss": 0.5227,
+      "step": 16831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92395486022057e-05,
+      "loss": 0.5875,
+      "step": 16832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923890741215697e-05,
+      "loss": 0.4353,
+      "step": 16833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9238266222108234e-05,
+      "loss": 0.4933,
+      "step": 16834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9237625032059506e-05,
+      "loss": 0.6206,
+      "step": 16835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923698384201077e-05,
+      "loss": 0.5259,
+      "step": 16836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923634265196204e-05,
+      "loss": 0.5051,
+      "step": 16837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9235701461913315e-05,
+      "loss": 0.5028,
+      "step": 16838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923506027186458e-05,
+      "loss": 0.5638,
+      "step": 16839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923441908181585e-05,
+      "loss": 0.6427,
+      "step": 16840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923377789176712e-05,
+      "loss": 0.5281,
+      "step": 16841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9233136701718395e-05,
+      "loss": 0.4617,
+      "step": 16842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923249551166966e-05,
+      "loss": 0.521,
+      "step": 16843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923185432162093e-05,
+      "loss": 0.5333,
+      "step": 16844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9231213131572196e-05,
+      "loss": 0.5665,
+      "step": 16845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923057194152347e-05,
+      "loss": 0.4678,
+      "step": 16846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922993075147474e-05,
+      "loss": 0.5438,
+      "step": 16847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922928956142601e-05,
+      "loss": 0.453,
+      "step": 16848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922864837137728e-05,
+      "loss": 0.4656,
+      "step": 16849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922800718132855e-05,
+      "loss": 0.5085,
+      "step": 16850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922736599127982e-05,
+      "loss": 0.4638,
+      "step": 16851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9226724801231085e-05,
+      "loss": 0.4859,
+      "step": 16852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9226083611182356e-05,
+      "loss": 0.5636,
+      "step": 16853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922544242113362e-05,
+      "loss": 0.5654,
+      "step": 16854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922480123108489e-05,
+      "loss": 0.4735,
+      "step": 16855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9224160041036165e-05,
+      "loss": 0.6318,
+      "step": 16856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9223518850987436e-05,
+      "loss": 0.4551,
+      "step": 16857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922287766093871e-05,
+      "loss": 0.5102,
+      "step": 16858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922223647088997e-05,
+      "loss": 0.3986,
+      "step": 16859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9221595280841245e-05,
+      "loss": 0.5568,
+      "step": 16860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922095409079251e-05,
+      "loss": 0.6608,
+      "step": 16861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922031290074378e-05,
+      "loss": 0.5331,
+      "step": 16862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9219671710695046e-05,
+      "loss": 0.422,
+      "step": 16863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9219030520646325e-05,
+      "loss": 0.5414,
+      "step": 16864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92183893305976e-05,
+      "loss": 0.5022,
+      "step": 16865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921774814054886e-05,
+      "loss": 0.5029,
+      "step": 16866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921710695050013e-05,
+      "loss": 0.5507,
+      "step": 16867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92164657604514e-05,
+      "loss": 0.4207,
+      "step": 16868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921582457040267e-05,
+      "loss": 0.545,
+      "step": 16869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9215183380353935e-05,
+      "loss": 0.4994,
+      "step": 16870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921454219030521e-05,
+      "loss": 0.5144,
+      "step": 16871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921390100025648e-05,
+      "loss": 0.5372,
+      "step": 16872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921325981020775e-05,
+      "loss": 0.543,
+      "step": 16873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921261862015902e-05,
+      "loss": 0.5043,
+      "step": 16874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921197743011029e-05,
+      "loss": 0.5958,
+      "step": 16875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921133624006156e-05,
+      "loss": 0.4842,
+      "step": 16876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921069505001282e-05,
+      "loss": 0.5748,
+      "step": 16877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9210053859964095e-05,
+      "loss": 0.5483,
+      "step": 16878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920941266991536e-05,
+      "loss": 0.4843,
+      "step": 16879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920877147986663e-05,
+      "loss": 0.5201,
+      "step": 16880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9208130289817903e-05,
+      "loss": 0.5107,
+      "step": 16881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9207489099769175e-05,
+      "loss": 0.5284,
+      "step": 16882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920684790972045e-05,
+      "loss": 0.4761,
+      "step": 16883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920620671967171e-05,
+      "loss": 0.5258,
+      "step": 16884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9205565529622984e-05,
+      "loss": 0.6007,
+      "step": 16885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920492433957425e-05,
+      "loss": 0.497,
+      "step": 16886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920428314952552e-05,
+      "loss": 0.5847,
+      "step": 16887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920364195947679e-05,
+      "loss": 0.6517,
+      "step": 16888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9203000769428064e-05,
+      "loss": 0.5004,
+      "step": 16889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920235957937933e-05,
+      "loss": 0.5344,
+      "step": 16890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92017183893306e-05,
+      "loss": 0.4717,
+      "step": 16891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920107719928187e-05,
+      "loss": 0.5614,
+      "step": 16892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920043600923314e-05,
+      "loss": 0.6614,
+      "step": 16893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919979481918441e-05,
+      "loss": 0.4299,
+      "step": 16894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9199153629135674e-05,
+      "loss": 0.6534,
+      "step": 16895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9198512439086945e-05,
+      "loss": 0.5977,
+      "step": 16896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919787124903822e-05,
+      "loss": 0.6489,
+      "step": 16897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919723005898949e-05,
+      "loss": 0.541,
+      "step": 16898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9196588868940754e-05,
+      "loss": 0.4427,
+      "step": 16899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9195947678892025e-05,
+      "loss": 0.568,
+      "step": 16900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91953064888433e-05,
+      "loss": 0.6264,
+      "step": 16901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919466529879456e-05,
+      "loss": 0.4856,
+      "step": 16902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9194024108745834e-05,
+      "loss": 0.6205,
+      "step": 16903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9193382918697106e-05,
+      "loss": 0.5313,
+      "step": 16904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919274172864838e-05,
+      "loss": 0.4556,
+      "step": 16905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919210053859964e-05,
+      "loss": 0.5378,
+      "step": 16906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9191459348550914e-05,
+      "loss": 0.4838,
+      "step": 16907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919081815850218e-05,
+      "loss": 0.5854,
+      "step": 16908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919017696845345e-05,
+      "loss": 0.5268,
+      "step": 16909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918953577840472e-05,
+      "loss": 0.5619,
+      "step": 16910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918889458835599e-05,
+      "loss": 0.5389,
+      "step": 16911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918825339830726e-05,
+      "loss": 0.5103,
+      "step": 16912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918761220825853e-05,
+      "loss": 0.4771,
+      "step": 16913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91869710182098e-05,
+      "loss": 0.5092,
+      "step": 16914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918632982816107e-05,
+      "loss": 0.4887,
+      "step": 16915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918568863811234e-05,
+      "loss": 0.4515,
+      "step": 16916
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9185047448063604e-05,
+      "loss": 0.5365,
+      "step": 16917
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9184406258014876e-05,
+      "loss": 0.5823,
+      "step": 16918
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918376506796615e-05,
+      "loss": 0.4836,
+      "step": 16919
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918312387791742e-05,
+      "loss": 0.5979,
+      "step": 16920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918248268786869e-05,
+      "loss": 0.5564,
+      "step": 16921
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9181841497819956e-05,
+      "loss": 0.5359,
+      "step": 16922
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918120030777123e-05,
+      "loss": 0.5239,
+      "step": 16923
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918055911772249e-05,
+      "loss": 0.6255,
+      "step": 16924
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9179917927673764e-05,
+      "loss": 0.4445,
+      "step": 16925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917927673762503e-05,
+      "loss": 0.5622,
+      "step": 16926
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91786355475763e-05,
+      "loss": 0.5222,
+      "step": 16927
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917799435752757e-05,
+      "loss": 0.6394,
+      "step": 16928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9177353167478844e-05,
+      "loss": 0.6195,
+      "step": 16929
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9176711977430116e-05,
+      "loss": 0.5756,
+      "step": 16930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917607078738138e-05,
+      "loss": 0.6447,
+      "step": 16931
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917542959733265e-05,
+      "loss": 0.4711,
+      "step": 16932
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917478840728392e-05,
+      "loss": 0.5966,
+      "step": 16933
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917414721723519e-05,
+      "loss": 0.4911,
+      "step": 16934
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9173506027186454e-05,
+      "loss": 0.4276,
+      "step": 16935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917286483713773e-05,
+      "loss": 0.6001,
+      "step": 16936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9172223647089004e-05,
+      "loss": 0.5403,
+      "step": 16937
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917158245704027e-05,
+      "loss": 0.5065,
+      "step": 16938
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917094126699154e-05,
+      "loss": 0.4965,
+      "step": 16939
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9170300076942806e-05,
+      "loss": 0.4641,
+      "step": 16940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916965888689408e-05,
+      "loss": 0.5196,
+      "step": 16941
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916901769684534e-05,
+      "loss": 0.4812,
+      "step": 16942
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9168376506796614e-05,
+      "loss": 0.611,
+      "step": 16943
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9167735316747886e-05,
+      "loss": 0.586,
+      "step": 16944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916709412669916e-05,
+      "loss": 0.5404,
+      "step": 16945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916645293665043e-05,
+      "loss": 0.4786,
+      "step": 16946
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9165811746601694e-05,
+      "loss": 0.4959,
+      "step": 16947
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9165170556552966e-05,
+      "loss": 0.5183,
+      "step": 16948
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916452936650423e-05,
+      "loss": 0.601,
+      "step": 16949
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91638881764555e-05,
+      "loss": 0.6084,
+      "step": 16950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916324698640677e-05,
+      "loss": 0.61,
+      "step": 16951
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916260579635804e-05,
+      "loss": 0.4364,
+      "step": 16952
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916196460630931e-05,
+      "loss": 0.4748,
+      "step": 16953
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916132341626058e-05,
+      "loss": 0.53,
+      "step": 16954
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9160682226211855e-05,
+      "loss": 0.5281,
+      "step": 16955
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916004103616312e-05,
+      "loss": 0.4826,
+      "step": 16956
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915939984611439e-05,
+      "loss": 0.5396,
+      "step": 16957
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9158758656065656e-05,
+      "loss": 0.4276,
+      "step": 16958
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915811746601693e-05,
+      "loss": 0.4837,
+      "step": 16959
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91574762759682e-05,
+      "loss": 0.4891,
+      "step": 16960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915683508591947e-05,
+      "loss": 0.52,
+      "step": 16961
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9156193895870736e-05,
+      "loss": 0.5218,
+      "step": 16962
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915555270582201e-05,
+      "loss": 0.5146,
+      "step": 16963
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915491151577328e-05,
+      "loss": 0.6021,
+      "step": 16964
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9154270325724545e-05,
+      "loss": 0.6102,
+      "step": 16965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9153629135675816e-05,
+      "loss": 0.4402,
+      "step": 16966
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915298794562708e-05,
+      "loss": 0.4004,
+      "step": 16967
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915234675557835e-05,
+      "loss": 0.4843,
+      "step": 16968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9151705565529625e-05,
+      "loss": 0.565,
+      "step": 16969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9151064375480896e-05,
+      "loss": 0.3628,
+      "step": 16970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915042318543216e-05,
+      "loss": 0.4751,
+      "step": 16971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914978199538343e-05,
+      "loss": 0.4937,
+      "step": 16972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9149140805334705e-05,
+      "loss": 0.5352,
+      "step": 16973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914849961528597e-05,
+      "loss": 0.5213,
+      "step": 16974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914785842523724e-05,
+      "loss": 0.5333,
+      "step": 16975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914721723518851e-05,
+      "loss": 0.4426,
+      "step": 16976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9146576045139785e-05,
+      "loss": 0.5067,
+      "step": 16977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914593485509105e-05,
+      "loss": 0.5031,
+      "step": 16978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914529366504232e-05,
+      "loss": 0.5405,
+      "step": 16979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9144652474993587e-05,
+      "loss": 0.4668,
+      "step": 16980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914401128494486e-05,
+      "loss": 0.4182,
+      "step": 16981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914337009489613e-05,
+      "loss": 0.63,
+      "step": 16982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9142728904847395e-05,
+      "loss": 0.4209,
+      "step": 16983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914208771479867e-05,
+      "loss": 0.5007,
+      "step": 16984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914144652474994e-05,
+      "loss": 0.5384,
+      "step": 16985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914080533470121e-05,
+      "loss": 0.7031,
+      "step": 16986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9140164144652475e-05,
+      "loss": 0.4983,
+      "step": 16987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913952295460375e-05,
+      "loss": 0.3968,
+      "step": 16988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913888176455501e-05,
+      "loss": 0.6151,
+      "step": 16989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9138240574506283e-05,
+      "loss": 0.5113,
+      "step": 16990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9137599384457555e-05,
+      "loss": 0.5381,
+      "step": 16991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913695819440883e-05,
+      "loss": 0.5264,
+      "step": 16992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91363170043601e-05,
+      "loss": 0.4888,
+      "step": 16993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9135675814311363e-05,
+      "loss": 0.5253,
+      "step": 16994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9135034624262635e-05,
+      "loss": 0.4513,
+      "step": 16995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91343934342139e-05,
+      "loss": 0.5356,
+      "step": 16996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913375224416517e-05,
+      "loss": 0.5917,
+      "step": 16997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9133111054116444e-05,
+      "loss": 0.6284,
+      "step": 16998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913246986406771e-05,
+      "loss": 0.5118,
+      "step": 16999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913182867401898e-05,
+      "loss": 0.5373,
+      "step": 17000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913118748397025e-05,
+      "loss": 0.5616,
+      "step": 17001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9130546293921524e-05,
+      "loss": 0.6254,
+      "step": 17002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912990510387279e-05,
+      "loss": 0.4471,
+      "step": 17003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912926391382406e-05,
+      "loss": 0.5542,
+      "step": 17004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9128622723775325e-05,
+      "loss": 0.5869,
+      "step": 17005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91279815337266e-05,
+      "loss": 0.5018,
+      "step": 17006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912734034367787e-05,
+      "loss": 0.4992,
+      "step": 17007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912669915362914e-05,
+      "loss": 0.5971,
+      "step": 17008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912605796358041e-05,
+      "loss": 0.5948,
+      "step": 17009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912541677353168e-05,
+      "loss": 0.53,
+      "step": 17010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912477558348295e-05,
+      "loss": 0.5254,
+      "step": 17011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9124134393434214e-05,
+      "loss": 0.6823,
+      "step": 17012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9123493203385485e-05,
+      "loss": 0.5171,
+      "step": 17013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912285201333675e-05,
+      "loss": 0.5835,
+      "step": 17014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912221082328802e-05,
+      "loss": 0.5654,
+      "step": 17015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9121569633239294e-05,
+      "loss": 0.595,
+      "step": 17016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9120928443190566e-05,
+      "loss": 0.5581,
+      "step": 17017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912028725314184e-05,
+      "loss": 0.4323,
+      "step": 17018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91196460630931e-05,
+      "loss": 0.4992,
+      "step": 17019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9119004873044374e-05,
+      "loss": 0.5449,
+      "step": 17020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.911836368299564e-05,
+      "loss": 0.5364,
+      "step": 17021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.911772249294691e-05,
+      "loss": 0.4796,
+      "step": 17022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9117081302898176e-05,
+      "loss": 0.5846,
+      "step": 17023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9116440112849454e-05,
+      "loss": 0.5787,
+      "step": 17024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9115798922800726e-05,
+      "loss": 0.4341,
+      "step": 17025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.911515773275199e-05,
+      "loss": 0.5112,
+      "step": 17026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.911451654270326e-05,
+      "loss": 0.5184,
+      "step": 17027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.911387535265453e-05,
+      "loss": 0.5246,
+      "step": 17028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91132341626058e-05,
+      "loss": 0.4601,
+      "step": 17029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9112592972557064e-05,
+      "loss": 0.5064,
+      "step": 17030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9111951782508336e-05,
+      "loss": 0.5036,
+      "step": 17031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.911131059245961e-05,
+      "loss": 0.4559,
+      "step": 17032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.911066940241088e-05,
+      "loss": 0.4684,
+      "step": 17033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.911002821236215e-05,
+      "loss": 0.5086,
+      "step": 17034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9109387022313416e-05,
+      "loss": 0.5666,
+      "step": 17035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.910874583226469e-05,
+      "loss": 0.585,
+      "step": 17036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910810464221595e-05,
+      "loss": 0.4488,
+      "step": 17037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9107463452167224e-05,
+      "loss": 0.481,
+      "step": 17038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910682226211849e-05,
+      "loss": 0.4873,
+      "step": 17039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910618107206976e-05,
+      "loss": 0.5454,
+      "step": 17040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910553988202103e-05,
+      "loss": 0.5487,
+      "step": 17041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9104898691972304e-05,
+      "loss": 0.5646,
+      "step": 17042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9104257501923576e-05,
+      "loss": 0.4079,
+      "step": 17043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910361631187484e-05,
+      "loss": 0.5737,
+      "step": 17044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910297512182611e-05,
+      "loss": 0.5229,
+      "step": 17045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910233393177738e-05,
+      "loss": 0.6064,
+      "step": 17046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910169274172865e-05,
+      "loss": 0.5459,
+      "step": 17047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910105155167992e-05,
+      "loss": 0.5369,
+      "step": 17048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910041036163119e-05,
+      "loss": 0.4498,
+      "step": 17049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909976917158246e-05,
+      "loss": 0.4505,
+      "step": 17050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909912798153373e-05,
+      "loss": 0.4298,
+      "step": 17051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9098486791485e-05,
+      "loss": 0.6111,
+      "step": 17052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9097845601436266e-05,
+      "loss": 0.4848,
+      "step": 17053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909720441138754e-05,
+      "loss": 0.5899,
+      "step": 17054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90965632213388e-05,
+      "loss": 0.6507,
+      "step": 17055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9095922031290074e-05,
+      "loss": 0.5203,
+      "step": 17056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9095280841241346e-05,
+      "loss": 0.5943,
+      "step": 17057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909463965119262e-05,
+      "loss": 0.5622,
+      "step": 17058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909399846114388e-05,
+      "loss": 0.5229,
+      "step": 17059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9093357271095154e-05,
+      "loss": 0.5449,
+      "step": 17060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9092716081046426e-05,
+      "loss": 0.4306,
+      "step": 17061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909207489099769e-05,
+      "loss": 0.5291,
+      "step": 17062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909143370094896e-05,
+      "loss": 0.6127,
+      "step": 17063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9090792510900235e-05,
+      "loss": 0.4664,
+      "step": 17064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9090151320851506e-05,
+      "loss": 0.6191,
+      "step": 17065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908951013080277e-05,
+      "loss": 0.579,
+      "step": 17066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908886894075404e-05,
+      "loss": 0.5213,
+      "step": 17067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908822775070531e-05,
+      "loss": 0.6828,
+      "step": 17068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908758656065658e-05,
+      "loss": 0.5416,
+      "step": 17069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908694537060785e-05,
+      "loss": 0.4385,
+      "step": 17070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9086304180559116e-05,
+      "loss": 0.6299,
+      "step": 17071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908566299051039e-05,
+      "loss": 0.5454,
+      "step": 17072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908502180046166e-05,
+      "loss": 0.4917,
+      "step": 17073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908438061041293e-05,
+      "loss": 0.5451,
+      "step": 17074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9083739420364196e-05,
+      "loss": 0.5494,
+      "step": 17075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908309823031547e-05,
+      "loss": 0.5952,
+      "step": 17076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908245704026673e-05,
+      "loss": 0.5967,
+      "step": 17077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9081815850218005e-05,
+      "loss": 0.629,
+      "step": 17078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9081174660169276e-05,
+      "loss": 0.5145,
+      "step": 17079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908053347012055e-05,
+      "loss": 0.5695,
+      "step": 17080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907989228007182e-05,
+      "loss": 0.5307,
+      "step": 17081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9079251090023085e-05,
+      "loss": 0.4859,
+      "step": 17082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9078609899974357e-05,
+      "loss": 0.5575,
+      "step": 17083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907796870992562e-05,
+      "loss": 0.5111,
+      "step": 17084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907732751987689e-05,
+      "loss": 0.4702,
+      "step": 17085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907668632982816e-05,
+      "loss": 0.6378,
+      "step": 17086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907604513977943e-05,
+      "loss": 0.538,
+      "step": 17087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90754039497307e-05,
+      "loss": 0.5058,
+      "step": 17088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907476275968197e-05,
+      "loss": 0.5263,
+      "step": 17089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9074121569633245e-05,
+      "loss": 0.5875,
+      "step": 17090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907348037958451e-05,
+      "loss": 0.4698,
+      "step": 17091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907283918953578e-05,
+      "loss": 0.5418,
+      "step": 17092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9072197999487047e-05,
+      "loss": 0.4621,
+      "step": 17093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907155680943832e-05,
+      "loss": 0.6803,
+      "step": 17094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907091561938958e-05,
+      "loss": 0.4782,
+      "step": 17095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907027442934086e-05,
+      "loss": 0.5956,
+      "step": 17096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9069633239292133e-05,
+      "loss": 0.4559,
+      "step": 17097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90689920492434e-05,
+      "loss": 0.5731,
+      "step": 17098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906835085919467e-05,
+      "loss": 0.5093,
+      "step": 17099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9067709669145935e-05,
+      "loss": 0.484,
+      "step": 17100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906706847909721e-05,
+      "loss": 0.6517,
+      "step": 17101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906642728904847e-05,
+      "loss": 0.6641,
+      "step": 17102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9065786098999743e-05,
+      "loss": 0.5393,
+      "step": 17103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9065144908951015e-05,
+      "loss": 0.4489,
+      "step": 17104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906450371890229e-05,
+      "loss": 0.5416,
+      "step": 17105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906386252885356e-05,
+      "loss": 0.4811,
+      "step": 17106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9063221338804824e-05,
+      "loss": 0.4629,
+      "step": 17107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9062580148756095e-05,
+      "loss": 0.5468,
+      "step": 17108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906193895870736e-05,
+      "loss": 0.4882,
+      "step": 17109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906129776865863e-05,
+      "loss": 0.5736,
+      "step": 17110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90606565786099e-05,
+      "loss": 0.5902,
+      "step": 17111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906001538856117e-05,
+      "loss": 0.4433,
+      "step": 17112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905937419851244e-05,
+      "loss": 0.6855,
+      "step": 17113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905873300846371e-05,
+      "loss": 0.5164,
+      "step": 17114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9058091818414984e-05,
+      "loss": 0.4798,
+      "step": 17115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905745062836625e-05,
+      "loss": 0.4909,
+      "step": 17116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905680943831752e-05,
+      "loss": 0.4961,
+      "step": 17117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9056168248268785e-05,
+      "loss": 0.5232,
+      "step": 17118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905552705822006e-05,
+      "loss": 0.4344,
+      "step": 17119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905488586817133e-05,
+      "loss": 0.5856,
+      "step": 17120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90542446781226e-05,
+      "loss": 0.3767,
+      "step": 17121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9053603488073865e-05,
+      "loss": 0.4879,
+      "step": 17122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905296229802514e-05,
+      "loss": 0.4634,
+      "step": 17123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905232110797641e-05,
+      "loss": 0.5634,
+      "step": 17124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9051679917927674e-05,
+      "loss": 0.612,
+      "step": 17125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9051038727878945e-05,
+      "loss": 0.5523,
+      "step": 17126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905039753783021e-05,
+      "loss": 0.5922,
+      "step": 17127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904975634778148e-05,
+      "loss": 0.5303,
+      "step": 17128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9049115157732754e-05,
+      "loss": 0.5674,
+      "step": 17129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9048473967684026e-05,
+      "loss": 0.533,
+      "step": 17130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904783277763529e-05,
+      "loss": 0.5754,
+      "step": 17131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904719158758656e-05,
+      "loss": 0.5876,
+      "step": 17132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9046550397537834e-05,
+      "loss": 0.473,
+      "step": 17133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90459092074891e-05,
+      "loss": 0.5229,
+      "step": 17134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904526801744037e-05,
+      "loss": 0.4266,
+      "step": 17135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904462682739164e-05,
+      "loss": 0.553,
+      "step": 17136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9043985637342914e-05,
+      "loss": 0.53,
+      "step": 17137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904334444729418e-05,
+      "loss": 0.5479,
+      "step": 17138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904270325724545e-05,
+      "loss": 0.4305,
+      "step": 17139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9042062067196716e-05,
+      "loss": 0.6807,
+      "step": 17140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904142087714799e-05,
+      "loss": 0.5831,
+      "step": 17141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904077968709926e-05,
+      "loss": 0.5633,
+      "step": 17142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9040138497050524e-05,
+      "loss": 0.4933,
+      "step": 17143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9039497307001796e-05,
+      "loss": 0.639,
+      "step": 17144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903885611695307e-05,
+      "loss": 0.522,
+      "step": 17145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903821492690434e-05,
+      "loss": 0.5415,
+      "step": 17146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9037573736855604e-05,
+      "loss": 0.4923,
+      "step": 17147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9036932546806876e-05,
+      "loss": 0.5441,
+      "step": 17148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903629135675814e-05,
+      "loss": 0.5546,
+      "step": 17149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903565016670941e-05,
+      "loss": 0.5118,
+      "step": 17150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9035008976660684e-05,
+      "loss": 0.4969,
+      "step": 17151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9034367786611956e-05,
+      "loss": 0.5237,
+      "step": 17152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903372659656323e-05,
+      "loss": 0.503,
+      "step": 17153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903308540651449e-05,
+      "loss": 0.52,
+      "step": 17154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9032444216465764e-05,
+      "loss": 0.5003,
+      "step": 17155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903180302641703e-05,
+      "loss": 0.5684,
+      "step": 17156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90311618363683e-05,
+      "loss": 0.4535,
+      "step": 17157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9030520646319566e-05,
+      "loss": 0.4871,
+      "step": 17158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902987945627084e-05,
+      "loss": 0.4754,
+      "step": 17159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902923826622211e-05,
+      "loss": 0.5213,
+      "step": 17160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902859707617338e-05,
+      "loss": 0.4965,
+      "step": 17161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902795588612465e-05,
+      "loss": 0.5092,
+      "step": 17162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902731469607592e-05,
+      "loss": 0.4412,
+      "step": 17163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902667350602719e-05,
+      "loss": 0.5046,
+      "step": 17164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9026032315978454e-05,
+      "loss": 0.4918,
+      "step": 17165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9025391125929726e-05,
+      "loss": 0.5161,
+      "step": 17166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902474993588099e-05,
+      "loss": 0.5195,
+      "step": 17167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902410874583227e-05,
+      "loss": 0.5892,
+      "step": 17168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902346755578354e-05,
+      "loss": 0.5553,
+      "step": 17169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9022826365734806e-05,
+      "loss": 0.443,
+      "step": 17170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902218517568608e-05,
+      "loss": 0.5915,
+      "step": 17171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902154398563734e-05,
+      "loss": 0.5834,
+      "step": 17172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9020902795588615e-05,
+      "loss": 0.4884,
+      "step": 17173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902026160553988e-05,
+      "loss": 0.5664,
+      "step": 17174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901962041549115e-05,
+      "loss": 0.6784,
+      "step": 17175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901897922544242e-05,
+      "loss": 0.6061,
+      "step": 17176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9018338035393695e-05,
+      "loss": 0.5889,
+      "step": 17177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9017696845344966e-05,
+      "loss": 0.4907,
+      "step": 17178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901705565529623e-05,
+      "loss": 0.4922,
+      "step": 17179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90164144652475e-05,
+      "loss": 0.4501,
+      "step": 17180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901577327519877e-05,
+      "loss": 0.5656,
+      "step": 17181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901513208515004e-05,
+      "loss": 0.5387,
+      "step": 17182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9014490895101305e-05,
+      "loss": 0.5251,
+      "step": 17183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901384970505258e-05,
+      "loss": 0.5163,
+      "step": 17184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9013208515003855e-05,
+      "loss": 0.5438,
+      "step": 17185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901256732495512e-05,
+      "loss": 0.5553,
+      "step": 17186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901192613490639e-05,
+      "loss": 0.4399,
+      "step": 17187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9011284944857656e-05,
+      "loss": 0.4926,
+      "step": 17188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901064375480893e-05,
+      "loss": 0.5093,
+      "step": 17189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901000256476019e-05,
+      "loss": 0.5477,
+      "step": 17190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9009361374711465e-05,
+      "loss": 0.5201,
+      "step": 17191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9008720184662736e-05,
+      "loss": 0.4921,
+      "step": 17192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900807899461401e-05,
+      "loss": 0.5376,
+      "step": 17193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900743780456528e-05,
+      "loss": 0.5749,
+      "step": 17194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9006796614516545e-05,
+      "loss": 0.3744,
+      "step": 17195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9006155424467817e-05,
+      "loss": 0.544,
+      "step": 17196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900551423441908e-05,
+      "loss": 0.4356,
+      "step": 17197
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900487304437035e-05,
+      "loss": 0.3903,
+      "step": 17198
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900423185432162e-05,
+      "loss": 0.4466,
+      "step": 17199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900359066427289e-05,
+      "loss": 0.4825,
+      "step": 17200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900294947422416e-05,
+      "loss": 0.4269,
+      "step": 17201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900230828417543e-05,
+      "loss": 0.5175,
+      "step": 17202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9001667094126705e-05,
+      "loss": 0.6214,
+      "step": 17203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900102590407797e-05,
+      "loss": 0.4581,
+      "step": 17204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900038471402924e-05,
+      "loss": 0.4701,
+      "step": 17205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899974352398051e-05,
+      "loss": 0.5621,
+      "step": 17206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899910233393178e-05,
+      "loss": 0.5722,
+      "step": 17207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899846114388305e-05,
+      "loss": 0.5957,
+      "step": 17208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899781995383432e-05,
+      "loss": 0.5853,
+      "step": 17209
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899717876378559e-05,
+      "loss": 0.3955,
+      "step": 17210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899653757373686e-05,
+      "loss": 0.5461,
+      "step": 17211
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899589638368813e-05,
+      "loss": 0.5205,
+      "step": 17212
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8995255193639395e-05,
+      "loss": 0.4543,
+      "step": 17213
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899461400359067e-05,
+      "loss": 0.5194,
+      "step": 17214
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899397281354193e-05,
+      "loss": 0.5307,
+      "step": 17215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8993331623493203e-05,
+      "loss": 0.5093,
+      "step": 17216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8992690433444475e-05,
+      "loss": 0.5481,
+      "step": 17217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899204924339575e-05,
+      "loss": 0.5563,
+      "step": 17218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899140805334701e-05,
+      "loss": 0.4247,
+      "step": 17219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8990766863298284e-05,
+      "loss": 0.5183,
+      "step": 17220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8990125673249555e-05,
+      "loss": 0.6224,
+      "step": 17221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898948448320082e-05,
+      "loss": 0.4936,
+      "step": 17222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898884329315209e-05,
+      "loss": 0.5212,
+      "step": 17223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8988202103103364e-05,
+      "loss": 0.5148,
+      "step": 17224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8987560913054635e-05,
+      "loss": 0.5391,
+      "step": 17225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.89869197230059e-05,
+      "loss": 0.5058,
+      "step": 17226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898627853295717e-05,
+      "loss": 0.4876,
+      "step": 17227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898563734290844e-05,
+      "loss": 0.5164,
+      "step": 17228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898499615285971e-05,
+      "loss": 0.5519,
+      "step": 17229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898435496281098e-05,
+      "loss": 0.4112,
+      "step": 17230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8983713772762245e-05,
+      "loss": 0.4356,
+      "step": 17231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898307258271352e-05,
+      "loss": 0.4912,
+      "step": 17232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898243139266479e-05,
+      "loss": 0.5055,
+      "step": 17233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898179020261606e-05,
+      "loss": 0.5308,
+      "step": 17234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8981149012567325e-05,
+      "loss": 0.5637,
+      "step": 17235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.89805078225186e-05,
+      "loss": 0.5697,
+      "step": 17236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897986663246986e-05,
+      "loss": 0.5803,
+      "step": 17237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8979225442421134e-05,
+      "loss": 0.5311,
+      "step": 17238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8978584252372406e-05,
+      "loss": 0.6067,
+      "step": 17239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897794306232368e-05,
+      "loss": 0.5164,
+      "step": 17240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897730187227495e-05,
+      "loss": 0.4965,
+      "step": 17241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8976660682226214e-05,
+      "loss": 0.4547,
+      "step": 17242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8976019492177486e-05,
+      "loss": 0.4845,
+      "step": 17243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897537830212875e-05,
+      "loss": 0.4929,
+      "step": 17244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897473711208002e-05,
+      "loss": 0.5916,
+      "step": 17245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897409592203129e-05,
+      "loss": 0.5459,
+      "step": 17246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897345473198256e-05,
+      "loss": 0.5229,
+      "step": 17247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897281354193383e-05,
+      "loss": 0.5499,
+      "step": 17248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.89721723518851e-05,
+      "loss": 0.6194,
+      "step": 17249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8971531161836374e-05,
+      "loss": 0.5527,
+      "step": 17250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897088997178764e-05,
+      "loss": 0.5045,
+      "step": 17251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897024878173891e-05,
+      "loss": 0.518,
+      "step": 17252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8969607591690176e-05,
+      "loss": 0.4945,
+      "step": 17253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896896640164145e-05,
+      "loss": 0.6265,
+      "step": 17254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896832521159271e-05,
+      "loss": 0.5952,
+      "step": 17255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896768402154399e-05,
+      "loss": 0.64,
+      "step": 17256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896704283149526e-05,
+      "loss": 0.5563,
+      "step": 17257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896640164144653e-05,
+      "loss": 0.5035,
+      "step": 17258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.89657604513978e-05,
+      "loss": 0.5829,
+      "step": 17259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8965119261349064e-05,
+      "loss": 0.4479,
+      "step": 17260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8964478071300336e-05,
+      "loss": 0.5024,
+      "step": 17261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.89638368812516e-05,
+      "loss": 0.4987,
+      "step": 17262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896319569120287e-05,
+      "loss": 0.5888,
+      "step": 17263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8962554501154144e-05,
+      "loss": 0.6661,
+      "step": 17264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8961913311105416e-05,
+      "loss": 0.6267,
+      "step": 17265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896127212105669e-05,
+      "loss": 0.4373,
+      "step": 17266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896063093100795e-05,
+      "loss": 0.5598,
+      "step": 17267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8959989740959224e-05,
+      "loss": 0.4398,
+      "step": 17268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895934855091049e-05,
+      "loss": 0.6598,
+      "step": 17269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895870736086176e-05,
+      "loss": 0.4336,
+      "step": 17270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8958066170813026e-05,
+      "loss": 0.5355,
+      "step": 17271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8957424980764304e-05,
+      "loss": 0.4827,
+      "step": 17272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895678379071557e-05,
+      "loss": 0.4543,
+      "step": 17273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895614260066684e-05,
+      "loss": 0.5972,
+      "step": 17274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895550141061811e-05,
+      "loss": 0.5495,
+      "step": 17275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895486022056938e-05,
+      "loss": 0.5486,
+      "step": 17276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895421903052065e-05,
+      "loss": 0.582,
+      "step": 17277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8953577840471914e-05,
+      "loss": 0.6376,
+      "step": 17278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8952936650423186e-05,
+      "loss": 0.4534,
+      "step": 17279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895229546037446e-05,
+      "loss": 0.5017,
+      "step": 17280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.895165427032573e-05,
+      "loss": 0.5571,
+      "step": 17281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8951013080276994e-05,
+      "loss": 0.5336,
+      "step": 17282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8950371890228266e-05,
+      "loss": 0.5235,
+      "step": 17283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894973070017954e-05,
+      "loss": 0.4323,
+      "step": 17284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.89490895101308e-05,
+      "loss": 0.5054,
+      "step": 17285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8948448320082075e-05,
+      "loss": 0.5807,
+      "step": 17286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894780713003334e-05,
+      "loss": 0.5373,
+      "step": 17287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894716593998461e-05,
+      "loss": 0.4932,
+      "step": 17288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894652474993588e-05,
+      "loss": 0.5088,
+      "step": 17289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8945883559887155e-05,
+      "loss": 0.5242,
+      "step": 17290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894524236983842e-05,
+      "loss": 0.5968,
+      "step": 17291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894460117978969e-05,
+      "loss": 0.6465,
+      "step": 17292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894395998974096e-05,
+      "loss": 0.4988,
+      "step": 17293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894331879969223e-05,
+      "loss": 0.5998,
+      "step": 17294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.89426776096435e-05,
+      "loss": 0.5374,
+      "step": 17295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894203641959477e-05,
+      "loss": 0.4351,
+      "step": 17296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894139522954604e-05,
+      "loss": 0.532,
+      "step": 17297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894075403949731e-05,
+      "loss": 0.6338,
+      "step": 17298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894011284944858e-05,
+      "loss": 0.4529,
+      "step": 17299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8939471659399845e-05,
+      "loss": 0.5602,
+      "step": 17300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8938830469351116e-05,
+      "loss": 0.5725,
+      "step": 17301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893818927930239e-05,
+      "loss": 0.5576,
+      "step": 17302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893754808925365e-05,
+      "loss": 0.5759,
+      "step": 17303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8936906899204925e-05,
+      "loss": 0.4889,
+      "step": 17304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8936265709156197e-05,
+      "loss": 0.6052,
+      "step": 17305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893562451910747e-05,
+      "loss": 0.5996,
+      "step": 17306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893498332905873e-05,
+      "loss": 0.6268,
+      "step": 17307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8934342139010005e-05,
+      "loss": 0.5218,
+      "step": 17308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893370094896127e-05,
+      "loss": 0.5244,
+      "step": 17309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893305975891254e-05,
+      "loss": 0.4725,
+      "step": 17310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893241856886381e-05,
+      "loss": 0.4496,
+      "step": 17311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8931777378815085e-05,
+      "loss": 0.596,
+      "step": 17312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893113618876636e-05,
+      "loss": 0.4461,
+      "step": 17313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893049499871762e-05,
+      "loss": 0.5595,
+      "step": 17314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892985380866889e-05,
+      "loss": 0.5392,
+      "step": 17315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892921261862016e-05,
+      "loss": 0.4643,
+      "step": 17316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892857142857143e-05,
+      "loss": 0.5986,
+      "step": 17317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8927930238522695e-05,
+      "loss": 0.4795,
+      "step": 17318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892728904847397e-05,
+      "loss": 0.5105,
+      "step": 17319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892664785842524e-05,
+      "loss": 0.4905,
+      "step": 17320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892600666837651e-05,
+      "loss": 0.5628,
+      "step": 17321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892536547832778e-05,
+      "loss": 0.4473,
+      "step": 17322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892472428827905e-05,
+      "loss": 0.4978,
+      "step": 17323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892408309823032e-05,
+      "loss": 0.5783,
+      "step": 17324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8923441908181583e-05,
+      "loss": 0.5056,
+      "step": 17325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8922800718132855e-05,
+      "loss": 0.4889,
+      "step": 17326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892215952808412e-05,
+      "loss": 0.4627,
+      "step": 17327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89215183380354e-05,
+      "loss": 0.4821,
+      "step": 17328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892087714798667e-05,
+      "loss": 0.5096,
+      "step": 17329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8920235957937935e-05,
+      "loss": 0.5524,
+      "step": 17330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891959476788921e-05,
+      "loss": 0.5029,
+      "step": 17331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891895357784047e-05,
+      "loss": 0.5511,
+      "step": 17332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8918312387791744e-05,
+      "loss": 0.4476,
+      "step": 17333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891767119774301e-05,
+      "loss": 0.5733,
+      "step": 17334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891703000769428e-05,
+      "loss": 0.5775,
+      "step": 17335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891638881764555e-05,
+      "loss": 0.4891,
+      "step": 17336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8915747627596824e-05,
+      "loss": 0.4757,
+      "step": 17337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8915106437548095e-05,
+      "loss": 0.4565,
+      "step": 17338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891446524749936e-05,
+      "loss": 0.5359,
+      "step": 17339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891382405745063e-05,
+      "loss": 0.5449,
+      "step": 17340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89131828674019e-05,
+      "loss": 0.6006,
+      "step": 17341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891254167735317e-05,
+      "loss": 0.4249,
+      "step": 17342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8911900487304434e-05,
+      "loss": 0.6339,
+      "step": 17343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891125929725571e-05,
+      "loss": 0.5066,
+      "step": 17344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891061810720698e-05,
+      "loss": 0.5179,
+      "step": 17345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890997691715825e-05,
+      "loss": 0.5266,
+      "step": 17346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890933572710952e-05,
+      "loss": 0.6307,
+      "step": 17347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8908694537060785e-05,
+      "loss": 0.6583,
+      "step": 17348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890805334701206e-05,
+      "loss": 0.5502,
+      "step": 17349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890741215696332e-05,
+      "loss": 0.5745,
+      "step": 17350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8906770966914594e-05,
+      "loss": 0.5381,
+      "step": 17351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8906129776865866e-05,
+      "loss": 0.4056,
+      "step": 17352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890548858681714e-05,
+      "loss": 0.4357,
+      "step": 17353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89048473967684e-05,
+      "loss": 0.51,
+      "step": 17354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8904206206719674e-05,
+      "loss": 0.6044,
+      "step": 17355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8903565016670946e-05,
+      "loss": 0.5561,
+      "step": 17356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890292382662221e-05,
+      "loss": 0.3362,
+      "step": 17357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890228263657348e-05,
+      "loss": 0.456,
+      "step": 17358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890164144652475e-05,
+      "loss": 0.4977,
+      "step": 17359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890100025647602e-05,
+      "loss": 0.5167,
+      "step": 17360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890035906642729e-05,
+      "loss": 0.5744,
+      "step": 17361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889971787637856e-05,
+      "loss": 0.4778,
+      "step": 17362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889907668632983e-05,
+      "loss": 0.4934,
+      "step": 17363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88984354962811e-05,
+      "loss": 0.518,
+      "step": 17364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889779430623237e-05,
+      "loss": 0.5048,
+      "step": 17365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8897153116183636e-05,
+      "loss": 0.4883,
+      "step": 17366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889651192613491e-05,
+      "loss": 0.4482,
+      "step": 17367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889587073608618e-05,
+      "loss": 0.5561,
+      "step": 17368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889522954603745e-05,
+      "loss": 0.6091,
+      "step": 17369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8894588355988716e-05,
+      "loss": 0.5076,
+      "step": 17370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889394716593999e-05,
+      "loss": 0.5205,
+      "step": 17371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889330597589126e-05,
+      "loss": 0.5108,
+      "step": 17372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8892664785842524e-05,
+      "loss": 0.4846,
+      "step": 17373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8892023595793796e-05,
+      "loss": 0.4661,
+      "step": 17374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889138240574506e-05,
+      "loss": 0.5194,
+      "step": 17375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889074121569633e-05,
+      "loss": 0.3926,
+      "step": 17376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8890100025647604e-05,
+      "loss": 0.4978,
+      "step": 17377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8889458835598876e-05,
+      "loss": 0.5795,
+      "step": 17378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888881764555014e-05,
+      "loss": 0.5617,
+      "step": 17379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888817645550141e-05,
+      "loss": 0.6181,
+      "step": 17380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8887535265452684e-05,
+      "loss": 0.4495,
+      "step": 17381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888689407540395e-05,
+      "loss": 0.4625,
+      "step": 17382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888625288535522e-05,
+      "loss": 0.5252,
+      "step": 17383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888561169530649e-05,
+      "loss": 0.538,
+      "step": 17384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8884970505257764e-05,
+      "loss": 0.5724,
+      "step": 17385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888432931520903e-05,
+      "loss": 0.4672,
+      "step": 17386
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88836881251603e-05,
+      "loss": 0.5618,
+      "step": 17387
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8883046935111566e-05,
+      "loss": 0.5129,
+      "step": 17388
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888240574506284e-05,
+      "loss": 0.5442,
+      "step": 17389
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888176455501411e-05,
+      "loss": 0.5417,
+      "step": 17390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8881123364965374e-05,
+      "loss": 0.5397,
+      "step": 17391
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8880482174916646e-05,
+      "loss": 0.5092,
+      "step": 17392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887984098486792e-05,
+      "loss": 0.4748,
+      "step": 17393
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887919979481919e-05,
+      "loss": 0.4943,
+      "step": 17394
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8878558604770454e-05,
+      "loss": 0.3912,
+      "step": 17395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8877917414721726e-05,
+      "loss": 0.5435,
+      "step": 17396
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887727622467299e-05,
+      "loss": 0.5125,
+      "step": 17397
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887663503462426e-05,
+      "loss": 0.5795,
+      "step": 17398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8875993844575535e-05,
+      "loss": 0.5246,
+      "step": 17399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8875352654526806e-05,
+      "loss": 0.5276,
+      "step": 17400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887471146447808e-05,
+      "loss": 0.4073,
+      "step": 17401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887407027442934e-05,
+      "loss": 0.5642,
+      "step": 17402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8873429084380615e-05,
+      "loss": 0.5281,
+      "step": 17403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887278789433188e-05,
+      "loss": 0.5193,
+      "step": 17404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887214670428315e-05,
+      "loss": 0.5707,
+      "step": 17405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8871505514234416e-05,
+      "loss": 0.6435,
+      "step": 17406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887086432418569e-05,
+      "loss": 0.5496,
+      "step": 17407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887022313413696e-05,
+      "loss": 0.4666,
+      "step": 17408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886958194408823e-05,
+      "loss": 0.4901,
+      "step": 17409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88689407540395e-05,
+      "loss": 0.4237,
+      "step": 17410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886829956399077e-05,
+      "loss": 0.6249,
+      "step": 17411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886765837394204e-05,
+      "loss": 0.5226,
+      "step": 17412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8867017183893305e-05,
+      "loss": 0.5109,
+      "step": 17413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8866375993844576e-05,
+      "loss": 0.5213,
+      "step": 17414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886573480379584e-05,
+      "loss": 0.4638,
+      "step": 17415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886509361374712e-05,
+      "loss": 0.5927,
+      "step": 17416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886445242369839e-05,
+      "loss": 0.4967,
+      "step": 17417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8863811233649657e-05,
+      "loss": 0.5128,
+      "step": 17418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886317004360093e-05,
+      "loss": 0.5964,
+      "step": 17419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886252885355219e-05,
+      "loss": 0.5066,
+      "step": 17420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8861887663503465e-05,
+      "loss": 0.4993,
+      "step": 17421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886124647345473e-05,
+      "loss": 0.5141,
+      "step": 17422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8860605283406e-05,
+      "loss": 0.5095,
+      "step": 17423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885996409335727e-05,
+      "loss": 0.4787,
+      "step": 17424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8859322903308545e-05,
+      "loss": 0.4767,
+      "step": 17425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885868171325982e-05,
+      "loss": 0.6252,
+      "step": 17426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885804052321108e-05,
+      "loss": 0.455,
+      "step": 17427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885739933316235e-05,
+      "loss": 0.4777,
+      "step": 17428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885675814311362e-05,
+      "loss": 0.4842,
+      "step": 17429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885611695306489e-05,
+      "loss": 0.4196,
+      "step": 17430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8855475763016155e-05,
+      "loss": 0.4797,
+      "step": 17431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8854834572967433e-05,
+      "loss": 0.5769,
+      "step": 17432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88541933829187e-05,
+      "loss": 0.5777,
+      "step": 17433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885355219286997e-05,
+      "loss": 0.4329,
+      "step": 17434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885291100282124e-05,
+      "loss": 0.4725,
+      "step": 17435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885226981277251e-05,
+      "loss": 0.4648,
+      "step": 17436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885162862272378e-05,
+      "loss": 0.4931,
+      "step": 17437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8850987432675043e-05,
+      "loss": 0.5187,
+      "step": 17438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8850346242626315e-05,
+      "loss": 0.5626,
+      "step": 17439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884970505257759e-05,
+      "loss": 0.5124,
+      "step": 17440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884906386252886e-05,
+      "loss": 0.5521,
+      "step": 17441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8848422672480124e-05,
+      "loss": 0.7139,
+      "step": 17442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8847781482431395e-05,
+      "loss": 0.4839,
+      "step": 17443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884714029238267e-05,
+      "loss": 0.5563,
+      "step": 17444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884649910233393e-05,
+      "loss": 0.4584,
+      "step": 17445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8845857912285204e-05,
+      "loss": 0.4941,
+      "step": 17446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884521672223647e-05,
+      "loss": 0.4492,
+      "step": 17447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884457553218774e-05,
+      "loss": 0.4644,
+      "step": 17448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884393434213901e-05,
+      "loss": 0.5449,
+      "step": 17449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8843293152090284e-05,
+      "loss": 0.4575,
+      "step": 17450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884265196204155e-05,
+      "loss": 0.5072,
+      "step": 17451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884201077199282e-05,
+      "loss": 0.5049,
+      "step": 17452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884136958194409e-05,
+      "loss": 0.6279,
+      "step": 17453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884072839189536e-05,
+      "loss": 0.4286,
+      "step": 17454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884008720184663e-05,
+      "loss": 0.6698,
+      "step": 17455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88394460117979e-05,
+      "loss": 0.4574,
+      "step": 17456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883880482174917e-05,
+      "loss": 0.5107,
+      "step": 17457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883816363170044e-05,
+      "loss": 0.5415,
+      "step": 17458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883752244165171e-05,
+      "loss": 0.5198,
+      "step": 17459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8836881251602974e-05,
+      "loss": 0.529,
+      "step": 17460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8836240061554245e-05,
+      "loss": 0.4389,
+      "step": 17461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883559887150552e-05,
+      "loss": 0.5093,
+      "step": 17462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883495768145678e-05,
+      "loss": 0.6256,
+      "step": 17463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8834316491408054e-05,
+      "loss": 0.541,
+      "step": 17464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8833675301359326e-05,
+      "loss": 0.5333,
+      "step": 17465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88330341113106e-05,
+      "loss": 0.4642,
+      "step": 17466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883239292126186e-05,
+      "loss": 0.5802,
+      "step": 17467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8831751731213134e-05,
+      "loss": 0.6078,
+      "step": 17468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88311105411644e-05,
+      "loss": 0.602,
+      "step": 17469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883046935111567e-05,
+      "loss": 0.5547,
+      "step": 17470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882982816106694e-05,
+      "loss": 0.4929,
+      "step": 17471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8829186971018214e-05,
+      "loss": 0.56,
+      "step": 17472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8828545780969486e-05,
+      "loss": 0.5524,
+      "step": 17473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882790459092075e-05,
+      "loss": 0.4215,
+      "step": 17474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882726340087202e-05,
+      "loss": 0.5048,
+      "step": 17475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882662221082329e-05,
+      "loss": 0.5057,
+      "step": 17476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882598102077456e-05,
+      "loss": 0.5547,
+      "step": 17477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8825339830725824e-05,
+      "loss": 0.7576,
+      "step": 17478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8824698640677096e-05,
+      "loss": 0.6253,
+      "step": 17479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882405745062837e-05,
+      "loss": 0.5474,
+      "step": 17480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882341626057964e-05,
+      "loss": 0.5152,
+      "step": 17481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882277507053091e-05,
+      "loss": 0.5308,
+      "step": 17482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8822133880482176e-05,
+      "loss": 0.5947,
+      "step": 17483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882149269043345e-05,
+      "loss": 0.5842,
+      "step": 17484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882085150038471e-05,
+      "loss": 0.4007,
+      "step": 17485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8820210310335984e-05,
+      "loss": 0.5963,
+      "step": 17486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881956912028725e-05,
+      "loss": 0.4655,
+      "step": 17487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881892793023853e-05,
+      "loss": 0.5785,
+      "step": 17488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88182867401898e-05,
+      "loss": 0.5003,
+      "step": 17489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8817645550141064e-05,
+      "loss": 0.6,
+      "step": 17490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8817004360092336e-05,
+      "loss": 0.5933,
+      "step": 17491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88163631700436e-05,
+      "loss": 0.4853,
+      "step": 17492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881572197999487e-05,
+      "loss": 0.59,
+      "step": 17493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881508078994614e-05,
+      "loss": 0.4743,
+      "step": 17494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881443959989741e-05,
+      "loss": 0.4195,
+      "step": 17495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881379840984868e-05,
+      "loss": 0.6154,
+      "step": 17496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881315721979995e-05,
+      "loss": 0.5362,
+      "step": 17497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8812516029751224e-05,
+      "loss": 0.3976,
+      "step": 17498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881187483970249e-05,
+      "loss": 0.4007,
+      "step": 17499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881123364965376e-05,
+      "loss": 0.5911,
+      "step": 17500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8810592459605026e-05,
+      "loss": 0.4579,
+      "step": 17501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88099512695563e-05,
+      "loss": 0.6188,
+      "step": 17502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880931007950756e-05,
+      "loss": 0.4969,
+      "step": 17503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880866888945884e-05,
+      "loss": 0.616,
+      "step": 17504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8808027699410106e-05,
+      "loss": 0.4425,
+      "step": 17505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880738650936138e-05,
+      "loss": 0.4699,
+      "step": 17506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880674531931265e-05,
+      "loss": 0.552,
+      "step": 17507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8806104129263915e-05,
+      "loss": 0.4745,
+      "step": 17508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8805462939215186e-05,
+      "loss": 0.5721,
+      "step": 17509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880482174916645e-05,
+      "loss": 0.5772,
+      "step": 17510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880418055911772e-05,
+      "loss": 0.5276,
+      "step": 17511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8803539369068995e-05,
+      "loss": 0.6028,
+      "step": 17512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8802898179020266e-05,
+      "loss": 0.4853,
+      "step": 17513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880225698897153e-05,
+      "loss": 0.4928,
+      "step": 17514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88016157989228e-05,
+      "loss": 0.4882,
+      "step": 17515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8800974608874075e-05,
+      "loss": 0.5607,
+      "step": 17516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880033341882534e-05,
+      "loss": 0.4936,
+      "step": 17517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879969222877661e-05,
+      "loss": 0.4847,
+      "step": 17518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8799051038727876e-05,
+      "loss": 0.5469,
+      "step": 17519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879840984867915e-05,
+      "loss": 0.5348,
+      "step": 17520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879776865863042e-05,
+      "loss": 0.4615,
+      "step": 17521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879712746858169e-05,
+      "loss": 0.5361,
+      "step": 17522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8796486278532956e-05,
+      "loss": 0.6262,
+      "step": 17523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879584508848423e-05,
+      "loss": 0.5354,
+      "step": 17524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.87952038984355e-05,
+      "loss": 0.5558,
+      "step": 17525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8794562708386765e-05,
+      "loss": 0.4793,
+      "step": 17526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8793921518338036e-05,
+      "loss": 0.4744,
+      "step": 17527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879328032828931e-05,
+      "loss": 0.5499,
+      "step": 17528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879263913824058e-05,
+      "loss": 0.5137,
+      "step": 17529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8791997948191845e-05,
+      "loss": 0.5454,
+      "step": 17530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8791356758143117e-05,
+      "loss": 0.4753,
+      "step": 17531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879071556809438e-05,
+      "loss": 0.4552,
+      "step": 17532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879007437804565e-05,
+      "loss": 0.5314,
+      "step": 17533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8789433187996925e-05,
+      "loss": 0.5479,
+      "step": 17534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878879199794819e-05,
+      "loss": 0.5622,
+      "step": 17535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878815080789946e-05,
+      "loss": 0.5187,
+      "step": 17536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878750961785073e-05,
+      "loss": 0.503,
+      "step": 17537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8786868427802005e-05,
+      "loss": 0.5512,
+      "step": 17538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878622723775327e-05,
+      "loss": 0.5858,
+      "step": 17539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878558604770454e-05,
+      "loss": 0.5491,
+      "step": 17540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878494485765581e-05,
+      "loss": 0.4918,
+      "step": 17541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878430366760708e-05,
+      "loss": 0.5649,
+      "step": 17542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878366247755835e-05,
+      "loss": 0.5166,
+      "step": 17543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878302128750962e-05,
+      "loss": 0.5053,
+      "step": 17544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8782380097460894e-05,
+      "loss": 0.519,
+      "step": 17545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878173890741216e-05,
+      "loss": 0.5165,
+      "step": 17546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878109771736343e-05,
+      "loss": 0.608,
+      "step": 17547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8780456527314695e-05,
+      "loss": 0.4931,
+      "step": 17548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.877981533726597e-05,
+      "loss": 0.5574,
+      "step": 17549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.877917414721723e-05,
+      "loss": 0.4914,
+      "step": 17550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8778532957168503e-05,
+      "loss": 0.4648,
+      "step": 17551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8777891767119775e-05,
+      "loss": 0.5238,
+      "step": 17552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.877725057707105e-05,
+      "loss": 0.5732,
+      "step": 17553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.877660938702232e-05,
+      "loss": 0.5226,
+      "step": 17554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8775968196973584e-05,
+      "loss": 0.5251,
+      "step": 17555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8775327006924855e-05,
+      "loss": 0.5083,
+      "step": 17556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.877468581687612e-05,
+      "loss": 0.6725,
+      "step": 17557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.877404462682739e-05,
+      "loss": 0.4824,
+      "step": 17558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8773403436778664e-05,
+      "loss": 0.3867,
+      "step": 17559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8772762246729935e-05,
+      "loss": 0.5171,
+      "step": 17560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.877212105668121e-05,
+      "loss": 0.4367,
+      "step": 17561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.877147986663247e-05,
+      "loss": 0.5174,
+      "step": 17562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8770838676583744e-05,
+      "loss": 0.4298,
+      "step": 17563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.877019748653501e-05,
+      "loss": 0.5322,
+      "step": 17564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876955629648628e-05,
+      "loss": 0.5923,
+      "step": 17565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8768915106437545e-05,
+      "loss": 0.5718,
+      "step": 17566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876827391638882e-05,
+      "loss": 0.454,
+      "step": 17567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876763272634009e-05,
+      "loss": 0.5682,
+      "step": 17568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876699153629136e-05,
+      "loss": 0.5,
+      "step": 17569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876635034624263e-05,
+      "loss": 0.4802,
+      "step": 17570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87657091561939e-05,
+      "loss": 0.4019,
+      "step": 17571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876506796614517e-05,
+      "loss": 0.5224,
+      "step": 17572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8764426776096434e-05,
+      "loss": 0.492,
+      "step": 17573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8763785586047706e-05,
+      "loss": 0.513,
+      "step": 17574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876314439599897e-05,
+      "loss": 0.6744,
+      "step": 17575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876250320595025e-05,
+      "loss": 0.4662,
+      "step": 17576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876186201590152e-05,
+      "loss": 0.4397,
+      "step": 17577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8761220825852786e-05,
+      "loss": 0.4956,
+      "step": 17578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876057963580406e-05,
+      "loss": 0.5064,
+      "step": 17579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875993844575532e-05,
+      "loss": 0.5543,
+      "step": 17580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8759297255706594e-05,
+      "loss": 0.6216,
+      "step": 17581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875865606565786e-05,
+      "loss": 0.3919,
+      "step": 17582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875801487560913e-05,
+      "loss": 0.5615,
+      "step": 17583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87573736855604e-05,
+      "loss": 0.5069,
+      "step": 17584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8756732495511674e-05,
+      "loss": 0.5305,
+      "step": 17585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8756091305462946e-05,
+      "loss": 0.5305,
+      "step": 17586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875545011541421e-05,
+      "loss": 0.5137,
+      "step": 17587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875480892536548e-05,
+      "loss": 0.5545,
+      "step": 17588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875416773531675e-05,
+      "loss": 0.6449,
+      "step": 17589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875352654526802e-05,
+      "loss": 0.4969,
+      "step": 17590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8752885355219284e-05,
+      "loss": 0.4913,
+      "step": 17591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875224416517056e-05,
+      "loss": 0.52,
+      "step": 17592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875160297512183e-05,
+      "loss": 0.5358,
+      "step": 17593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87509617850731e-05,
+      "loss": 0.5175,
+      "step": 17594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875032059502437e-05,
+      "loss": 0.5562,
+      "step": 17595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8749679404975636e-05,
+      "loss": 0.5135,
+      "step": 17596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874903821492691e-05,
+      "loss": 0.5559,
+      "step": 17597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874839702487817e-05,
+      "loss": 0.4286,
+      "step": 17598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8747755834829444e-05,
+      "loss": 0.4581,
+      "step": 17599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8747114644780716e-05,
+      "loss": 0.6123,
+      "step": 17600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874647345473199e-05,
+      "loss": 0.578,
+      "step": 17601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874583226468325e-05,
+      "loss": 0.4813,
+      "step": 17602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8745191074634524e-05,
+      "loss": 0.5405,
+      "step": 17603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8744549884585796e-05,
+      "loss": 0.576,
+      "step": 17604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874390869453706e-05,
+      "loss": 0.4921,
+      "step": 17605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874326750448833e-05,
+      "loss": 0.4349,
+      "step": 17606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87426263144396e-05,
+      "loss": 0.4534,
+      "step": 17607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874198512439087e-05,
+      "loss": 0.5553,
+      "step": 17608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874134393434214e-05,
+      "loss": 0.5307,
+      "step": 17609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874070274429341e-05,
+      "loss": 0.4846,
+      "step": 17610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874006155424468e-05,
+      "loss": 0.4509,
+      "step": 17611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873942036419595e-05,
+      "loss": 0.6752,
+      "step": 17612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873877917414722e-05,
+      "loss": 0.5303,
+      "step": 17613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8738137984098486e-05,
+      "loss": 0.574,
+      "step": 17614
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873749679404976e-05,
+      "loss": 0.561,
+      "step": 17615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873685560400103e-05,
+      "loss": 0.5257,
+      "step": 17616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87362144139523e-05,
+      "loss": 0.6176,
+      "step": 17617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8735573223903566e-05,
+      "loss": 0.5496,
+      "step": 17618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873493203385484e-05,
+      "loss": 0.5531,
+      "step": 17619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87342908438061e-05,
+      "loss": 0.5814,
+      "step": 17620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8733649653757375e-05,
+      "loss": 0.4985,
+      "step": 17621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8733008463708646e-05,
+      "loss": 0.4617,
+      "step": 17622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873236727365991e-05,
+      "loss": 0.4725,
+      "step": 17623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873172608361118e-05,
+      "loss": 0.5386,
+      "step": 17624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8731084893562455e-05,
+      "loss": 0.505,
+      "step": 17625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8730443703513726e-05,
+      "loss": 0.4853,
+      "step": 17626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872980251346499e-05,
+      "loss": 0.5388,
+      "step": 17627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872916132341626e-05,
+      "loss": 0.4534,
+      "step": 17628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872852013336753e-05,
+      "loss": 0.5097,
+      "step": 17629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87278789433188e-05,
+      "loss": 0.4407,
+      "step": 17630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872723775327007e-05,
+      "loss": 0.5213,
+      "step": 17631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872659656322134e-05,
+      "loss": 0.6288,
+      "step": 17632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8725955373172615e-05,
+      "loss": 0.5175,
+      "step": 17633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872531418312388e-05,
+      "loss": 0.5054,
+      "step": 17634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872467299307515e-05,
+      "loss": 0.5794,
+      "step": 17635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8724031803026416e-05,
+      "loss": 0.5467,
+      "step": 17636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872339061297769e-05,
+      "loss": 0.5115,
+      "step": 17637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872274942292895e-05,
+      "loss": 0.5412,
+      "step": 17638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8722108232880225e-05,
+      "loss": 0.5372,
+      "step": 17639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8721467042831497e-05,
+      "loss": 0.5877,
+      "step": 17640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872082585278277e-05,
+      "loss": 0.4971,
+      "step": 17641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872018466273404e-05,
+      "loss": 0.4964,
+      "step": 17642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8719543472685305e-05,
+      "loss": 0.4655,
+      "step": 17643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8718902282636577e-05,
+      "loss": 0.525,
+      "step": 17644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871826109258784e-05,
+      "loss": 0.4914,
+      "step": 17645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871761990253911e-05,
+      "loss": 0.5183,
+      "step": 17646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871697871249038e-05,
+      "loss": 0.4262,
+      "step": 17647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871633752244166e-05,
+      "loss": 0.5258,
+      "step": 17648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871569633239293e-05,
+      "loss": 0.4482,
+      "step": 17649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871505514234419e-05,
+      "loss": 0.6693,
+      "step": 17650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8714413952295465e-05,
+      "loss": 0.4618,
+      "step": 17651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871377276224673e-05,
+      "loss": 0.657,
+      "step": 17652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8713131572198e-05,
+      "loss": 0.611,
+      "step": 17653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871249038214927e-05,
+      "loss": 0.5465,
+      "step": 17654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871184919210054e-05,
+      "loss": 0.5278,
+      "step": 17655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871120800205181e-05,
+      "loss": 0.5409,
+      "step": 17656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871056681200308e-05,
+      "loss": 0.4974,
+      "step": 17657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8709925621954354e-05,
+      "loss": 0.4549,
+      "step": 17658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870928443190562e-05,
+      "loss": 0.547,
+      "step": 17659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870864324185689e-05,
+      "loss": 0.6778,
+      "step": 17660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8708002051808155e-05,
+      "loss": 0.5656,
+      "step": 17661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870736086175943e-05,
+      "loss": 0.5713,
+      "step": 17662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870671967171069e-05,
+      "loss": 0.4946,
+      "step": 17663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870607848166197e-05,
+      "loss": 0.4124,
+      "step": 17664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8705437291613235e-05,
+      "loss": 0.5469,
+      "step": 17665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870479610156451e-05,
+      "loss": 0.4558,
+      "step": 17666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870415491151578e-05,
+      "loss": 0.5039,
+      "step": 17667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8703513721467044e-05,
+      "loss": 0.5179,
+      "step": 17668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8702872531418315e-05,
+      "loss": 0.5119,
+      "step": 17669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870223134136958e-05,
+      "loss": 0.5106,
+      "step": 17670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870159015132085e-05,
+      "loss": 0.4308,
+      "step": 17671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8700948961272124e-05,
+      "loss": 0.5487,
+      "step": 17672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8700307771223395e-05,
+      "loss": 0.4718,
+      "step": 17673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869966658117466e-05,
+      "loss": 0.5389,
+      "step": 17674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869902539112593e-05,
+      "loss": 0.5221,
+      "step": 17675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8698384201077204e-05,
+      "loss": 0.5794,
+      "step": 17676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869774301102847e-05,
+      "loss": 0.4672,
+      "step": 17677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869710182097974e-05,
+      "loss": 0.5005,
+      "step": 17678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8696460630931005e-05,
+      "loss": 0.4773,
+      "step": 17679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8695819440882284e-05,
+      "loss": 0.5112,
+      "step": 17680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869517825083355e-05,
+      "loss": 0.5545,
+      "step": 17681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869453706078482e-05,
+      "loss": 0.4032,
+      "step": 17682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8693895870736085e-05,
+      "loss": 0.6359,
+      "step": 17683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869325468068736e-05,
+      "loss": 0.4453,
+      "step": 17684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869261349063863e-05,
+      "loss": 0.4722,
+      "step": 17685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8691972300589894e-05,
+      "loss": 0.5345,
+      "step": 17686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8691331110541166e-05,
+      "loss": 0.3978,
+      "step": 17687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869068992049244e-05,
+      "loss": 0.5376,
+      "step": 17688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869004873044371e-05,
+      "loss": 0.4837,
+      "step": 17689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8689407540394974e-05,
+      "loss": 0.646,
+      "step": 17690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8688766350346246e-05,
+      "loss": 0.4994,
+      "step": 17691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868812516029751e-05,
+      "loss": 0.3896,
+      "step": 17692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868748397024878e-05,
+      "loss": 0.4809,
+      "step": 17693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8686842780200054e-05,
+      "loss": 0.5338,
+      "step": 17694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868620159015132e-05,
+      "loss": 0.5017,
+      "step": 17695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868556040010259e-05,
+      "loss": 0.551,
+      "step": 17696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868491921005386e-05,
+      "loss": 0.5178,
+      "step": 17697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8684278020005134e-05,
+      "loss": 0.5038,
+      "step": 17698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86836368299564e-05,
+      "loss": 0.5028,
+      "step": 17699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868299563990767e-05,
+      "loss": 0.4999,
+      "step": 17700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8682354449858936e-05,
+      "loss": 0.4161,
+      "step": 17701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868171325981021e-05,
+      "loss": 0.6871,
+      "step": 17702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868107206976148e-05,
+      "loss": 0.5045,
+      "step": 17703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868043087971275e-05,
+      "loss": 0.5625,
+      "step": 17704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867978968966402e-05,
+      "loss": 0.6448,
+      "step": 17705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867914849961529e-05,
+      "loss": 0.4598,
+      "step": 17706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867850730956656e-05,
+      "loss": 0.4935,
+      "step": 17707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8677866119517824e-05,
+      "loss": 0.5082,
+      "step": 17708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8677224929469096e-05,
+      "loss": 0.5247,
+      "step": 17709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867658373942036e-05,
+      "loss": 0.5008,
+      "step": 17710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867594254937163e-05,
+      "loss": 0.4362,
+      "step": 17711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8675301359322904e-05,
+      "loss": 0.5287,
+      "step": 17712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8674660169274176e-05,
+      "loss": 0.4785,
+      "step": 17713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867401897922545e-05,
+      "loss": 0.3952,
+      "step": 17714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867337778917671e-05,
+      "loss": 0.5395,
+      "step": 17715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8672736599127984e-05,
+      "loss": 0.5186,
+      "step": 17716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867209540907925e-05,
+      "loss": 0.4948,
+      "step": 17717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867145421903052e-05,
+      "loss": 0.5151,
+      "step": 17718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867081302898179e-05,
+      "loss": 0.5778,
+      "step": 17719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8670171838933064e-05,
+      "loss": 0.4573,
+      "step": 17720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8669530648884336e-05,
+      "loss": 0.5531,
+      "step": 17721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86688894588356e-05,
+      "loss": 0.5332,
+      "step": 17722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866824826878687e-05,
+      "loss": 0.5704,
+      "step": 17723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866760707873814e-05,
+      "loss": 0.4928,
+      "step": 17724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866696588868941e-05,
+      "loss": 0.5469,
+      "step": 17725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8666324698640674e-05,
+      "loss": 0.4733,
+      "step": 17726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8665683508591946e-05,
+      "loss": 0.6298,
+      "step": 17727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866504231854322e-05,
+      "loss": 0.434,
+      "step": 17728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866440112849449e-05,
+      "loss": 0.5057,
+      "step": 17729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866375993844576e-05,
+      "loss": 0.5805,
+      "step": 17730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8663118748397026e-05,
+      "loss": 0.4919,
+      "step": 17731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86624775583483e-05,
+      "loss": 0.4382,
+      "step": 17732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866183636829956e-05,
+      "loss": 0.501,
+      "step": 17733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8661195178250835e-05,
+      "loss": 0.4842,
+      "step": 17734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86605539882021e-05,
+      "loss": 0.5514,
+      "step": 17735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865991279815338e-05,
+      "loss": 0.5851,
+      "step": 17736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865927160810464e-05,
+      "loss": 0.4263,
+      "step": 17737
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8658630418055915e-05,
+      "loss": 0.4557,
+      "step": 17738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8657989228007186e-05,
+      "loss": 0.4711,
+      "step": 17739
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865734803795845e-05,
+      "loss": 0.4972,
+      "step": 17740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865670684790972e-05,
+      "loss": 0.5297,
+      "step": 17741
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865606565786099e-05,
+      "loss": 0.4383,
+      "step": 17742
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865542446781226e-05,
+      "loss": 0.5051,
+      "step": 17743
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865478327776353e-05,
+      "loss": 0.5553,
+      "step": 17744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86541420877148e-05,
+      "loss": 0.4929,
+      "step": 17745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865350089766607e-05,
+      "loss": 0.4429,
+      "step": 17746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865285970761734e-05,
+      "loss": 0.6154,
+      "step": 17747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865221851756861e-05,
+      "loss": 0.6278,
+      "step": 17748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8651577327519876e-05,
+      "loss": 0.4961,
+      "step": 17749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865093613747115e-05,
+      "loss": 0.428,
+      "step": 17750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865029494742241e-05,
+      "loss": 0.4883,
+      "step": 17751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864965375737369e-05,
+      "loss": 0.402,
+      "step": 17752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8649012567324957e-05,
+      "loss": 0.4966,
+      "step": 17753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864837137727623e-05,
+      "loss": 0.5992,
+      "step": 17754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86477301872275e-05,
+      "loss": 0.4707,
+      "step": 17755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8647088997178765e-05,
+      "loss": 0.4748,
+      "step": 17756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864644780713004e-05,
+      "loss": 0.5826,
+      "step": 17757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86458066170813e-05,
+      "loss": 0.4931,
+      "step": 17758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864516542703257e-05,
+      "loss": 0.4605,
+      "step": 17759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8644524236983845e-05,
+      "loss": 0.5938,
+      "step": 17760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864388304693512e-05,
+      "loss": 0.4779,
+      "step": 17761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864324185688638e-05,
+      "loss": 0.4361,
+      "step": 17762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864260066683765e-05,
+      "loss": 0.5421,
+      "step": 17763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8641959476788925e-05,
+      "loss": 0.6218,
+      "step": 17764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864131828674019e-05,
+      "loss": 0.5539,
+      "step": 17765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864067709669146e-05,
+      "loss": 0.5299,
+      "step": 17766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864003590664273e-05,
+      "loss": 0.524,
+      "step": 17767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8639394716594e-05,
+      "loss": 0.5426,
+      "step": 17768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863875352654527e-05,
+      "loss": 0.4712,
+      "step": 17769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863811233649654e-05,
+      "loss": 0.5693,
+      "step": 17770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863747114644781e-05,
+      "loss": 0.4454,
+      "step": 17771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863682995639908e-05,
+      "loss": 0.5225,
+      "step": 17772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863618876635035e-05,
+      "loss": 0.643,
+      "step": 17773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8635547576301615e-05,
+      "loss": 0.4344,
+      "step": 17774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863490638625289e-05,
+      "loss": 0.3754,
+      "step": 17775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863426519620416e-05,
+      "loss": 0.4686,
+      "step": 17776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863362400615543e-05,
+      "loss": 0.6292,
+      "step": 17777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8632982816106695e-05,
+      "loss": 0.5431,
+      "step": 17778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863234162605797e-05,
+      "loss": 0.5107,
+      "step": 17779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863170043600923e-05,
+      "loss": 0.5317,
+      "step": 17780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8631059245960504e-05,
+      "loss": 0.4578,
+      "step": 17781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8630418055911775e-05,
+      "loss": 0.5744,
+      "step": 17782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862977686586304e-05,
+      "loss": 0.5657,
+      "step": 17783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862913567581431e-05,
+      "loss": 0.5778,
+      "step": 17784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8628494485765584e-05,
+      "loss": 0.4955,
+      "step": 17785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8627853295716855e-05,
+      "loss": 0.5887,
+      "step": 17786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862721210566812e-05,
+      "loss": 0.5261,
+      "step": 17787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862657091561939e-05,
+      "loss": 0.5422,
+      "step": 17788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862592972557066e-05,
+      "loss": 0.5062,
+      "step": 17789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862528853552193e-05,
+      "loss": 0.4869,
+      "step": 17790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86246473454732e-05,
+      "loss": 0.4708,
+      "step": 17791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862400615542447e-05,
+      "loss": 0.5042,
+      "step": 17792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8623364965375744e-05,
+      "loss": 0.5323,
+      "step": 17793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862272377532701e-05,
+      "loss": 0.483,
+      "step": 17794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862208258527828e-05,
+      "loss": 0.526,
+      "step": 17795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8621441395229545e-05,
+      "loss": 0.6196,
+      "step": 17796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862080020518082e-05,
+      "loss": 0.5498,
+      "step": 17797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862015901513208e-05,
+      "loss": 0.4817,
+      "step": 17798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8619517825083354e-05,
+      "loss": 0.5155,
+      "step": 17799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8618876635034626e-05,
+      "loss": 0.5407,
+      "step": 17800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86182354449859e-05,
+      "loss": 0.6439,
+      "step": 17801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861759425493717e-05,
+      "loss": 0.5804,
+      "step": 17802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8616953064888434e-05,
+      "loss": 0.5178,
+      "step": 17803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8616311874839706e-05,
+      "loss": 0.4879,
+      "step": 17804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861567068479097e-05,
+      "loss": 0.5845,
+      "step": 17805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861502949474224e-05,
+      "loss": 0.709,
+      "step": 17806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861438830469351e-05,
+      "loss": 0.5312,
+      "step": 17807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8613747114644786e-05,
+      "loss": 0.4834,
+      "step": 17808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861310592459606e-05,
+      "loss": 0.4824,
+      "step": 17809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861246473454732e-05,
+      "loss": 0.4769,
+      "step": 17810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8611823544498594e-05,
+      "loss": 0.4869,
+      "step": 17811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861118235444986e-05,
+      "loss": 0.5015,
+      "step": 17812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861054116440113e-05,
+      "loss": 0.5448,
+      "step": 17813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8609899974352396e-05,
+      "loss": 0.5232,
+      "step": 17814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.860925878430367e-05,
+      "loss": 0.4864,
+      "step": 17815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.860861759425494e-05,
+      "loss": 0.5626,
+      "step": 17816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860797640420621e-05,
+      "loss": 0.6056,
+      "step": 17817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860733521415748e-05,
+      "loss": 0.5453,
+      "step": 17818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860669402410875e-05,
+      "loss": 0.5076,
+      "step": 17819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860605283406002e-05,
+      "loss": 0.4733,
+      "step": 17820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8605411644011284e-05,
+      "loss": 0.4441,
+      "step": 17821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8604770453962556e-05,
+      "loss": 0.5818,
+      "step": 17822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860412926391382e-05,
+      "loss": 0.6361,
+      "step": 17823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.86034880738651e-05,
+      "loss": 0.4869,
+      "step": 17824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8602846883816364e-05,
+      "loss": 0.4368,
+      "step": 17825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8602205693767636e-05,
+      "loss": 0.4816,
+      "step": 17826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860156450371891e-05,
+      "loss": 0.5564,
+      "step": 17827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860092331367017e-05,
+      "loss": 0.4347,
+      "step": 17828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8600282123621444e-05,
+      "loss": 0.5093,
+      "step": 17829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859964093357271e-05,
+      "loss": 0.4884,
+      "step": 17830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859899974352398e-05,
+      "loss": 0.4675,
+      "step": 17831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859835855347525e-05,
+      "loss": 0.4794,
+      "step": 17832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8597717363426524e-05,
+      "loss": 0.59,
+      "step": 17833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859707617337779e-05,
+      "loss": 0.434,
+      "step": 17834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859643498332906e-05,
+      "loss": 0.4998,
+      "step": 17835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859579379328033e-05,
+      "loss": 0.6201,
+      "step": 17836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85951526032316e-05,
+      "loss": 0.5433,
+      "step": 17837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859451141318287e-05,
+      "loss": 0.5677,
+      "step": 17838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8593870223134134e-05,
+      "loss": 0.5549,
+      "step": 17839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859322903308541e-05,
+      "loss": 0.5243,
+      "step": 17840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859258784303668e-05,
+      "loss": 0.5023,
+      "step": 17841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859194665298795e-05,
+      "loss": 0.5646,
+      "step": 17842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8591305462939215e-05,
+      "loss": 0.5877,
+      "step": 17843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8590664272890486e-05,
+      "loss": 0.5499,
+      "step": 17844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859002308284176e-05,
+      "loss": 0.5016,
+      "step": 17845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858938189279302e-05,
+      "loss": 0.4575,
+      "step": 17846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8588740702744295e-05,
+      "loss": 0.4777,
+      "step": 17847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8588099512695566e-05,
+      "loss": 0.5477,
+      "step": 17848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858745832264684e-05,
+      "loss": 0.6133,
+      "step": 17849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85868171325981e-05,
+      "loss": 0.4741,
+      "step": 17850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8586175942549375e-05,
+      "loss": 0.6232,
+      "step": 17851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858553475250064e-05,
+      "loss": 0.467,
+      "step": 17852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858489356245191e-05,
+      "loss": 0.5169,
+      "step": 17853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858425237240318e-05,
+      "loss": 0.4875,
+      "step": 17854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858361118235445e-05,
+      "loss": 0.5234,
+      "step": 17855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858296999230572e-05,
+      "loss": 0.5438,
+      "step": 17856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858232880225699e-05,
+      "loss": 0.5478,
+      "step": 17857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858168761220826e-05,
+      "loss": 0.4731,
+      "step": 17858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858104642215953e-05,
+      "loss": 0.6193,
+      "step": 17859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85804052321108e-05,
+      "loss": 0.5821,
+      "step": 17860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8579764042062065e-05,
+      "loss": 0.5059,
+      "step": 17861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8579122852013336e-05,
+      "loss": 0.4773,
+      "step": 17862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857848166196461e-05,
+      "loss": 0.6205,
+      "step": 17863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857784047191588e-05,
+      "loss": 0.5618,
+      "step": 17864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857719928186715e-05,
+      "loss": 0.5095,
+      "step": 17865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8576558091818417e-05,
+      "loss": 0.4812,
+      "step": 17866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857591690176969e-05,
+      "loss": 0.4486,
+      "step": 17867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857527571172095e-05,
+      "loss": 0.5483,
+      "step": 17868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8574634521672225e-05,
+      "loss": 0.4506,
+      "step": 17869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857399333162349e-05,
+      "loss": 0.5584,
+      "step": 17870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857335214157476e-05,
+      "loss": 0.4722,
+      "step": 17871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857271095152603e-05,
+      "loss": 0.4956,
+      "step": 17872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8572069761477305e-05,
+      "loss": 0.5101,
+      "step": 17873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857142857142858e-05,
+      "loss": 0.7272,
+      "step": 17874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857078738137984e-05,
+      "loss": 0.4403,
+      "step": 17875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8570146191331113e-05,
+      "loss": 0.4004,
+      "step": 17876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856950500128238e-05,
+      "loss": 0.5971,
+      "step": 17877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856886381123365e-05,
+      "loss": 0.588,
+      "step": 17878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856822262118492e-05,
+      "loss": 0.554,
+      "step": 17879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8567581431136194e-05,
+      "loss": 0.6195,
+      "step": 17880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8566940241087465e-05,
+      "loss": 0.5997,
+      "step": 17881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856629905103873e-05,
+      "loss": 0.4613,
+      "step": 17882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856565786099e-05,
+      "loss": 0.5721,
+      "step": 17883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856501667094127e-05,
+      "loss": 0.5352,
+      "step": 17884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856437548089254e-05,
+      "loss": 0.548,
+      "step": 17885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8563734290843803e-05,
+      "loss": 0.4952,
+      "step": 17886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8563093100795075e-05,
+      "loss": 0.4391,
+      "step": 17887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856245191074635e-05,
+      "loss": 0.5611,
+      "step": 17888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856181072069762e-05,
+      "loss": 0.4867,
+      "step": 17889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856116953064889e-05,
+      "loss": 0.5014,
+      "step": 17890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8560528340600155e-05,
+      "loss": 0.5635,
+      "step": 17891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855988715055143e-05,
+      "loss": 0.4987,
+      "step": 17892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855924596050269e-05,
+      "loss": 0.5274,
+      "step": 17893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8558604770453964e-05,
+      "loss": 0.4773,
+      "step": 17894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855796358040523e-05,
+      "loss": 0.4365,
+      "step": 17895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855732239035651e-05,
+      "loss": 0.5628,
+      "step": 17896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855668120030777e-05,
+      "loss": 0.5128,
+      "step": 17897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8556040010259044e-05,
+      "loss": 0.5154,
+      "step": 17898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8555398820210315e-05,
+      "loss": 0.565,
+      "step": 17899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855475763016158e-05,
+      "loss": 0.5223,
+      "step": 17900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855411644011285e-05,
+      "loss": 0.5484,
+      "step": 17901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855347525006412e-05,
+      "loss": 0.4027,
+      "step": 17902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855283406001539e-05,
+      "loss": 0.5727,
+      "step": 17903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855219286996666e-05,
+      "loss": 0.529,
+      "step": 17904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855155167991793e-05,
+      "loss": 0.4691,
+      "step": 17905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85509104898692e-05,
+      "loss": 0.5968,
+      "step": 17906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855026929982047e-05,
+      "loss": 0.5255,
+      "step": 17907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854962810977174e-05,
+      "loss": 0.5262,
+      "step": 17908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8548986919723006e-05,
+      "loss": 0.5765,
+      "step": 17909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854834572967428e-05,
+      "loss": 0.4376,
+      "step": 17910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854770453962554e-05,
+      "loss": 0.4826,
+      "step": 17911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854706334957682e-05,
+      "loss": 0.4857,
+      "step": 17912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8546422159528086e-05,
+      "loss": 0.5158,
+      "step": 17913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854578096947936e-05,
+      "loss": 0.5129,
+      "step": 17914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854513977943062e-05,
+      "loss": 0.4864,
+      "step": 17915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8544498589381894e-05,
+      "loss": 0.4939,
+      "step": 17916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8543857399333166e-05,
+      "loss": 0.5653,
+      "step": 17917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854321620928443e-05,
+      "loss": 0.5535,
+      "step": 17918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85425750192357e-05,
+      "loss": 0.5844,
+      "step": 17919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8541933829186974e-05,
+      "loss": 0.4153,
+      "step": 17920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8541292639138246e-05,
+      "loss": 0.4489,
+      "step": 17921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854065144908951e-05,
+      "loss": 0.491,
+      "step": 17922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854001025904078e-05,
+      "loss": 0.617,
+      "step": 17923
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853936906899205e-05,
+      "loss": 0.5303,
+      "step": 17924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853872787894332e-05,
+      "loss": 0.4803,
+      "step": 17925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853808668889459e-05,
+      "loss": 0.5474,
+      "step": 17926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8537445498845856e-05,
+      "loss": 0.5289,
+      "step": 17927
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8536804308797134e-05,
+      "loss": 0.5382,
+      "step": 17928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85361631187484e-05,
+      "loss": 0.5339,
+      "step": 17929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853552192869967e-05,
+      "loss": 0.5817,
+      "step": 17930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8534880738650936e-05,
+      "loss": 0.4542,
+      "step": 17931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853423954860221e-05,
+      "loss": 0.4837,
+      "step": 17932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853359835855347e-05,
+      "loss": 0.5612,
+      "step": 17933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8532957168504744e-05,
+      "loss": 0.4894,
+      "step": 17934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8532315978456016e-05,
+      "loss": 0.5057,
+      "step": 17935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853167478840729e-05,
+      "loss": 0.4229,
+      "step": 17936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853103359835856e-05,
+      "loss": 0.4886,
+      "step": 17937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8530392408309824e-05,
+      "loss": 0.5622,
+      "step": 17938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8529751218261096e-05,
+      "loss": 0.5267,
+      "step": 17939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852911002821236e-05,
+      "loss": 0.4501,
+      "step": 17940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852846883816363e-05,
+      "loss": 0.5278,
+      "step": 17941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8527827648114904e-05,
+      "loss": 0.4776,
+      "step": 17942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852718645806617e-05,
+      "loss": 0.4742,
+      "step": 17943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852654526801744e-05,
+      "loss": 0.6058,
+      "step": 17944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852590407796871e-05,
+      "loss": 0.6028,
+      "step": 17945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8525262887919985e-05,
+      "loss": 0.5425,
+      "step": 17946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852462169787125e-05,
+      "loss": 0.5782,
+      "step": 17947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852398050782252e-05,
+      "loss": 0.5446,
+      "step": 17948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8523339317773786e-05,
+      "loss": 0.6825,
+      "step": 17949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852269812772506e-05,
+      "loss": 0.5257,
+      "step": 17950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852205693767633e-05,
+      "loss": 0.5785,
+      "step": 17951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85214157476276e-05,
+      "loss": 0.6923,
+      "step": 17952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852077455757887e-05,
+      "loss": 0.4765,
+      "step": 17953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852013336753014e-05,
+      "loss": 0.4802,
+      "step": 17954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851949217748141e-05,
+      "loss": 0.4504,
+      "step": 17955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8518850987432675e-05,
+      "loss": 0.5816,
+      "step": 17956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8518209797383946e-05,
+      "loss": 0.5837,
+      "step": 17957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851756860733521e-05,
+      "loss": 0.5183,
+      "step": 17958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851692741728648e-05,
+      "loss": 0.5083,
+      "step": 17959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8516286227237755e-05,
+      "loss": 0.4666,
+      "step": 17960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8515645037189026e-05,
+      "loss": 0.4768,
+      "step": 17961
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85150038471403e-05,
+      "loss": 0.4972,
+      "step": 17962
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851436265709156e-05,
+      "loss": 0.5123,
+      "step": 17963
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8513721467042835e-05,
+      "loss": 0.6266,
+      "step": 17964
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85130802769941e-05,
+      "loss": 0.5558,
+      "step": 17965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851243908694537e-05,
+      "loss": 0.5098,
+      "step": 17966
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851179789689664e-05,
+      "loss": 0.6863,
+      "step": 17967
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8511156706847915e-05,
+      "loss": 0.5367,
+      "step": 17968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8510515516799187e-05,
+      "loss": 0.4949,
+      "step": 17969
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850987432675045e-05,
+      "loss": 0.6615,
+      "step": 17970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850923313670172e-05,
+      "loss": 0.503,
+      "step": 17971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850859194665299e-05,
+      "loss": 0.4763,
+      "step": 17972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850795075660426e-05,
+      "loss": 0.5753,
+      "step": 17973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8507309566555525e-05,
+      "loss": 0.4041,
+      "step": 17974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8506668376506797e-05,
+      "loss": 0.7223,
+      "step": 17975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850602718645807e-05,
+      "loss": 0.4732,
+      "step": 17976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850538599640934e-05,
+      "loss": 0.4693,
+      "step": 17977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850474480636061e-05,
+      "loss": 0.4504,
+      "step": 17978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850410361631188e-05,
+      "loss": 0.5025,
+      "step": 17979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850346242626315e-05,
+      "loss": 0.4572,
+      "step": 17980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850282123621441e-05,
+      "loss": 0.5208,
+      "step": 17981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8502180046165685e-05,
+      "loss": 0.5557,
+      "step": 17982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850153885611695e-05,
+      "loss": 0.4105,
+      "step": 17983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850089766606823e-05,
+      "loss": 0.6045,
+      "step": 17984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850025647601949e-05,
+      "loss": 0.4881,
+      "step": 17985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8499615285970765e-05,
+      "loss": 0.4523,
+      "step": 17986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849897409592204e-05,
+      "loss": 0.5265,
+      "step": 17987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.84983329058733e-05,
+      "loss": 0.4339,
+      "step": 17988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8497691715824573e-05,
+      "loss": 0.504,
+      "step": 17989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849705052577584e-05,
+      "loss": 0.658,
+      "step": 17990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849640933572711e-05,
+      "loss": 0.6479,
+      "step": 17991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849576814567838e-05,
+      "loss": 0.6413,
+      "step": 17992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8495126955629654e-05,
+      "loss": 0.4762,
+      "step": 17993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849448576558092e-05,
+      "loss": 0.5855,
+      "step": 17994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849384457553219e-05,
+      "loss": 0.5709,
+      "step": 17995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849320338548346e-05,
+      "loss": 0.5612,
+      "step": 17996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849256219543473e-05,
+      "loss": 0.4877,
+      "step": 17997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8491921005386e-05,
+      "loss": 0.5991,
+      "step": 17998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8491279815337264e-05,
+      "loss": 0.6326,
+      "step": 17999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849063862528854e-05,
+      "loss": 0.5035,
+      "step": 18000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848999743523981e-05,
+      "loss": 0.49,
+      "step": 18001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848935624519108e-05,
+      "loss": 0.5199,
+      "step": 18002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8488715055142344e-05,
+      "loss": 0.5529,
+      "step": 18003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8488073865093615e-05,
+      "loss": 0.595,
+      "step": 18004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848743267504489e-05,
+      "loss": 0.5346,
+      "step": 18005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848679148499615e-05,
+      "loss": 0.5486,
+      "step": 18006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8486150294947424e-05,
+      "loss": 0.5111,
+      "step": 18007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8485509104898695e-05,
+      "loss": 0.5681,
+      "step": 18008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848486791484997e-05,
+      "loss": 0.4939,
+      "step": 18009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848422672480123e-05,
+      "loss": 0.5408,
+      "step": 18010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8483585534752504e-05,
+      "loss": 0.5706,
+      "step": 18011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848294434470377e-05,
+      "loss": 0.5934,
+      "step": 18012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848230315465504e-05,
+      "loss": 0.5044,
+      "step": 18013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848166196460631e-05,
+      "loss": 0.5638,
+      "step": 18014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848102077455758e-05,
+      "loss": 0.6312,
+      "step": 18015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.848037958450885e-05,
+      "loss": 0.5376,
+      "step": 18016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847973839446012e-05,
+      "loss": 0.504,
+      "step": 18017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847909720441139e-05,
+      "loss": 0.465,
+      "step": 18018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847845601436266e-05,
+      "loss": 0.5737,
+      "step": 18019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847781482431393e-05,
+      "loss": 0.5183,
+      "step": 18020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8477173634265194e-05,
+      "loss": 0.5659,
+      "step": 18021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8476532444216466e-05,
+      "loss": 0.5812,
+      "step": 18022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847589125416774e-05,
+      "loss": 0.5,
+      "step": 18023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847525006411901e-05,
+      "loss": 0.5135,
+      "step": 18024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847460887407028e-05,
+      "loss": 0.5222,
+      "step": 18025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8473967684021546e-05,
+      "loss": 0.6212,
+      "step": 18026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847332649397282e-05,
+      "loss": 0.4904,
+      "step": 18027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847268530392408e-05,
+      "loss": 0.5362,
+      "step": 18028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8472044113875354e-05,
+      "loss": 0.5122,
+      "step": 18029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847140292382662e-05,
+      "loss": 0.5472,
+      "step": 18030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847076173377789e-05,
+      "loss": 0.4993,
+      "step": 18031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847012054372916e-05,
+      "loss": 0.5523,
+      "step": 18032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8469479353680434e-05,
+      "loss": 0.4627,
+      "step": 18033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8468838163631706e-05,
+      "loss": 0.4534,
+      "step": 18034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846819697358297e-05,
+      "loss": 0.5625,
+      "step": 18035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846755578353424e-05,
+      "loss": 0.4692,
+      "step": 18036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846691459348551e-05,
+      "loss": 0.4618,
+      "step": 18037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846627340343678e-05,
+      "loss": 0.5821,
+      "step": 18038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846563221338805e-05,
+      "loss": 0.5486,
+      "step": 18039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846499102333932e-05,
+      "loss": 0.5563,
+      "step": 18040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8464349833290594e-05,
+      "loss": 0.6188,
+      "step": 18041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846370864324186e-05,
+      "loss": 0.5871,
+      "step": 18042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846306745319313e-05,
+      "loss": 0.4628,
+      "step": 18043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8462426263144396e-05,
+      "loss": 0.4827,
+      "step": 18044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846178507309567e-05,
+      "loss": 0.5313,
+      "step": 18045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846114388304693e-05,
+      "loss": 0.6048,
+      "step": 18046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8460502692998204e-05,
+      "loss": 0.4778,
+      "step": 18047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8459861502949476e-05,
+      "loss": 0.673,
+      "step": 18048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845922031290075e-05,
+      "loss": 0.6361,
+      "step": 18049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845857912285202e-05,
+      "loss": 0.5268,
+      "step": 18050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8457937932803284e-05,
+      "loss": 0.5674,
+      "step": 18051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8457296742754556e-05,
+      "loss": 0.6396,
+      "step": 18052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845665555270582e-05,
+      "loss": 0.548,
+      "step": 18053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845601436265709e-05,
+      "loss": 0.4666,
+      "step": 18054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845537317260836e-05,
+      "loss": 0.539,
+      "step": 18055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8454731982559636e-05,
+      "loss": 0.5308,
+      "step": 18056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.84540907925109e-05,
+      "loss": 0.4303,
+      "step": 18057
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845344960246217e-05,
+      "loss": 0.5561,
+      "step": 18058
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8452808412413445e-05,
+      "loss": 0.5342,
+      "step": 18059
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845216722236471e-05,
+      "loss": 0.4978,
+      "step": 18060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845152603231598e-05,
+      "loss": 0.4923,
+      "step": 18061
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8450884842267246e-05,
+      "loss": 0.529,
+      "step": 18062
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.845024365221852e-05,
+      "loss": 0.5361,
+      "step": 18063
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.844960246216979e-05,
+      "loss": 0.6384,
+      "step": 18064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.844896127212106e-05,
+      "loss": 0.4705,
+      "step": 18065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8448320082072326e-05,
+      "loss": 0.5372,
+      "step": 18066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.84476788920236e-05,
+      "loss": 0.4996,
+      "step": 18067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.844703770197487e-05,
+      "loss": 0.447,
+      "step": 18068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8446396511926135e-05,
+      "loss": 0.527,
+      "step": 18069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8445755321877406e-05,
+      "loss": 0.5648,
+      "step": 18070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.844511413182867e-05,
+      "loss": 0.552,
+      "step": 18071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.844447294177995e-05,
+      "loss": 0.6014,
+      "step": 18072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8443831751731215e-05,
+      "loss": 0.4386,
+      "step": 18073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8443190561682486e-05,
+      "loss": 0.4692,
+      "step": 18074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.844254937163375e-05,
+      "loss": 0.4614,
+      "step": 18075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.844190818158502e-05,
+      "loss": 0.5178,
+      "step": 18076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8441266991536295e-05,
+      "loss": 0.4953,
+      "step": 18077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.844062580148756e-05,
+      "loss": 0.5354,
+      "step": 18078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843998461143883e-05,
+      "loss": 0.5241,
+      "step": 18079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84393434213901e-05,
+      "loss": 0.5081,
+      "step": 18080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8438702231341375e-05,
+      "loss": 0.4343,
+      "step": 18081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843806104129264e-05,
+      "loss": 0.4352,
+      "step": 18082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843741985124391e-05,
+      "loss": 0.5791,
+      "step": 18083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8436778661195176e-05,
+      "loss": 0.5243,
+      "step": 18084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843613747114645e-05,
+      "loss": 0.5563,
+      "step": 18085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843549628109772e-05,
+      "loss": 0.4651,
+      "step": 18086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8434855091048985e-05,
+      "loss": 0.4707,
+      "step": 18087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843421390100026e-05,
+      "loss": 0.5829,
+      "step": 18088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843357271095153e-05,
+      "loss": 0.5453,
+      "step": 18089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84329315209028e-05,
+      "loss": 0.4641,
+      "step": 18090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8432290330854065e-05,
+      "loss": 0.5222,
+      "step": 18091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843164914080534e-05,
+      "loss": 0.4181,
+      "step": 18092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84310079507566e-05,
+      "loss": 0.7471,
+      "step": 18093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843036676070787e-05,
+      "loss": 0.6384,
+      "step": 18094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8429725570659145e-05,
+      "loss": 0.5329,
+      "step": 18095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842908438061042e-05,
+      "loss": 0.5134,
+      "step": 18096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842844319056169e-05,
+      "loss": 0.4941,
+      "step": 18097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8427802000512953e-05,
+      "loss": 0.537,
+      "step": 18098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8427160810464225e-05,
+      "loss": 0.5505,
+      "step": 18099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842651962041549e-05,
+      "loss": 0.5613,
+      "step": 18100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842587843036676e-05,
+      "loss": 0.4361,
+      "step": 18101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842523724031803e-05,
+      "loss": 0.5237,
+      "step": 18102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84245960502693e-05,
+      "loss": 0.5466,
+      "step": 18103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842395486022057e-05,
+      "loss": 0.4533,
+      "step": 18104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842331367017184e-05,
+      "loss": 0.4374,
+      "step": 18105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8422672480123114e-05,
+      "loss": 0.6049,
+      "step": 18106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842203129007438e-05,
+      "loss": 0.632,
+      "step": 18107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842139010002565e-05,
+      "loss": 0.6205,
+      "step": 18108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8420748909976915e-05,
+      "loss": 0.4489,
+      "step": 18109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842010771992819e-05,
+      "loss": 0.5566,
+      "step": 18110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841946652987946e-05,
+      "loss": 0.6731,
+      "step": 18111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841882533983073e-05,
+      "loss": 0.4367,
+      "step": 18112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8418184149782e-05,
+      "loss": 0.5053,
+      "step": 18113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841754295973327e-05,
+      "loss": 0.434,
+      "step": 18114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841690176968454e-05,
+      "loss": 0.5725,
+      "step": 18115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8416260579635804e-05,
+      "loss": 0.5385,
+      "step": 18116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8415619389587075e-05,
+      "loss": 0.4967,
+      "step": 18117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841497819953834e-05,
+      "loss": 0.5979,
+      "step": 18118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841433700948961e-05,
+      "loss": 0.5081,
+      "step": 18119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8413695819440884e-05,
+      "loss": 0.5006,
+      "step": 18120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8413054629392155e-05,
+      "loss": 0.4967,
+      "step": 18121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841241343934343e-05,
+      "loss": 0.4993,
+      "step": 18122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841177224929469e-05,
+      "loss": 0.5051,
+      "step": 18123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8411131059245964e-05,
+      "loss": 0.5067,
+      "step": 18124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841048986919723e-05,
+      "loss": 0.4561,
+      "step": 18125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84098486791485e-05,
+      "loss": 0.5318,
+      "step": 18126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840920748909977e-05,
+      "loss": 0.505,
+      "step": 18127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8408566299051044e-05,
+      "loss": 0.4773,
+      "step": 18128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8407925109002316e-05,
+      "loss": 0.4851,
+      "step": 18129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840728391895358e-05,
+      "loss": 0.4939,
+      "step": 18130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840664272890485e-05,
+      "loss": 0.4653,
+      "step": 18131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840600153885612e-05,
+      "loss": 0.4795,
+      "step": 18132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840536034880739e-05,
+      "loss": 0.6099,
+      "step": 18133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8404719158758654e-05,
+      "loss": 0.5782,
+      "step": 18134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8404077968709926e-05,
+      "loss": 0.5763,
+      "step": 18135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84034367786612e-05,
+      "loss": 0.5049,
+      "step": 18136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840279558861247e-05,
+      "loss": 0.4548,
+      "step": 18137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840215439856374e-05,
+      "loss": 0.4794,
+      "step": 18138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8401513208515006e-05,
+      "loss": 0.6308,
+      "step": 18139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840087201846628e-05,
+      "loss": 0.5622,
+      "step": 18140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840023082841754e-05,
+      "loss": 0.5378,
+      "step": 18141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8399589638368814e-05,
+      "loss": 0.4275,
+      "step": 18142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839894844832008e-05,
+      "loss": 0.5504,
+      "step": 18143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839830725827136e-05,
+      "loss": 0.4801,
+      "step": 18144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839766606822262e-05,
+      "loss": 0.5512,
+      "step": 18145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8397024878173894e-05,
+      "loss": 0.5026,
+      "step": 18146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8396383688125166e-05,
+      "loss": 0.5249,
+      "step": 18147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839574249807643e-05,
+      "loss": 0.543,
+      "step": 18148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83951013080277e-05,
+      "loss": 0.5782,
+      "step": 18149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839446011797897e-05,
+      "loss": 0.5247,
+      "step": 18150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839381892793024e-05,
+      "loss": 0.4427,
+      "step": 18151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839317773788151e-05,
+      "loss": 0.4796,
+      "step": 18152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839253654783278e-05,
+      "loss": 0.4044,
+      "step": 18153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839189535778405e-05,
+      "loss": 0.4918,
+      "step": 18154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839125416773532e-05,
+      "loss": 0.4436,
+      "step": 18155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839061297768659e-05,
+      "loss": 0.508,
+      "step": 18156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8389971787637856e-05,
+      "loss": 0.4848,
+      "step": 18157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838933059758913e-05,
+      "loss": 0.5639,
+      "step": 18158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838868940754039e-05,
+      "loss": 0.6512,
+      "step": 18159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838804821749167e-05,
+      "loss": 0.5254,
+      "step": 18160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8387407027442936e-05,
+      "loss": 0.5886,
+      "step": 18161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838676583739421e-05,
+      "loss": 0.4849,
+      "step": 18162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838612464734547e-05,
+      "loss": 0.508,
+      "step": 18163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8385483457296744e-05,
+      "loss": 0.6258,
+      "step": 18164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8384842267248016e-05,
+      "loss": 0.6222,
+      "step": 18165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838420107719928e-05,
+      "loss": 0.5565,
+      "step": 18166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838355988715055e-05,
+      "loss": 0.4939,
+      "step": 18167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8382918697101824e-05,
+      "loss": 0.4538,
+      "step": 18168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8382277507053096e-05,
+      "loss": 0.5503,
+      "step": 18169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838163631700436e-05,
+      "loss": 0.5971,
+      "step": 18170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838099512695563e-05,
+      "loss": 0.4997,
+      "step": 18171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83803539369069e-05,
+      "loss": 0.5018,
+      "step": 18172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837971274685817e-05,
+      "loss": 0.4028,
+      "step": 18173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837907155680944e-05,
+      "loss": 0.4439,
+      "step": 18174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8378430366760706e-05,
+      "loss": 0.4788,
+      "step": 18175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837778917671198e-05,
+      "loss": 0.5471,
+      "step": 18176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837714798666325e-05,
+      "loss": 0.5436,
+      "step": 18177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837650679661452e-05,
+      "loss": 0.5424,
+      "step": 18178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8375865606565786e-05,
+      "loss": 0.4922,
+      "step": 18179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837522441651706e-05,
+      "loss": 0.5308,
+      "step": 18180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837458322646832e-05,
+      "loss": 0.5223,
+      "step": 18181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8373942036419595e-05,
+      "loss": 0.579,
+      "step": 18182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8373300846370866e-05,
+      "loss": 0.5366,
+      "step": 18183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837265965632214e-05,
+      "loss": 0.6025,
+      "step": 18184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837201846627341e-05,
+      "loss": 0.4248,
+      "step": 18185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8371377276224675e-05,
+      "loss": 0.5476,
+      "step": 18186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8370736086175946e-05,
+      "loss": 0.5922,
+      "step": 18187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837009489612721e-05,
+      "loss": 0.4829,
+      "step": 18188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836945370607848e-05,
+      "loss": 0.4553,
+      "step": 18189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836881251602975e-05,
+      "loss": 0.5683,
+      "step": 18190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836817132598102e-05,
+      "loss": 0.4849,
+      "step": 18191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836753013593229e-05,
+      "loss": 0.5109,
+      "step": 18192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836688894588356e-05,
+      "loss": 0.5453,
+      "step": 18193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8366247755834835e-05,
+      "loss": 0.4958,
+      "step": 18194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83656065657861e-05,
+      "loss": 0.446,
+      "step": 18195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836496537573737e-05,
+      "loss": 0.4145,
+      "step": 18196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8364324185688636e-05,
+      "loss": 0.466,
+      "step": 18197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836368299563991e-05,
+      "loss": 0.5315,
+      "step": 18198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836304180559118e-05,
+      "loss": 0.482,
+      "step": 18199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836240061554245e-05,
+      "loss": 0.4989,
+      "step": 18200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836175942549372e-05,
+      "loss": 0.5344,
+      "step": 18201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836111823544499e-05,
+      "loss": 0.5884,
+      "step": 18202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836047704539626e-05,
+      "loss": 0.597,
+      "step": 18203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8359835855347525e-05,
+      "loss": 0.4161,
+      "step": 18204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83591946652988e-05,
+      "loss": 0.5728,
+      "step": 18205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835855347525006e-05,
+      "loss": 0.5862,
+      "step": 18206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835791228520133e-05,
+      "loss": 0.444,
+      "step": 18207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8357271095152605e-05,
+      "loss": 0.4037,
+      "step": 18208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835662990510388e-05,
+      "loss": 0.4539,
+      "step": 18209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835598871505515e-05,
+      "loss": 0.5017,
+      "step": 18210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8355347525006413e-05,
+      "loss": 0.4876,
+      "step": 18211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8354706334957685e-05,
+      "loss": 0.5795,
+      "step": 18212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835406514490895e-05,
+      "loss": 0.6621,
+      "step": 18213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835342395486022e-05,
+      "loss": 0.552,
+      "step": 18214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8352782764811494e-05,
+      "loss": 0.5539,
+      "step": 18215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8352141574762765e-05,
+      "loss": 0.4935,
+      "step": 18216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835150038471403e-05,
+      "loss": 0.4313,
+      "step": 18217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83508591946653e-05,
+      "loss": 0.6063,
+      "step": 18218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8350218004616574e-05,
+      "loss": 0.4382,
+      "step": 18219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834957681456784e-05,
+      "loss": 0.5583,
+      "step": 18220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834893562451911e-05,
+      "loss": 0.6546,
+      "step": 18221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8348294434470375e-05,
+      "loss": 0.6326,
+      "step": 18222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834765324442165e-05,
+      "loss": 0.5235,
+      "step": 18223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834701205437292e-05,
+      "loss": 0.5011,
+      "step": 18224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834637086432419e-05,
+      "loss": 0.5246,
+      "step": 18225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8345729674275455e-05,
+      "loss": 0.4745,
+      "step": 18226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834508848422673e-05,
+      "loss": 0.5431,
+      "step": 18227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8344447294178e-05,
+      "loss": 0.5486,
+      "step": 18228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8343806104129264e-05,
+      "loss": 0.5659,
+      "step": 18229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8343164914080535e-05,
+      "loss": 0.5358,
+      "step": 18230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83425237240318e-05,
+      "loss": 0.5642,
+      "step": 18231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834188253398308e-05,
+      "loss": 0.447,
+      "step": 18232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8341241343934344e-05,
+      "loss": 0.4466,
+      "step": 18233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8340600153885615e-05,
+      "loss": 0.5284,
+      "step": 18234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833995896383688e-05,
+      "loss": 0.4991,
+      "step": 18235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833931777378815e-05,
+      "loss": 0.5148,
+      "step": 18236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8338676583739424e-05,
+      "loss": 0.4628,
+      "step": 18237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833803539369069e-05,
+      "loss": 0.5856,
+      "step": 18238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833739420364196e-05,
+      "loss": 0.6658,
+      "step": 18239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833675301359323e-05,
+      "loss": 0.5233,
+      "step": 18240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8336111823544504e-05,
+      "loss": 0.579,
+      "step": 18241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833547063349577e-05,
+      "loss": 0.5983,
+      "step": 18242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833482944344704e-05,
+      "loss": 0.4875,
+      "step": 18243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8334188253398306e-05,
+      "loss": 0.4571,
+      "step": 18244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833354706334958e-05,
+      "loss": 0.5837,
+      "step": 18245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833290587330085e-05,
+      "loss": 0.5709,
+      "step": 18246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8332264683252114e-05,
+      "loss": 0.5185,
+      "step": 18247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833162349320339e-05,
+      "loss": 0.5565,
+      "step": 18248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833098230315466e-05,
+      "loss": 0.4616,
+      "step": 18249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.833034111310593e-05,
+      "loss": 0.5451,
+      "step": 18250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8329699923057194e-05,
+      "loss": 0.422,
+      "step": 18251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8329058733008466e-05,
+      "loss": 0.5963,
+      "step": 18252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832841754295973e-05,
+      "loss": 0.5449,
+      "step": 18253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8327776352911e-05,
+      "loss": 0.5039,
+      "step": 18254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8327135162862274e-05,
+      "loss": 0.5349,
+      "step": 18255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8326493972813546e-05,
+      "loss": 0.5425,
+      "step": 18256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832585278276482e-05,
+      "loss": 0.4125,
+      "step": 18257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832521159271608e-05,
+      "loss": 0.5866,
+      "step": 18258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8324570402667354e-05,
+      "loss": 0.5704,
+      "step": 18259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832392921261862e-05,
+      "loss": 0.5844,
+      "step": 18260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832328802256989e-05,
+      "loss": 0.5202,
+      "step": 18261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8322646832521156e-05,
+      "loss": 0.4733,
+      "step": 18262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832200564247243e-05,
+      "loss": 0.5001,
+      "step": 18263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83213644524237e-05,
+      "loss": 0.5413,
+      "step": 18264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832072326237497e-05,
+      "loss": 0.5339,
+      "step": 18265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832008207232624e-05,
+      "loss": 0.5154,
+      "step": 18266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831944088227751e-05,
+      "loss": 0.5788,
+      "step": 18267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831879969222878e-05,
+      "loss": 0.443,
+      "step": 18268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8318158502180044e-05,
+      "loss": 0.5226,
+      "step": 18269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8317517312131316e-05,
+      "loss": 0.442,
+      "step": 18270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831687612208259e-05,
+      "loss": 0.5391,
+      "step": 18271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831623493203386e-05,
+      "loss": 0.6129,
+      "step": 18272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831559374198513e-05,
+      "loss": 0.4493,
+      "step": 18273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8314952551936396e-05,
+      "loss": 0.503,
+      "step": 18274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831431136188767e-05,
+      "loss": 0.4377,
+      "step": 18275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831367017183893e-05,
+      "loss": 0.4611,
+      "step": 18276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8313028981790204e-05,
+      "loss": 0.4699,
+      "step": 18277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831238779174147e-05,
+      "loss": 0.6517,
+      "step": 18278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831174660169274e-05,
+      "loss": 0.4823,
+      "step": 18279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831110541164401e-05,
+      "loss": 0.4582,
+      "step": 18280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8310464221595285e-05,
+      "loss": 0.5478,
+      "step": 18281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8309823031546556e-05,
+      "loss": 0.4664,
+      "step": 18282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830918184149782e-05,
+      "loss": 0.55,
+      "step": 18283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830854065144909e-05,
+      "loss": 0.491,
+      "step": 18284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830789946140036e-05,
+      "loss": 0.5184,
+      "step": 18285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830725827135163e-05,
+      "loss": 0.5092,
+      "step": 18286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83066170813029e-05,
+      "loss": 0.5167,
+      "step": 18287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830597589125417e-05,
+      "loss": 0.5088,
+      "step": 18288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830533470120544e-05,
+      "loss": 0.4842,
+      "step": 18289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830469351115671e-05,
+      "loss": 0.4904,
+      "step": 18290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830405232110798e-05,
+      "loss": 0.5668,
+      "step": 18291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8303411131059246e-05,
+      "loss": 0.4906,
+      "step": 18292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830276994101052e-05,
+      "loss": 0.4191,
+      "step": 18293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830212875096178e-05,
+      "loss": 0.4904,
+      "step": 18294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8301487560913055e-05,
+      "loss": 0.5467,
+      "step": 18295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8300846370864326e-05,
+      "loss": 0.4821,
+      "step": 18296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83002051808156e-05,
+      "loss": 0.5117,
+      "step": 18297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829956399076686e-05,
+      "loss": 0.6581,
+      "step": 18298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8298922800718135e-05,
+      "loss": 0.5338,
+      "step": 18299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8298281610669406e-05,
+      "loss": 0.6321,
+      "step": 18300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829764042062067e-05,
+      "loss": 0.6853,
+      "step": 18301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829699923057194e-05,
+      "loss": 0.4721,
+      "step": 18302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829635804052321e-05,
+      "loss": 0.4644,
+      "step": 18303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8295716850474487e-05,
+      "loss": 0.5295,
+      "step": 18304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829507566042575e-05,
+      "loss": 0.6147,
+      "step": 18305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829443447037702e-05,
+      "loss": 0.5772,
+      "step": 18306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829379328032829e-05,
+      "loss": 0.4714,
+      "step": 18307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829315209027956e-05,
+      "loss": 0.51,
+      "step": 18308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829251090023083e-05,
+      "loss": 0.4827,
+      "step": 18309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8291869710182097e-05,
+      "loss": 0.4466,
+      "step": 18310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829122852013337e-05,
+      "loss": 0.515,
+      "step": 18311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829058733008464e-05,
+      "loss": 0.536,
+      "step": 18312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828994614003591e-05,
+      "loss": 0.5027,
+      "step": 18313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828930494998718e-05,
+      "loss": 0.5672,
+      "step": 18314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828866375993845e-05,
+      "loss": 0.5785,
+      "step": 18315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828802256988972e-05,
+      "loss": 0.4244,
+      "step": 18316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8287381379840985e-05,
+      "loss": 0.5652,
+      "step": 18317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828674018979226e-05,
+      "loss": 0.4585,
+      "step": 18318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828609899974352e-05,
+      "loss": 0.5334,
+      "step": 18319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.82854578096948e-05,
+      "loss": 0.5798,
+      "step": 18320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8284816619646065e-05,
+      "loss": 0.5572,
+      "step": 18321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828417542959734e-05,
+      "loss": 0.4982,
+      "step": 18322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.82835342395486e-05,
+      "loss": 0.4914,
+      "step": 18323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8282893049499873e-05,
+      "loss": 0.5518,
+      "step": 18324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8282251859451145e-05,
+      "loss": 0.442,
+      "step": 18325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828161066940241e-05,
+      "loss": 0.452,
+      "step": 18326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828096947935368e-05,
+      "loss": 0.5853,
+      "step": 18327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8280328289304954e-05,
+      "loss": 0.4779,
+      "step": 18328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8279687099256225e-05,
+      "loss": 0.5289,
+      "step": 18329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.827904590920749e-05,
+      "loss": 0.6204,
+      "step": 18330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.827840471915876e-05,
+      "loss": 0.566,
+      "step": 18331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.827776352911003e-05,
+      "loss": 0.545,
+      "step": 18332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.82771223390613e-05,
+      "loss": 0.5813,
+      "step": 18333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.827648114901257e-05,
+      "loss": 0.5049,
+      "step": 18334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8275839958963835e-05,
+      "loss": 0.4601,
+      "step": 18335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8275198768915114e-05,
+      "loss": 0.4946,
+      "step": 18336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.827455757886638e-05,
+      "loss": 0.4489,
+      "step": 18337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827391638881765e-05,
+      "loss": 0.5133,
+      "step": 18338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8273275198768915e-05,
+      "loss": 0.5725,
+      "step": 18339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827263400872019e-05,
+      "loss": 0.5143,
+      "step": 18340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827199281867145e-05,
+      "loss": 0.5342,
+      "step": 18341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8271351628622724e-05,
+      "loss": 0.5489,
+      "step": 18342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8270710438573995e-05,
+      "loss": 0.6103,
+      "step": 18343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827006924852527e-05,
+      "loss": 0.5461,
+      "step": 18344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826942805847654e-05,
+      "loss": 0.4614,
+      "step": 18345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8268786868427804e-05,
+      "loss": 0.4908,
+      "step": 18346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8268145678379076e-05,
+      "loss": 0.5046,
+      "step": 18347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826750448833034e-05,
+      "loss": 0.5914,
+      "step": 18348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826686329828161e-05,
+      "loss": 0.5301,
+      "step": 18349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826622210823288e-05,
+      "loss": 0.5753,
+      "step": 18350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826558091818415e-05,
+      "loss": 0.4935,
+      "step": 18351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826493972813542e-05,
+      "loss": 0.5411,
+      "step": 18352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826429853808669e-05,
+      "loss": 0.4356,
+      "step": 18353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8263657348037964e-05,
+      "loss": 0.5096,
+      "step": 18354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826301615798923e-05,
+      "loss": 0.5114,
+      "step": 18355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82623749679405e-05,
+      "loss": 0.5426,
+      "step": 18356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8261733777891766e-05,
+      "loss": 0.4656,
+      "step": 18357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826109258784304e-05,
+      "loss": 0.467,
+      "step": 18358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826045139779431e-05,
+      "loss": 0.4653,
+      "step": 18359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825981020774558e-05,
+      "loss": 0.581,
+      "step": 18360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825916901769685e-05,
+      "loss": 0.487,
+      "step": 18361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825852782764812e-05,
+      "loss": 0.4622,
+      "step": 18362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825788663759939e-05,
+      "loss": 0.4846,
+      "step": 18363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8257245447550654e-05,
+      "loss": 0.5476,
+      "step": 18364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8256604257501926e-05,
+      "loss": 0.4987,
+      "step": 18365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825596306745319e-05,
+      "loss": 0.4622,
+      "step": 18366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825532187740446e-05,
+      "loss": 0.5296,
+      "step": 18367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8254680687355734e-05,
+      "loss": 0.4934,
+      "step": 18368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8254039497307006e-05,
+      "loss": 0.5493,
+      "step": 18369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825339830725828e-05,
+      "loss": 0.5029,
+      "step": 18370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825275711720954e-05,
+      "loss": 0.4988,
+      "step": 18371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8252115927160814e-05,
+      "loss": 0.6135,
+      "step": 18372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825147473711208e-05,
+      "loss": 0.5126,
+      "step": 18373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825083354706335e-05,
+      "loss": 0.5433,
+      "step": 18374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825019235701462e-05,
+      "loss": 0.6851,
+      "step": 18375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8249551166965894e-05,
+      "loss": 0.6216,
+      "step": 18376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824890997691716e-05,
+      "loss": 0.5548,
+      "step": 18377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824826878686843e-05,
+      "loss": 0.5043,
+      "step": 18378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82476275968197e-05,
+      "loss": 0.506,
+      "step": 18379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824698640677097e-05,
+      "loss": 0.5991,
+      "step": 18380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824634521672224e-05,
+      "loss": 0.5482,
+      "step": 18381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8245704026673504e-05,
+      "loss": 0.5719,
+      "step": 18382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8245062836624776e-05,
+      "loss": 0.6514,
+      "step": 18383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824442164657605e-05,
+      "loss": 0.5531,
+      "step": 18384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824378045652732e-05,
+      "loss": 0.5158,
+      "step": 18385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8243139266478584e-05,
+      "loss": 0.4828,
+      "step": 18386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8242498076429856e-05,
+      "loss": 0.4854,
+      "step": 18387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824185688638113e-05,
+      "loss": 0.5224,
+      "step": 18388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824121569633239e-05,
+      "loss": 0.5642,
+      "step": 18389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8240574506283664e-05,
+      "loss": 0.5518,
+      "step": 18390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823993331623493e-05,
+      "loss": 0.5676,
+      "step": 18391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823929212618621e-05,
+      "loss": 0.5526,
+      "step": 18392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823865093613747e-05,
+      "loss": 0.6126,
+      "step": 18393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8238009746088745e-05,
+      "loss": 0.5339,
+      "step": 18394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823736855604001e-05,
+      "loss": 0.5854,
+      "step": 18395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823672736599128e-05,
+      "loss": 0.498,
+      "step": 18396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823608617594255e-05,
+      "loss": 0.4524,
+      "step": 18397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823544498589382e-05,
+      "loss": 0.5321,
+      "step": 18398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823480379584509e-05,
+      "loss": 0.5198,
+      "step": 18399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823416260579636e-05,
+      "loss": 0.4652,
+      "step": 18400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823352141574763e-05,
+      "loss": 0.4706,
+      "step": 18401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82328802256989e-05,
+      "loss": 0.6196,
+      "step": 18402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823223903565017e-05,
+      "loss": 0.4168,
+      "step": 18403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8231597845601435e-05,
+      "loss": 0.4973,
+      "step": 18404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8230956655552706e-05,
+      "loss": 0.5881,
+      "step": 18405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823031546550398e-05,
+      "loss": 0.5372,
+      "step": 18406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822967427545524e-05,
+      "loss": 0.4547,
+      "step": 18407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822903308540652e-05,
+      "loss": 0.4999,
+      "step": 18408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8228391895357786e-05,
+      "loss": 0.513,
+      "step": 18409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822775070530906e-05,
+      "loss": 0.4948,
+      "step": 18410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822710951526032e-05,
+      "loss": 0.4715,
+      "step": 18411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8226468325211595e-05,
+      "loss": 0.5796,
+      "step": 18412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822582713516286e-05,
+      "loss": 0.4733,
+      "step": 18413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822518594511413e-05,
+      "loss": 0.46,
+      "step": 18414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82245447550654e-05,
+      "loss": 0.4553,
+      "step": 18415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8223903565016675e-05,
+      "loss": 0.6193,
+      "step": 18416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8223262374967947e-05,
+      "loss": 0.5744,
+      "step": 18417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822262118491921e-05,
+      "loss": 0.5475,
+      "step": 18418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822197999487048e-05,
+      "loss": 0.3776,
+      "step": 18419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822133880482175e-05,
+      "loss": 0.4782,
+      "step": 18420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822069761477302e-05,
+      "loss": 0.5079,
+      "step": 18421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8220056424724285e-05,
+      "loss": 0.4291,
+      "step": 18422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8219415234675557e-05,
+      "loss": 0.4904,
+      "step": 18423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821877404462683e-05,
+      "loss": 0.6036,
+      "step": 18424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82181328545781e-05,
+      "loss": 0.6088,
+      "step": 18425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821749166452937e-05,
+      "loss": 0.4562,
+      "step": 18426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821685047448064e-05,
+      "loss": 0.5344,
+      "step": 18427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821620928443191e-05,
+      "loss": 0.4397,
+      "step": 18428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821556809438317e-05,
+      "loss": 0.5597,
+      "step": 18429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8214926904334445e-05,
+      "loss": 0.5225,
+      "step": 18430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821428571428572e-05,
+      "loss": 0.5257,
+      "step": 18431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821364452423699e-05,
+      "loss": 0.4146,
+      "step": 18432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821300333418826e-05,
+      "loss": 0.4971,
+      "step": 18433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8212362144139525e-05,
+      "loss": 0.4513,
+      "step": 18434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82117209540908e-05,
+      "loss": 0.5277,
+      "step": 18435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821107976404206e-05,
+      "loss": 0.5493,
+      "step": 18436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8210438573993333e-05,
+      "loss": 0.6237,
+      "step": 18437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82097973839446e-05,
+      "loss": 0.4208,
+      "step": 18438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820915619389587e-05,
+      "loss": 0.5338,
+      "step": 18439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820851500384714e-05,
+      "loss": 0.4695,
+      "step": 18440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8207873813798414e-05,
+      "loss": 0.5342,
+      "step": 18441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8207232623749685e-05,
+      "loss": 0.5411,
+      "step": 18442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820659143370095e-05,
+      "loss": 0.4836,
+      "step": 18443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820595024365222e-05,
+      "loss": 0.423,
+      "step": 18444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820530905360349e-05,
+      "loss": 0.5857,
+      "step": 18445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820466786355476e-05,
+      "loss": 0.5771,
+      "step": 18446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820402667350603e-05,
+      "loss": 0.4462,
+      "step": 18447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82033854834573e-05,
+      "loss": 0.5349,
+      "step": 18448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820274429340857e-05,
+      "loss": 0.3963,
+      "step": 18449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820210310335984e-05,
+      "loss": 0.4368,
+      "step": 18450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820146191331111e-05,
+      "loss": 0.4876,
+      "step": 18451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8200820723262375e-05,
+      "loss": 0.5201,
+      "step": 18452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820017953321365e-05,
+      "loss": 0.5278,
+      "step": 18453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819953834316491e-05,
+      "loss": 0.4748,
+      "step": 18454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8198897153116184e-05,
+      "loss": 0.3847,
+      "step": 18455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8198255963067455e-05,
+      "loss": 0.475,
+      "step": 18456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819761477301873e-05,
+      "loss": 0.5249,
+      "step": 18457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819697358296999e-05,
+      "loss": 0.6197,
+      "step": 18458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8196332392921264e-05,
+      "loss": 0.5618,
+      "step": 18459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8195691202872536e-05,
+      "loss": 0.5225,
+      "step": 18460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81950500128238e-05,
+      "loss": 0.5842,
+      "step": 18461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819440882277507e-05,
+      "loss": 0.5257,
+      "step": 18462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819376763272634e-05,
+      "loss": 0.5529,
+      "step": 18463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8193126442677616e-05,
+      "loss": 0.4379,
+      "step": 18464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819248525262888e-05,
+      "loss": 0.4895,
+      "step": 18465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819184406258015e-05,
+      "loss": 0.4885,
+      "step": 18466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819120287253142e-05,
+      "loss": 0.5705,
+      "step": 18467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819056168248269e-05,
+      "loss": 0.6663,
+      "step": 18468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818992049243396e-05,
+      "loss": 0.528,
+      "step": 18469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8189279302385226e-05,
+      "loss": 0.606,
+      "step": 18470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81886381123365e-05,
+      "loss": 0.4174,
+      "step": 18471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818799692228777e-05,
+      "loss": 0.492,
+      "step": 18472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818735573223904e-05,
+      "loss": 0.5547,
+      "step": 18473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8186714542190306e-05,
+      "loss": 0.5049,
+      "step": 18474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818607335214158e-05,
+      "loss": 0.5052,
+      "step": 18475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818543216209284e-05,
+      "loss": 0.4459,
+      "step": 18476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8184790972044114e-05,
+      "loss": 0.4489,
+      "step": 18477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8184149781995386e-05,
+      "loss": 0.5569,
+      "step": 18478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818350859194665e-05,
+      "loss": 0.4963,
+      "step": 18479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818286740189793e-05,
+      "loss": 0.4328,
+      "step": 18480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8182226211849194e-05,
+      "loss": 0.4454,
+      "step": 18481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8181585021800466e-05,
+      "loss": 0.4665,
+      "step": 18482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818094383175173e-05,
+      "loss": 0.478,
+      "step": 18483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8180302641703e-05,
+      "loss": 0.6564,
+      "step": 18484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817966145165427e-05,
+      "loss": 0.4245,
+      "step": 18485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817902026160554e-05,
+      "loss": 0.536,
+      "step": 18486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817837907155681e-05,
+      "loss": 0.6911,
+      "step": 18487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817773788150808e-05,
+      "loss": 0.421,
+      "step": 18488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8177096691459354e-05,
+      "loss": 0.5456,
+      "step": 18489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817645550141062e-05,
+      "loss": 0.4497,
+      "step": 18490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817581431136189e-05,
+      "loss": 0.4814,
+      "step": 18491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8175173121313156e-05,
+      "loss": 0.48,
+      "step": 18492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817453193126443e-05,
+      "loss": 0.5535,
+      "step": 18493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817389074121569e-05,
+      "loss": 0.4992,
+      "step": 18494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8173249551166964e-05,
+      "loss": 0.5093,
+      "step": 18495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817260836111824e-05,
+      "loss": 0.4223,
+      "step": 18496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817196717106951e-05,
+      "loss": 0.5782,
+      "step": 18497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817132598102078e-05,
+      "loss": 0.5328,
+      "step": 18498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8170684790972044e-05,
+      "loss": 0.4403,
+      "step": 18499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8170043600923316e-05,
+      "loss": 0.5635,
+      "step": 18500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816940241087458e-05,
+      "loss": 0.4694,
+      "step": 18501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816876122082585e-05,
+      "loss": 0.5189,
+      "step": 18502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8168120030777124e-05,
+      "loss": 0.5665,
+      "step": 18503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8167478840728396e-05,
+      "loss": 0.5145,
+      "step": 18504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816683765067967e-05,
+      "loss": 0.7517,
+      "step": 18505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816619646063093e-05,
+      "loss": 0.5539,
+      "step": 18506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8165555270582205e-05,
+      "loss": 0.5952,
+      "step": 18507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816491408053347e-05,
+      "loss": 0.5498,
+      "step": 18508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816427289048474e-05,
+      "loss": 0.4986,
+      "step": 18509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8163631700436006e-05,
+      "loss": 0.5227,
+      "step": 18510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816299051038728e-05,
+      "loss": 0.6048,
+      "step": 18511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816234932033855e-05,
+      "loss": 0.3732,
+      "step": 18512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816170813028982e-05,
+      "loss": 0.5406,
+      "step": 18513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816106694024109e-05,
+      "loss": 0.4905,
+      "step": 18514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816042575019236e-05,
+      "loss": 0.4109,
+      "step": 18515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815978456014363e-05,
+      "loss": 0.4016,
+      "step": 18516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8159143370094895e-05,
+      "loss": 0.4763,
+      "step": 18517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8158502180046166e-05,
+      "loss": 0.4403,
+      "step": 18518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815786098999744e-05,
+      "loss": 0.423,
+      "step": 18519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815721979994871e-05,
+      "loss": 0.6258,
+      "step": 18520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815657860989998e-05,
+      "loss": 0.4017,
+      "step": 18521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8155937419851246e-05,
+      "loss": 0.6296,
+      "step": 18522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815529622980252e-05,
+      "loss": 0.4644,
+      "step": 18523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815465503975378e-05,
+      "loss": 0.4982,
+      "step": 18524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8154013849705055e-05,
+      "loss": 0.4793,
+      "step": 18525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815337265965632e-05,
+      "loss": 0.4859,
+      "step": 18526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815273146960759e-05,
+      "loss": 0.5648,
+      "step": 18527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815209027955886e-05,
+      "loss": 0.6346,
+      "step": 18528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8151449089510135e-05,
+      "loss": 0.6464,
+      "step": 18529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815080789946141e-05,
+      "loss": 0.4464,
+      "step": 18530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815016670941267e-05,
+      "loss": 0.4779,
+      "step": 18531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814952551936394e-05,
+      "loss": 0.5347,
+      "step": 18532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814888432931521e-05,
+      "loss": 0.5321,
+      "step": 18533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814824313926648e-05,
+      "loss": 0.4752,
+      "step": 18534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814760194921775e-05,
+      "loss": 0.3789,
+      "step": 18535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814696075916902e-05,
+      "loss": 0.5785,
+      "step": 18536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814631956912029e-05,
+      "loss": 0.5087,
+      "step": 18537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814567837907156e-05,
+      "loss": 0.5356,
+      "step": 18538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814503718902283e-05,
+      "loss": 0.5831,
+      "step": 18539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81443959989741e-05,
+      "loss": 0.5709,
+      "step": 18540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814375480892537e-05,
+      "loss": 0.4786,
+      "step": 18541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814311361887663e-05,
+      "loss": 0.5151,
+      "step": 18542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8142472428827905e-05,
+      "loss": 0.5386,
+      "step": 18543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814183123877918e-05,
+      "loss": 0.4869,
+      "step": 18544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814119004873045e-05,
+      "loss": 0.538,
+      "step": 18545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8140548858681713e-05,
+      "loss": 0.4237,
+      "step": 18546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8139907668632985e-05,
+      "loss": 0.5163,
+      "step": 18547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813926647858426e-05,
+      "loss": 0.5633,
+      "step": 18548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813862528853552e-05,
+      "loss": 0.5531,
+      "step": 18549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8137984098486794e-05,
+      "loss": 0.4756,
+      "step": 18550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813734290843806e-05,
+      "loss": 0.4975,
+      "step": 18551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813670171838934e-05,
+      "loss": 0.5986,
+      "step": 18552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81360605283406e-05,
+      "loss": 0.4586,
+      "step": 18553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8135419338291874e-05,
+      "loss": 0.5081,
+      "step": 18554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813477814824314e-05,
+      "loss": 0.6209,
+      "step": 18555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813413695819441e-05,
+      "loss": 0.6451,
+      "step": 18556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813349576814568e-05,
+      "loss": 0.5955,
+      "step": 18557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813285457809695e-05,
+      "loss": 0.4849,
+      "step": 18558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813221338804822e-05,
+      "loss": 0.542,
+      "step": 18559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813157219799949e-05,
+      "loss": 0.4995,
+      "step": 18560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813093100795076e-05,
+      "loss": 0.4138,
+      "step": 18561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813028981790203e-05,
+      "loss": 0.489,
+      "step": 18562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81296486278533e-05,
+      "loss": 0.4496,
+      "step": 18563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8129007437804564e-05,
+      "loss": 0.5358,
+      "step": 18564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8128366247755835e-05,
+      "loss": 0.5421,
+      "step": 18565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812772505770711e-05,
+      "loss": 0.4728,
+      "step": 18566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812708386765837e-05,
+      "loss": 0.4197,
+      "step": 18567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812644267760965e-05,
+      "loss": 0.5378,
+      "step": 18568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8125801487560915e-05,
+      "loss": 0.4381,
+      "step": 18569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812516029751219e-05,
+      "loss": 0.5561,
+      "step": 18570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812451910746345e-05,
+      "loss": 0.4066,
+      "step": 18571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8123877917414724e-05,
+      "loss": 0.5013,
+      "step": 18572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812323672736599e-05,
+      "loss": 0.508,
+      "step": 18573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812259553731726e-05,
+      "loss": 0.4523,
+      "step": 18574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812195434726853e-05,
+      "loss": 0.4917,
+      "step": 18575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8121313157219804e-05,
+      "loss": 0.5476,
+      "step": 18576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8120671967171076e-05,
+      "loss": 0.5267,
+      "step": 18577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812003077712234e-05,
+      "loss": 0.4613,
+      "step": 18578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811938958707361e-05,
+      "loss": 0.5199,
+      "step": 18579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811874839702488e-05,
+      "loss": 0.5759,
+      "step": 18580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811810720697615e-05,
+      "loss": 0.5357,
+      "step": 18581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8117466016927414e-05,
+      "loss": 0.5185,
+      "step": 18582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8116824826878686e-05,
+      "loss": 0.6213,
+      "step": 18583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811618363682996e-05,
+      "loss": 0.5344,
+      "step": 18584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811554244678123e-05,
+      "loss": 0.6488,
+      "step": 18585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81149012567325e-05,
+      "loss": 0.507,
+      "step": 18586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8114260066683766e-05,
+      "loss": 0.4488,
+      "step": 18587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811361887663504e-05,
+      "loss": 0.6832,
+      "step": 18588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81129776865863e-05,
+      "loss": 0.6617,
+      "step": 18589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8112336496537574e-05,
+      "loss": 0.6309,
+      "step": 18590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8111695306488846e-05,
+      "loss": 0.567,
+      "step": 18591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811105411644012e-05,
+      "loss": 0.4245,
+      "step": 18592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811041292639139e-05,
+      "loss": 0.4486,
+      "step": 18593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8109771736342654e-05,
+      "loss": 0.4353,
+      "step": 18594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8109130546293926e-05,
+      "loss": 0.4774,
+      "step": 18595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.810848935624519e-05,
+      "loss": 0.5516,
+      "step": 18596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.810784816619646e-05,
+      "loss": 0.434,
+      "step": 18597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810720697614773e-05,
+      "loss": 0.5027,
+      "step": 18598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8106565786099e-05,
+      "loss": 0.5815,
+      "step": 18599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810592459605027e-05,
+      "loss": 0.5542,
+      "step": 18600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810528340600154e-05,
+      "loss": 0.5255,
+      "step": 18601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8104642215952814e-05,
+      "loss": 0.5614,
+      "step": 18602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810400102590408e-05,
+      "loss": 0.5178,
+      "step": 18603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810335983585535e-05,
+      "loss": 0.6012,
+      "step": 18604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8102718645806616e-05,
+      "loss": 0.5447,
+      "step": 18605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810207745575789e-05,
+      "loss": 0.6133,
+      "step": 18606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810143626570916e-05,
+      "loss": 0.4001,
+      "step": 18607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810079507566043e-05,
+      "loss": 0.4444,
+      "step": 18608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8100153885611696e-05,
+      "loss": 0.4829,
+      "step": 18609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809951269556297e-05,
+      "loss": 0.5527,
+      "step": 18610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809887150551424e-05,
+      "loss": 0.5139,
+      "step": 18611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8098230315465504e-05,
+      "loss": 0.4888,
+      "step": 18612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8097589125416776e-05,
+      "loss": 0.4649,
+      "step": 18613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809694793536804e-05,
+      "loss": 0.5813,
+      "step": 18614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809630674531931e-05,
+      "loss": 0.5556,
+      "step": 18615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8095665555270585e-05,
+      "loss": 0.4999,
+      "step": 18616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8095024365221856e-05,
+      "loss": 0.4299,
+      "step": 18617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809438317517312e-05,
+      "loss": 0.505,
+      "step": 18618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809374198512439e-05,
+      "loss": 0.4724,
+      "step": 18619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8093100795075665e-05,
+      "loss": 0.5472,
+      "step": 18620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809245960502693e-05,
+      "loss": 0.4578,
+      "step": 18621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80918184149782e-05,
+      "loss": 0.5937,
+      "step": 18622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809117722492947e-05,
+      "loss": 0.4646,
+      "step": 18623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8090536034880745e-05,
+      "loss": 0.5024,
+      "step": 18624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808989484483201e-05,
+      "loss": 0.5417,
+      "step": 18625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808925365478328e-05,
+      "loss": 0.5357,
+      "step": 18626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8088612464734546e-05,
+      "loss": 0.64,
+      "step": 18627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808797127468582e-05,
+      "loss": 0.4687,
+      "step": 18628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808733008463709e-05,
+      "loss": 0.5048,
+      "step": 18629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8086688894588355e-05,
+      "loss": 0.5024,
+      "step": 18630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8086047704539626e-05,
+      "loss": 0.5456,
+      "step": 18631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80854065144909e-05,
+      "loss": 0.7135,
+      "step": 18632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808476532444217e-05,
+      "loss": 0.3701,
+      "step": 18633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8084124134393435e-05,
+      "loss": 0.5728,
+      "step": 18634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8083482944344706e-05,
+      "loss": 0.5165,
+      "step": 18635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808284175429597e-05,
+      "loss": 0.5586,
+      "step": 18636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808220056424724e-05,
+      "loss": 0.6193,
+      "step": 18637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8081559374198515e-05,
+      "loss": 0.5717,
+      "step": 18638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808091818414978e-05,
+      "loss": 0.533,
+      "step": 18639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808027699410106e-05,
+      "loss": 0.638,
+      "step": 18640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807963580405232e-05,
+      "loss": 0.5414,
+      "step": 18641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8078994614003595e-05,
+      "loss": 0.5471,
+      "step": 18642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807835342395486e-05,
+      "loss": 0.4358,
+      "step": 18643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807771223390613e-05,
+      "loss": 0.4986,
+      "step": 18644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8077071043857397e-05,
+      "loss": 0.5509,
+      "step": 18645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807642985380867e-05,
+      "loss": 0.577,
+      "step": 18646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807578866375994e-05,
+      "loss": 0.5277,
+      "step": 18647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807514747371121e-05,
+      "loss": 0.568,
+      "step": 18648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8074506283662483e-05,
+      "loss": 0.6164,
+      "step": 18649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807386509361375e-05,
+      "loss": 0.5023,
+      "step": 18650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807322390356502e-05,
+      "loss": 0.5015,
+      "step": 18651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8072582713516285e-05,
+      "loss": 0.608,
+      "step": 18652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807194152346756e-05,
+      "loss": 0.4917,
+      "step": 18653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807130033341882e-05,
+      "loss": 0.5266,
+      "step": 18654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807065914337009e-05,
+      "loss": 0.5232,
+      "step": 18655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807001795332137e-05,
+      "loss": 0.4543,
+      "step": 18656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806937676327264e-05,
+      "loss": 0.4779,
+      "step": 18657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806873557322391e-05,
+      "loss": 0.4632,
+      "step": 18658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8068094383175173e-05,
+      "loss": 0.5032,
+      "step": 18659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8067453193126445e-05,
+      "loss": 0.5075,
+      "step": 18660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806681200307771e-05,
+      "loss": 0.4823,
+      "step": 18661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806617081302898e-05,
+      "loss": 0.5617,
+      "step": 18662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8065529622980254e-05,
+      "loss": 0.4889,
+      "step": 18663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8064888432931525e-05,
+      "loss": 0.5637,
+      "step": 18664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80642472428828e-05,
+      "loss": 0.5662,
+      "step": 18665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806360605283406e-05,
+      "loss": 0.5442,
+      "step": 18666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8062964862785334e-05,
+      "loss": 0.4902,
+      "step": 18667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80623236727366e-05,
+      "loss": 0.5939,
+      "step": 18668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806168248268787e-05,
+      "loss": 0.5579,
+      "step": 18669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8061041292639135e-05,
+      "loss": 0.5303,
+      "step": 18670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806040010259041e-05,
+      "loss": 0.3896,
+      "step": 18671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805975891254168e-05,
+      "loss": 0.4987,
+      "step": 18672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805911772249295e-05,
+      "loss": 0.4735,
+      "step": 18673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805847653244422e-05,
+      "loss": 0.5595,
+      "step": 18674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805783534239549e-05,
+      "loss": 0.5724,
+      "step": 18675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805719415234676e-05,
+      "loss": 0.46,
+      "step": 18676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8056552962298024e-05,
+      "loss": 0.583,
+      "step": 18677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8055911772249295e-05,
+      "loss": 0.5075,
+      "step": 18678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805527058220057e-05,
+      "loss": 0.5132,
+      "step": 18679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805462939215184e-05,
+      "loss": 0.5672,
+      "step": 18680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8053988202103104e-05,
+      "loss": 0.4294,
+      "step": 18681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8053347012054376e-05,
+      "loss": 0.4749,
+      "step": 18682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805270582200565e-05,
+      "loss": 0.5266,
+      "step": 18683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805206463195691e-05,
+      "loss": 0.5827,
+      "step": 18684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8051423441908184e-05,
+      "loss": 0.6399,
+      "step": 18685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805078225185945e-05,
+      "loss": 0.4754,
+      "step": 18686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805014106181072e-05,
+      "loss": 0.4731,
+      "step": 18687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804949987176199e-05,
+      "loss": 0.5325,
+      "step": 18688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8048858681713264e-05,
+      "loss": 0.4102,
+      "step": 18689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804821749166453e-05,
+      "loss": 0.5578,
+      "step": 18690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80475763016158e-05,
+      "loss": 0.6347,
+      "step": 18691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804693511156707e-05,
+      "loss": 0.6201,
+      "step": 18692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804629392151834e-05,
+      "loss": 0.4558,
+      "step": 18693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804565273146961e-05,
+      "loss": 0.4426,
+      "step": 18694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804501154142088e-05,
+      "loss": 0.4881,
+      "step": 18695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804437035137215e-05,
+      "loss": 0.4882,
+      "step": 18696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804372916132342e-05,
+      "loss": 0.5257,
+      "step": 18697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804308797127469e-05,
+      "loss": 0.4614,
+      "step": 18698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804244678122596e-05,
+      "loss": 0.4652,
+      "step": 18699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8041805591177226e-05,
+      "loss": 0.6429,
+      "step": 18700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80411644011285e-05,
+      "loss": 0.5411,
+      "step": 18701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804052321107976e-05,
+      "loss": 0.4885,
+      "step": 18702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8039882021031034e-05,
+      "loss": 0.6211,
+      "step": 18703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8039240830982306e-05,
+      "loss": 0.5091,
+      "step": 18704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803859964093358e-05,
+      "loss": 0.481,
+      "step": 18705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803795845088484e-05,
+      "loss": 0.6212,
+      "step": 18706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8037317260836114e-05,
+      "loss": 0.4871,
+      "step": 18707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8036676070787386e-05,
+      "loss": 0.5769,
+      "step": 18708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803603488073865e-05,
+      "loss": 0.5421,
+      "step": 18709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803539369068992e-05,
+      "loss": 0.4864,
+      "step": 18710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803475250064119e-05,
+      "loss": 0.4566,
+      "step": 18711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8034111310592466e-05,
+      "loss": 0.4953,
+      "step": 18712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803347012054373e-05,
+      "loss": 0.5006,
+      "step": 18713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8032828930495e-05,
+      "loss": 0.533,
+      "step": 18714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803218774044627e-05,
+      "loss": 0.5118,
+      "step": 18715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803154655039754e-05,
+      "loss": 0.4711,
+      "step": 18716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803090536034881e-05,
+      "loss": 0.4893,
+      "step": 18717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8030264170300076e-05,
+      "loss": 0.5099,
+      "step": 18718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802962298025135e-05,
+      "loss": 0.5223,
+      "step": 18719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802898179020262e-05,
+      "loss": 0.5,
+      "step": 18720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802834060015389e-05,
+      "loss": 0.4576,
+      "step": 18721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8027699410105156e-05,
+      "loss": 0.5356,
+      "step": 18722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802705822005643e-05,
+      "loss": 0.4884,
+      "step": 18723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802641703000769e-05,
+      "loss": 0.6521,
+      "step": 18724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8025775839958964e-05,
+      "loss": 0.4715,
+      "step": 18725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8025134649910236e-05,
+      "loss": 0.5213,
+      "step": 18726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80244934598615e-05,
+      "loss": 0.4761,
+      "step": 18727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802385226981278e-05,
+      "loss": 0.511,
+      "step": 18728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8023211079764045e-05,
+      "loss": 0.5508,
+      "step": 18729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8022569889715316e-05,
+      "loss": 0.5315,
+      "step": 18730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802192869966658e-05,
+      "loss": 0.4468,
+      "step": 18731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802128750961785e-05,
+      "loss": 0.5499,
+      "step": 18732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802064631956912e-05,
+      "loss": 0.4594,
+      "step": 18733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802000512952039e-05,
+      "loss": 0.5036,
+      "step": 18734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801936393947166e-05,
+      "loss": 0.4544,
+      "step": 18735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801872274942293e-05,
+      "loss": 0.4559,
+      "step": 18736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8018081559374205e-05,
+      "loss": 0.4659,
+      "step": 18737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801744036932547e-05,
+      "loss": 0.6515,
+      "step": 18738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801679917927674e-05,
+      "loss": 0.5221,
+      "step": 18739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8016157989228006e-05,
+      "loss": 0.5534,
+      "step": 18740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801551679917928e-05,
+      "loss": 0.4745,
+      "step": 18741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801487560913054e-05,
+      "loss": 0.5221,
+      "step": 18742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8014234419081815e-05,
+      "loss": 0.5767,
+      "step": 18743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801359322903309e-05,
+      "loss": 0.5107,
+      "step": 18744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801295203898436e-05,
+      "loss": 0.5612,
+      "step": 18745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801231084893563e-05,
+      "loss": 0.5915,
+      "step": 18746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8011669658886895e-05,
+      "loss": 0.4802,
+      "step": 18747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8011028468838167e-05,
+      "loss": 0.4986,
+      "step": 18748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801038727878943e-05,
+      "loss": 0.5723,
+      "step": 18749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80097460887407e-05,
+      "loss": 0.5535,
+      "step": 18750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8009104898691975e-05,
+      "loss": 0.4441,
+      "step": 18751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8008463708643247e-05,
+      "loss": 0.5394,
+      "step": 18752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800782251859452e-05,
+      "loss": 0.4923,
+      "step": 18753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800718132854578e-05,
+      "loss": 0.4879,
+      "step": 18754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8006540138497055e-05,
+      "loss": 0.5374,
+      "step": 18755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800589894844832e-05,
+      "loss": 0.5785,
+      "step": 18756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800525775839959e-05,
+      "loss": 0.4924,
+      "step": 18757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8004616568350857e-05,
+      "loss": 0.5135,
+      "step": 18758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800397537830213e-05,
+      "loss": 0.5322,
+      "step": 18759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80033341882534e-05,
+      "loss": 0.5274,
+      "step": 18760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800269299820467e-05,
+      "loss": 0.4157,
+      "step": 18761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8002051808155943e-05,
+      "loss": 0.4286,
+      "step": 18762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800141061810721e-05,
+      "loss": 0.4649,
+      "step": 18763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800076942805848e-05,
+      "loss": 0.4779,
+      "step": 18764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8000128238009745e-05,
+      "loss": 0.5102,
+      "step": 18765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799948704796102e-05,
+      "loss": 0.4668,
+      "step": 18766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799884585791229e-05,
+      "loss": 0.4935,
+      "step": 18767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799820466786356e-05,
+      "loss": 0.5106,
+      "step": 18768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7997563477814825e-05,
+      "loss": 0.5296,
+      "step": 18769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79969222877661e-05,
+      "loss": 0.4471,
+      "step": 18770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799628109771737e-05,
+      "loss": 0.4883,
+      "step": 18771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7995639907668634e-05,
+      "loss": 0.5379,
+      "step": 18772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7994998717619905e-05,
+      "loss": 0.5695,
+      "step": 18773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799435752757117e-05,
+      "loss": 0.4985,
+      "step": 18774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799371633752244e-05,
+      "loss": 0.4922,
+      "step": 18775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7993075147473714e-05,
+      "loss": 0.5734,
+      "step": 18776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7992433957424985e-05,
+      "loss": 0.4646,
+      "step": 18777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799179276737625e-05,
+      "loss": 0.5995,
+      "step": 18778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799115157732752e-05,
+      "loss": 0.5083,
+      "step": 18779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7990510387278794e-05,
+      "loss": 0.5451,
+      "step": 18780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798986919723006e-05,
+      "loss": 0.4626,
+      "step": 18781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798922800718133e-05,
+      "loss": 0.4532,
+      "step": 18782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79885868171326e-05,
+      "loss": 0.5815,
+      "step": 18783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7987945627083874e-05,
+      "loss": 0.4728,
+      "step": 18784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798730443703514e-05,
+      "loss": 0.575,
+      "step": 18785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798666324698641e-05,
+      "loss": 0.6582,
+      "step": 18786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7986022056937675e-05,
+      "loss": 0.5263,
+      "step": 18787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798538086688895e-05,
+      "loss": 0.5161,
+      "step": 18788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798473967684022e-05,
+      "loss": 0.3985,
+      "step": 18789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7984098486791484e-05,
+      "loss": 0.5109,
+      "step": 18790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7983457296742755e-05,
+      "loss": 0.5136,
+      "step": 18791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798281610669403e-05,
+      "loss": 0.5472,
+      "step": 18792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79821749166453e-05,
+      "loss": 0.5423,
+      "step": 18793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7981533726596564e-05,
+      "loss": 0.4845,
+      "step": 18794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7980892536547836e-05,
+      "loss": 0.4862,
+      "step": 18795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79802513464991e-05,
+      "loss": 0.6183,
+      "step": 18796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797961015645037e-05,
+      "loss": 0.5073,
+      "step": 18797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7978968966401644e-05,
+      "loss": 0.4321,
+      "step": 18798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797832777635291e-05,
+      "loss": 0.5503,
+      "step": 18799
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797768658630419e-05,
+      "loss": 0.4473,
+      "step": 18800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797704539625545e-05,
+      "loss": 0.5529,
+      "step": 18801
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7976404206206724e-05,
+      "loss": 0.4634,
+      "step": 18802
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797576301615799e-05,
+      "loss": 0.5316,
+      "step": 18803
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797512182610926e-05,
+      "loss": 0.5391,
+      "step": 18804
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7974480636060526e-05,
+      "loss": 0.4205,
+      "step": 18805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79738394460118e-05,
+      "loss": 0.5666,
+      "step": 18806
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797319825596307e-05,
+      "loss": 0.5645,
+      "step": 18807
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797255706591434e-05,
+      "loss": 0.4926,
+      "step": 18808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797191587586561e-05,
+      "loss": 0.5067,
+      "step": 18809
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797127468581688e-05,
+      "loss": 0.5384,
+      "step": 18810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797063349576815e-05,
+      "loss": 0.4348,
+      "step": 18811
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7969992305719414e-05,
+      "loss": 0.5757,
+      "step": 18812
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7969351115670686e-05,
+      "loss": 0.5505,
+      "step": 18813
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796870992562195e-05,
+      "loss": 0.4564,
+      "step": 18814
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796806873557322e-05,
+      "loss": 0.44,
+      "step": 18815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79674275455245e-05,
+      "loss": 0.5725,
+      "step": 18816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7966786355475766e-05,
+      "loss": 0.3732,
+      "step": 18817
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796614516542704e-05,
+      "loss": 0.5141,
+      "step": 18818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79655039753783e-05,
+      "loss": 0.4921,
+      "step": 18819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7964862785329574e-05,
+      "loss": 0.5117,
+      "step": 18820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796422159528084e-05,
+      "loss": 0.534,
+      "step": 18821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796358040523211e-05,
+      "loss": 0.5555,
+      "step": 18822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796293921518338e-05,
+      "loss": 0.6055,
+      "step": 18823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7962298025134654e-05,
+      "loss": 0.511,
+      "step": 18824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7961656835085926e-05,
+      "loss": 0.5032,
+      "step": 18825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796101564503719e-05,
+      "loss": 0.5542,
+      "step": 18826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796037445498846e-05,
+      "loss": 0.5705,
+      "step": 18827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795973326493973e-05,
+      "loss": 0.4815,
+      "step": 18828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7959092074891e-05,
+      "loss": 0.4318,
+      "step": 18829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7958450884842264e-05,
+      "loss": 0.5742,
+      "step": 18830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7957809694793536e-05,
+      "loss": 0.5974,
+      "step": 18831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795716850474481e-05,
+      "loss": 0.5427,
+      "step": 18832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795652731469608e-05,
+      "loss": 0.4985,
+      "step": 18833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795588612464735e-05,
+      "loss": 0.6254,
+      "step": 18834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7955244934598616e-05,
+      "loss": 0.5893,
+      "step": 18835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795460374454989e-05,
+      "loss": 0.4866,
+      "step": 18836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795396255450115e-05,
+      "loss": 0.4681,
+      "step": 18837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7953321364452424e-05,
+      "loss": 0.5388,
+      "step": 18838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7952680174403696e-05,
+      "loss": 0.5113,
+      "step": 18839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795203898435497e-05,
+      "loss": 0.5182,
+      "step": 18840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795139779430623e-05,
+      "loss": 0.5201,
+      "step": 18841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7950756604257505e-05,
+      "loss": 0.4952,
+      "step": 18842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7950115414208776e-05,
+      "loss": 0.45,
+      "step": 18843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794947422416004e-05,
+      "loss": 0.4818,
+      "step": 18844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794883303411131e-05,
+      "loss": 0.5334,
+      "step": 18845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794819184406258e-05,
+      "loss": 0.4867,
+      "step": 18846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794755065401385e-05,
+      "loss": 0.4722,
+      "step": 18847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794690946396512e-05,
+      "loss": 0.5007,
+      "step": 18848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794626827391639e-05,
+      "loss": 0.5476,
+      "step": 18849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794562708386766e-05,
+      "loss": 0.4763,
+      "step": 18850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794498589381893e-05,
+      "loss": 0.4163,
+      "step": 18851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79443447037702e-05,
+      "loss": 0.4816,
+      "step": 18852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7943703513721466e-05,
+      "loss": 0.6884,
+      "step": 18853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794306232367274e-05,
+      "loss": 0.4976,
+      "step": 18854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794242113362401e-05,
+      "loss": 0.511,
+      "step": 18855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794177994357528e-05,
+      "loss": 0.5119,
+      "step": 18856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7941138753526546e-05,
+      "loss": 0.5052,
+      "step": 18857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.794049756347782e-05,
+      "loss": 0.4623,
+      "step": 18858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793985637342908e-05,
+      "loss": 0.5981,
+      "step": 18859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7939215183380355e-05,
+      "loss": 0.3626,
+      "step": 18860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7938573993331627e-05,
+      "loss": 0.4364,
+      "step": 18861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793793280328289e-05,
+      "loss": 0.5494,
+      "step": 18862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793729161323416e-05,
+      "loss": 0.5138,
+      "step": 18863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7936650423185435e-05,
+      "loss": 0.3633,
+      "step": 18864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793600923313671e-05,
+      "loss": 0.5844,
+      "step": 18865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793536804308797e-05,
+      "loss": 0.4523,
+      "step": 18866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793472685303924e-05,
+      "loss": 0.5713,
+      "step": 18867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793408566299051e-05,
+      "loss": 0.5063,
+      "step": 18868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793344447294178e-05,
+      "loss": 0.5578,
+      "step": 18869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793280328289305e-05,
+      "loss": 0.5523,
+      "step": 18870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7932162092844317e-05,
+      "loss": 0.4426,
+      "step": 18871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7931520902795595e-05,
+      "loss": 0.5623,
+      "step": 18872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793087971274686e-05,
+      "loss": 0.4675,
+      "step": 18873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793023852269813e-05,
+      "loss": 0.4831,
+      "step": 18874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.79295973326494e-05,
+      "loss": 0.4522,
+      "step": 18875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792895614260067e-05,
+      "loss": 0.6649,
+      "step": 18876
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792831495255193e-05,
+      "loss": 0.4427,
+      "step": 18877
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7927673762503205e-05,
+      "loss": 0.5427,
+      "step": 18878
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792703257245448e-05,
+      "loss": 0.5232,
+      "step": 18879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792639138240575e-05,
+      "loss": 0.4546,
+      "step": 18880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792575019235702e-05,
+      "loss": 0.5055,
+      "step": 18881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7925109002308285e-05,
+      "loss": 0.5338,
+      "step": 18882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792446781225956e-05,
+      "loss": 0.511,
+      "step": 18883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792382662221082e-05,
+      "loss": 0.4889,
+      "step": 18884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7923185432162094e-05,
+      "loss": 0.6418,
+      "step": 18885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7922544242113365e-05,
+      "loss": 0.5413,
+      "step": 18886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792190305206463e-05,
+      "loss": 0.5055,
+      "step": 18887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792126186201591e-05,
+      "loss": 0.5346,
+      "step": 18888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7920620671967174e-05,
+      "loss": 0.5298,
+      "step": 18889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7919979481918445e-05,
+      "loss": 0.5474,
+      "step": 18890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791933829186971e-05,
+      "loss": 0.5612,
+      "step": 18891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791869710182098e-05,
+      "loss": 0.479,
+      "step": 18892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791805591177225e-05,
+      "loss": 0.5371,
+      "step": 18893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791741472172352e-05,
+      "loss": 0.523,
+      "step": 18894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791677353167479e-05,
+      "loss": 0.386,
+      "step": 18895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791613234162606e-05,
+      "loss": 0.4257,
+      "step": 18896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7915491151577334e-05,
+      "loss": 0.4805,
+      "step": 18897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.79148499615286e-05,
+      "loss": 0.5296,
+      "step": 18898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791420877147987e-05,
+      "loss": 0.5913,
+      "step": 18899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7913567581431135e-05,
+      "loss": 0.5594,
+      "step": 18900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791292639138241e-05,
+      "loss": 0.4898,
+      "step": 18901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791228520133367e-05,
+      "loss": 0.5368,
+      "step": 18902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7911644011284944e-05,
+      "loss": 0.5612,
+      "step": 18903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791100282123622e-05,
+      "loss": 0.5654,
+      "step": 18904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791036163118749e-05,
+      "loss": 0.5551,
+      "step": 18905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790972044113876e-05,
+      "loss": 0.5588,
+      "step": 18906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7909079251090024e-05,
+      "loss": 0.4896,
+      "step": 18907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7908438061041296e-05,
+      "loss": 0.5315,
+      "step": 18908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790779687099256e-05,
+      "loss": 0.4848,
+      "step": 18909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790715568094383e-05,
+      "loss": 0.4601,
+      "step": 18910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7906514490895104e-05,
+      "loss": 0.4792,
+      "step": 18911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7905873300846376e-05,
+      "loss": 0.4732,
+      "step": 18912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790523211079765e-05,
+      "loss": 0.5228,
+      "step": 18913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790459092074891e-05,
+      "loss": 0.4943,
+      "step": 18914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7903949730700184e-05,
+      "loss": 0.5211,
+      "step": 18915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790330854065145e-05,
+      "loss": 0.6238,
+      "step": 18916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790266735060272e-05,
+      "loss": 0.4248,
+      "step": 18917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7902026160553986e-05,
+      "loss": 0.5793,
+      "step": 18918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790138497050526e-05,
+      "loss": 0.5645,
+      "step": 18919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790074378045653e-05,
+      "loss": 0.521,
+      "step": 18920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.79001025904078e-05,
+      "loss": 0.5476,
+      "step": 18921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789946140035907e-05,
+      "loss": 0.5448,
+      "step": 18922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789882021031034e-05,
+      "loss": 0.5445,
+      "step": 18923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789817902026161e-05,
+      "loss": 0.4899,
+      "step": 18924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7897537830212874e-05,
+      "loss": 0.5192,
+      "step": 18925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7896896640164146e-05,
+      "loss": 0.4852,
+      "step": 18926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789625545011542e-05,
+      "loss": 0.4953,
+      "step": 18927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789561426006669e-05,
+      "loss": 0.5603,
+      "step": 18928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7894973070017954e-05,
+      "loss": 0.5483,
+      "step": 18929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7894331879969226e-05,
+      "loss": 0.4915,
+      "step": 18930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78936906899205e-05,
+      "loss": 0.4828,
+      "step": 18931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789304949987176e-05,
+      "loss": 0.5223,
+      "step": 18932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7892408309823034e-05,
+      "loss": 0.5506,
+      "step": 18933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78917671197743e-05,
+      "loss": 0.4534,
+      "step": 18934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789112592972557e-05,
+      "loss": 0.5478,
+      "step": 18935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789048473967684e-05,
+      "loss": 0.4696,
+      "step": 18936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7889843549628114e-05,
+      "loss": 0.518,
+      "step": 18937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788920235957938e-05,
+      "loss": 0.44,
+      "step": 18938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788856116953065e-05,
+      "loss": 0.5461,
+      "step": 18939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788791997948192e-05,
+      "loss": 0.5342,
+      "step": 18940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788727878943319e-05,
+      "loss": 0.4091,
+      "step": 18941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788663759938446e-05,
+      "loss": 0.5164,
+      "step": 18942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788599640933573e-05,
+      "loss": 0.483,
+      "step": 18943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7885355219287e-05,
+      "loss": 0.4955,
+      "step": 18944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788471402923827e-05,
+      "loss": 0.5437,
+      "step": 18945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788407283918954e-05,
+      "loss": 0.6054,
+      "step": 18946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7883431649140804e-05,
+      "loss": 0.5791,
+      "step": 18947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7882790459092076e-05,
+      "loss": 0.4735,
+      "step": 18948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788214926904335e-05,
+      "loss": 0.4692,
+      "step": 18949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788150807899461e-05,
+      "loss": 0.4752,
+      "step": 18950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7880866888945885e-05,
+      "loss": 0.5058,
+      "step": 18951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7880225698897156e-05,
+      "loss": 0.5363,
+      "step": 18952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787958450884843e-05,
+      "loss": 0.5598,
+      "step": 18953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787894331879969e-05,
+      "loss": 0.5274,
+      "step": 18954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7878302128750965e-05,
+      "loss": 0.5268,
+      "step": 18955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787766093870223e-05,
+      "loss": 0.5828,
+      "step": 18956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78770197486535e-05,
+      "loss": 0.5099,
+      "step": 18957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787637855860477e-05,
+      "loss": 0.5465,
+      "step": 18958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787573736855604e-05,
+      "loss": 0.5163,
+      "step": 18959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7875096178507316e-05,
+      "loss": 0.5526,
+      "step": 18960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787445498845858e-05,
+      "loss": 0.5017,
+      "step": 18961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787381379840985e-05,
+      "loss": 0.4183,
+      "step": 18962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787317260836112e-05,
+      "loss": 0.4401,
+      "step": 18963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787253141831239e-05,
+      "loss": 0.5363,
+      "step": 18964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7871890228263655e-05,
+      "loss": 0.5537,
+      "step": 18965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7871249038214926e-05,
+      "loss": 0.5719,
+      "step": 18966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78706078481662e-05,
+      "loss": 0.5067,
+      "step": 18967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786996665811747e-05,
+      "loss": 0.5569,
+      "step": 18968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786932546806874e-05,
+      "loss": 0.5521,
+      "step": 18969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7868684278020006e-05,
+      "loss": 0.5607,
+      "step": 18970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786804308797128e-05,
+      "loss": 0.5298,
+      "step": 18971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786740189792254e-05,
+      "loss": 0.3796,
+      "step": 18972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7866760707873815e-05,
+      "loss": 0.5353,
+      "step": 18973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786611951782508e-05,
+      "loss": 0.4797,
+      "step": 18974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786547832777635e-05,
+      "loss": 0.5634,
+      "step": 18975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786483713772763e-05,
+      "loss": 0.5501,
+      "step": 18976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7864195947678895e-05,
+      "loss": 0.6709,
+      "step": 18977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786355475763017e-05,
+      "loss": 0.5083,
+      "step": 18978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786291356758143e-05,
+      "loss": 0.5251,
+      "step": 18979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78622723775327e-05,
+      "loss": 0.4917,
+      "step": 18980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786163118748397e-05,
+      "loss": 0.5374,
+      "step": 18981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786098999743524e-05,
+      "loss": 0.589,
+      "step": 18982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786034880738651e-05,
+      "loss": 0.538,
+      "step": 18983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7859707617337783e-05,
+      "loss": 0.54,
+      "step": 18984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7859066427289055e-05,
+      "loss": 0.4958,
+      "step": 18985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785842523724032e-05,
+      "loss": 0.5042,
+      "step": 18986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785778404719159e-05,
+      "loss": 0.5605,
+      "step": 18987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785714285714286e-05,
+      "loss": 0.4879,
+      "step": 18988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785650166709413e-05,
+      "loss": 0.5405,
+      "step": 18989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785586047704539e-05,
+      "loss": 0.4088,
+      "step": 18990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7855219286996665e-05,
+      "loss": 0.6999,
+      "step": 18991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785457809694794e-05,
+      "loss": 0.4604,
+      "step": 18992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785393690689921e-05,
+      "loss": 0.5364,
+      "step": 18993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785329571685048e-05,
+      "loss": 0.5104,
+      "step": 18994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7852654526801745e-05,
+      "loss": 0.5161,
+      "step": 18995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785201333675302e-05,
+      "loss": 0.4823,
+      "step": 18996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785137214670428e-05,
+      "loss": 0.4075,
+      "step": 18997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7850730956655554e-05,
+      "loss": 0.3757,
+      "step": 18998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7850089766606825e-05,
+      "loss": 0.4425,
+      "step": 18999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78494485765581e-05,
+      "loss": 0.4306,
+      "step": 19000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784880738650936e-05,
+      "loss": 0.5299,
+      "step": 19001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7848166196460634e-05,
+      "loss": 0.5652,
+      "step": 19002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7847525006411905e-05,
+      "loss": 0.4684,
+      "step": 19003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784688381636317e-05,
+      "loss": 0.5466,
+      "step": 19004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784624262631444e-05,
+      "loss": 0.4555,
+      "step": 19005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784560143626571e-05,
+      "loss": 0.5726,
+      "step": 19006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784496024621698e-05,
+      "loss": 0.4693,
+      "step": 19007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784431905616825e-05,
+      "loss": 0.4665,
+      "step": 19008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784367786611952e-05,
+      "loss": 0.5344,
+      "step": 19009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784303667607079e-05,
+      "loss": 0.5862,
+      "step": 19010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784239548602206e-05,
+      "loss": 0.5605,
+      "step": 19011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784175429597333e-05,
+      "loss": 0.452,
+      "step": 19012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7841113105924595e-05,
+      "loss": 0.4626,
+      "step": 19013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784047191587587e-05,
+      "loss": 0.442,
+      "step": 19014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783983072582714e-05,
+      "loss": 0.4736,
+      "step": 19015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783918953577841e-05,
+      "loss": 0.3904,
+      "step": 19016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7838548345729676e-05,
+      "loss": 0.4568,
+      "step": 19017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783790715568095e-05,
+      "loss": 0.5115,
+      "step": 19018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783726596563221e-05,
+      "loss": 0.5411,
+      "step": 19019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7836624775583484e-05,
+      "loss": 0.5109,
+      "step": 19020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7835983585534756e-05,
+      "loss": 0.5033,
+      "step": 19021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783534239548602e-05,
+      "loss": 0.4104,
+      "step": 19022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783470120543729e-05,
+      "loss": 0.4782,
+      "step": 19023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7834060015388564e-05,
+      "loss": 0.6092,
+      "step": 19024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7833418825339836e-05,
+      "loss": 0.4742,
+      "step": 19025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78327776352911e-05,
+      "loss": 0.5322,
+      "step": 19026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783213644524237e-05,
+      "loss": 0.4496,
+      "step": 19027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783149525519364e-05,
+      "loss": 0.544,
+      "step": 19028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783085406514491e-05,
+      "loss": 0.5096,
+      "step": 19029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783021287509618e-05,
+      "loss": 0.549,
+      "step": 19030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782957168504745e-05,
+      "loss": 0.5317,
+      "step": 19031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7828930494998724e-05,
+      "loss": 0.5338,
+      "step": 19032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782828930494999e-05,
+      "loss": 0.6211,
+      "step": 19033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782764811490126e-05,
+      "loss": 0.4816,
+      "step": 19034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7827006924852526e-05,
+      "loss": 0.5772,
+      "step": 19035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78263657348038e-05,
+      "loss": 0.3915,
+      "step": 19036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782572454475506e-05,
+      "loss": 0.4916,
+      "step": 19037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7825083354706334e-05,
+      "loss": 0.5828,
+      "step": 19038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7824442164657606e-05,
+      "loss": 0.475,
+      "step": 19039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782380097460888e-05,
+      "loss": 0.4754,
+      "step": 19040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782315978456015e-05,
+      "loss": 0.4516,
+      "step": 19041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7822518594511414e-05,
+      "loss": 0.4192,
+      "step": 19042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7821877404462686e-05,
+      "loss": 0.425,
+      "step": 19043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782123621441395e-05,
+      "loss": 0.4636,
+      "step": 19044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782059502436522e-05,
+      "loss": 0.5389,
+      "step": 19045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781995383431649e-05,
+      "loss": 0.4675,
+      "step": 19046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781931264426776e-05,
+      "loss": 0.4812,
+      "step": 19047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781867145421904e-05,
+      "loss": 0.3879,
+      "step": 19048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78180302641703e-05,
+      "loss": 0.4624,
+      "step": 19049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7817389074121574e-05,
+      "loss": 0.5528,
+      "step": 19050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781674788407284e-05,
+      "loss": 0.5999,
+      "step": 19051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781610669402411e-05,
+      "loss": 0.543,
+      "step": 19052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7815465503975376e-05,
+      "loss": 0.5968,
+      "step": 19053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781482431392665e-05,
+      "loss": 0.4578,
+      "step": 19054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781418312387792e-05,
+      "loss": 0.5858,
+      "step": 19055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781354193382919e-05,
+      "loss": 0.5166,
+      "step": 19056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781290074378046e-05,
+      "loss": 0.543,
+      "step": 19057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781225955373173e-05,
+      "loss": 0.4656,
+      "step": 19058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7811618363683e-05,
+      "loss": 0.4616,
+      "step": 19059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7810977173634264e-05,
+      "loss": 0.4965,
+      "step": 19060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7810335983585536e-05,
+      "loss": 0.482,
+      "step": 19061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78096947935368e-05,
+      "loss": 0.4806,
+      "step": 19062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780905360348807e-05,
+      "loss": 0.5131,
+      "step": 19063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7808412413439345e-05,
+      "loss": 0.3819,
+      "step": 19064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7807771223390616e-05,
+      "loss": 0.5698,
+      "step": 19065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780713003334189e-05,
+      "loss": 0.4502,
+      "step": 19066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780648884329315e-05,
+      "loss": 0.4632,
+      "step": 19067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7805847653244425e-05,
+      "loss": 0.5369,
+      "step": 19068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780520646319569e-05,
+      "loss": 0.5152,
+      "step": 19069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780456527314696e-05,
+      "loss": 0.5371,
+      "step": 19070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780392408309823e-05,
+      "loss": 0.5839,
+      "step": 19071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7803282893049505e-05,
+      "loss": 0.4844,
+      "step": 19072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7802641703000776e-05,
+      "loss": 0.511,
+      "step": 19073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780200051295204e-05,
+      "loss": 0.4978,
+      "step": 19074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780135932290331e-05,
+      "loss": 0.4626,
+      "step": 19075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780071813285458e-05,
+      "loss": 0.4806,
+      "step": 19076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780007694280585e-05,
+      "loss": 0.4718,
+      "step": 19077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7799435752757115e-05,
+      "loss": 0.507,
+      "step": 19078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7798794562708386e-05,
+      "loss": 0.5286,
+      "step": 19079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779815337265966e-05,
+      "loss": 0.4437,
+      "step": 19080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779751218261093e-05,
+      "loss": 0.4272,
+      "step": 19081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.77968709925622e-05,
+      "loss": 0.5408,
+      "step": 19082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7796229802513467e-05,
+      "loss": 0.5906,
+      "step": 19083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779558861246474e-05,
+      "loss": 0.5262,
+      "step": 19084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7794947422416e-05,
+      "loss": 0.5671,
+      "step": 19085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7794306232367275e-05,
+      "loss": 0.5729,
+      "step": 19086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779366504231855e-05,
+      "loss": 0.4803,
+      "step": 19087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779302385226982e-05,
+      "loss": 0.5067,
+      "step": 19088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779238266222108e-05,
+      "loss": 0.5124,
+      "step": 19089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7791741472172355e-05,
+      "loss": 0.452,
+      "step": 19090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779110028212363e-05,
+      "loss": 0.6059,
+      "step": 19091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779045909207489e-05,
+      "loss": 0.5937,
+      "step": 19092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778981790202616e-05,
+      "loss": 0.533,
+      "step": 19093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778917671197743e-05,
+      "loss": 0.4653,
+      "step": 19094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.77885355219287e-05,
+      "loss": 0.5499,
+      "step": 19095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778789433187997e-05,
+      "loss": 0.5141,
+      "step": 19096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7787253141831243e-05,
+      "loss": 0.5539,
+      "step": 19097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778661195178251e-05,
+      "loss": 0.4499,
+      "step": 19098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778597076173378e-05,
+      "loss": 0.5907,
+      "step": 19099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778532957168505e-05,
+      "loss": 0.422,
+      "step": 19100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778468838163632e-05,
+      "loss": 0.4796,
+      "step": 19101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778404719158759e-05,
+      "loss": 0.4983,
+      "step": 19102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778340600153886e-05,
+      "loss": 0.4633,
+      "step": 19103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778276481149013e-05,
+      "loss": 0.5135,
+      "step": 19104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.77821236214414e-05,
+      "loss": 0.4759,
+      "step": 19105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778148243139267e-05,
+      "loss": 0.5059,
+      "step": 19106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7780841241343934e-05,
+      "loss": 0.3842,
+      "step": 19107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7780200051295205e-05,
+      "loss": 0.5326,
+      "step": 19108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777955886124648e-05,
+      "loss": 0.4769,
+      "step": 19109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777891767119774e-05,
+      "loss": 0.4869,
+      "step": 19110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7778276481149014e-05,
+      "loss": 0.6052,
+      "step": 19111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7777635291100285e-05,
+      "loss": 0.5863,
+      "step": 19112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777699410105156e-05,
+      "loss": 0.467,
+      "step": 19113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777635291100282e-05,
+      "loss": 0.5229,
+      "step": 19114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7775711720954094e-05,
+      "loss": 0.6594,
+      "step": 19115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777507053090536e-05,
+      "loss": 0.4569,
+      "step": 19116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777442934085663e-05,
+      "loss": 0.5396,
+      "step": 19117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77737881508079e-05,
+      "loss": 0.3499,
+      "step": 19118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777314696075917e-05,
+      "loss": 0.6182,
+      "step": 19119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7772505770710445e-05,
+      "loss": 0.5539,
+      "step": 19120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777186458066171e-05,
+      "loss": 0.5432,
+      "step": 19121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777122339061298e-05,
+      "loss": 0.4894,
+      "step": 19122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777058220056425e-05,
+      "loss": 0.5941,
+      "step": 19123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776994101051552e-05,
+      "loss": 0.5403,
+      "step": 19124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7769299820466784e-05,
+      "loss": 0.5165,
+      "step": 19125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7768658630418055e-05,
+      "loss": 0.4586,
+      "step": 19126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776801744036933e-05,
+      "loss": 0.5831,
+      "step": 19127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77673762503206e-05,
+      "loss": 0.6027,
+      "step": 19128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776673506027187e-05,
+      "loss": 0.5135,
+      "step": 19129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7766093870223136e-05,
+      "loss": 0.5435,
+      "step": 19130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776545268017441e-05,
+      "loss": 0.488,
+      "step": 19131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776481149012567e-05,
+      "loss": 0.4006,
+      "step": 19132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7764170300076944e-05,
+      "loss": 0.6412,
+      "step": 19133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776352911002821e-05,
+      "loss": 0.6119,
+      "step": 19134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776288791997948e-05,
+      "loss": 0.4643,
+      "step": 19135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776224672993076e-05,
+      "loss": 0.4865,
+      "step": 19136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7761605539882024e-05,
+      "loss": 0.6329,
+      "step": 19137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7760964349833296e-05,
+      "loss": 0.5596,
+      "step": 19138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776032315978456e-05,
+      "loss": 0.5293,
+      "step": 19139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775968196973583e-05,
+      "loss": 0.5462,
+      "step": 19140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77590407796871e-05,
+      "loss": 0.5175,
+      "step": 19141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775839958963837e-05,
+      "loss": 0.4787,
+      "step": 19142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775775839958964e-05,
+      "loss": 0.4852,
+      "step": 19143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775711720954091e-05,
+      "loss": 0.4359,
+      "step": 19144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7756476019492184e-05,
+      "loss": 0.518,
+      "step": 19145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775583482944345e-05,
+      "loss": 0.5565,
+      "step": 19146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775519363939472e-05,
+      "loss": 0.5713,
+      "step": 19147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7754552449345986e-05,
+      "loss": 0.5266,
+      "step": 19148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775391125929726e-05,
+      "loss": 0.4625,
+      "step": 19149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775327006924852e-05,
+      "loss": 0.6007,
+      "step": 19150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7752628879199794e-05,
+      "loss": 0.5291,
+      "step": 19151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7751987689151066e-05,
+      "loss": 0.511,
+      "step": 19152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775134649910234e-05,
+      "loss": 0.4908,
+      "step": 19153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775070530905361e-05,
+      "loss": 0.4826,
+      "step": 19154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7750064119004874e-05,
+      "loss": 0.4462,
+      "step": 19155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7749422928956146e-05,
+      "loss": 0.5907,
+      "step": 19156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774878173890741e-05,
+      "loss": 0.5434,
+      "step": 19157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774814054885868e-05,
+      "loss": 0.5643,
+      "step": 19158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7747499358809954e-05,
+      "loss": 0.4671,
+      "step": 19159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7746858168761226e-05,
+      "loss": 0.5035,
+      "step": 19160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774621697871249e-05,
+      "loss": 0.4694,
+      "step": 19161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774557578866376e-05,
+      "loss": 0.436,
+      "step": 19162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7744934598615034e-05,
+      "loss": 0.5534,
+      "step": 19163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77442934085663e-05,
+      "loss": 0.5067,
+      "step": 19164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774365221851757e-05,
+      "loss": 0.6765,
+      "step": 19165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7743011028468836e-05,
+      "loss": 0.4933,
+      "step": 19166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774236983842011e-05,
+      "loss": 0.4582,
+      "step": 19167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774172864837138e-05,
+      "loss": 0.4888,
+      "step": 19168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774108745832265e-05,
+      "loss": 0.517,
+      "step": 19169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7740446268273916e-05,
+      "loss": 0.5691,
+      "step": 19170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773980507822519e-05,
+      "loss": 0.7017,
+      "step": 19171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773916388817646e-05,
+      "loss": 0.5781,
+      "step": 19172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7738522698127725e-05,
+      "loss": 0.5666,
+      "step": 19173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7737881508078996e-05,
+      "loss": 0.4241,
+      "step": 19174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773724031803027e-05,
+      "loss": 0.4889,
+      "step": 19175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773659912798154e-05,
+      "loss": 0.5609,
+      "step": 19176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7735957937932805e-05,
+      "loss": 0.4995,
+      "step": 19177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7735316747884076e-05,
+      "loss": 0.599,
+      "step": 19178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773467555783534e-05,
+      "loss": 0.4834,
+      "step": 19179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773403436778661e-05,
+      "loss": 0.5392,
+      "step": 19180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7733393177737885e-05,
+      "loss": 0.4959,
+      "step": 19181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773275198768915e-05,
+      "loss": 0.5263,
+      "step": 19182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773211079764042e-05,
+      "loss": 0.4795,
+      "step": 19183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773146960759169e-05,
+      "loss": 0.3704,
+      "step": 19184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7730828417542965e-05,
+      "loss": 0.5816,
+      "step": 19185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773018722749423e-05,
+      "loss": 0.4579,
+      "step": 19186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77295460374455e-05,
+      "loss": 0.5176,
+      "step": 19187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7728904847396766e-05,
+      "loss": 0.4775,
+      "step": 19188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772826365734804e-05,
+      "loss": 0.477,
+      "step": 19189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772762246729931e-05,
+      "loss": 0.5384,
+      "step": 19190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772698127725058e-05,
+      "loss": 0.5223,
+      "step": 19191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772634008720185e-05,
+      "loss": 0.6739,
+      "step": 19192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772569889715312e-05,
+      "loss": 0.4662,
+      "step": 19193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772505770710439e-05,
+      "loss": 0.5069,
+      "step": 19194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7724416517055655e-05,
+      "loss": 0.5188,
+      "step": 19195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7723775327006927e-05,
+      "loss": 0.4567,
+      "step": 19196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772313413695819e-05,
+      "loss": 0.5333,
+      "step": 19197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772249294690946e-05,
+      "loss": 0.3837,
+      "step": 19198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7721851756860735e-05,
+      "loss": 0.5404,
+      "step": 19199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772121056681201e-05,
+      "loss": 0.4799,
+      "step": 19200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772056937676328e-05,
+      "loss": 0.6078,
+      "step": 19201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771992818671454e-05,
+      "loss": 0.6024,
+      "step": 19202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7719286996665815e-05,
+      "loss": 0.6358,
+      "step": 19203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771864580661708e-05,
+      "loss": 0.6626,
+      "step": 19204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771800461656835e-05,
+      "loss": 0.4061,
+      "step": 19205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771736342651962e-05,
+      "loss": 0.4676,
+      "step": 19206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771672223647089e-05,
+      "loss": 0.5361,
+      "step": 19207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771608104642217e-05,
+      "loss": 0.4125,
+      "step": 19208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771543985637343e-05,
+      "loss": 0.5258,
+      "step": 19209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7714798666324703e-05,
+      "loss": 0.45,
+      "step": 19210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771415747627597e-05,
+      "loss": 0.5985,
+      "step": 19211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771351628622724e-05,
+      "loss": 0.4933,
+      "step": 19212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7712875096178505e-05,
+      "loss": 0.464,
+      "step": 19213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771223390612978e-05,
+      "loss": 0.5025,
+      "step": 19214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771159271608105e-05,
+      "loss": 0.5688,
+      "step": 19215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771095152603232e-05,
+      "loss": 0.5059,
+      "step": 19216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771031033598359e-05,
+      "loss": 0.6218,
+      "step": 19217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770966914593486e-05,
+      "loss": 0.492,
+      "step": 19218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770902795588613e-05,
+      "loss": 0.47,
+      "step": 19219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7708386765837394e-05,
+      "loss": 0.5265,
+      "step": 19220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7707745575788665e-05,
+      "loss": 0.5527,
+      "step": 19221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770710438573993e-05,
+      "loss": 0.5042,
+      "step": 19222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77064631956912e-05,
+      "loss": 0.5369,
+      "step": 19223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7705822005642474e-05,
+      "loss": 0.544,
+      "step": 19224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7705180815593745e-05,
+      "loss": 0.5559,
+      "step": 19225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770453962554502e-05,
+      "loss": 0.5804,
+      "step": 19226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770389843549628e-05,
+      "loss": 0.5314,
+      "step": 19227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7703257245447554e-05,
+      "loss": 0.5974,
+      "step": 19228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770261605539882e-05,
+      "loss": 0.5161,
+      "step": 19229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770197486535009e-05,
+      "loss": 0.5623,
+      "step": 19230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770133367530136e-05,
+      "loss": 0.5065,
+      "step": 19231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7700692485252634e-05,
+      "loss": 0.6101,
+      "step": 19232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77000512952039e-05,
+      "loss": 0.6745,
+      "step": 19233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769941010515517e-05,
+      "loss": 0.5648,
+      "step": 19234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769876891510644e-05,
+      "loss": 0.4478,
+      "step": 19235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769812772505771e-05,
+      "loss": 0.564,
+      "step": 19236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769748653500898e-05,
+      "loss": 0.4685,
+      "step": 19237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7696845344960244e-05,
+      "loss": 0.4764,
+      "step": 19238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7696204154911516e-05,
+      "loss": 0.5416,
+      "step": 19239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769556296486279e-05,
+      "loss": 0.476,
+      "step": 19240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769492177481406e-05,
+      "loss": 0.4252,
+      "step": 19241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7694280584765324e-05,
+      "loss": 0.5373,
+      "step": 19242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7693639394716596e-05,
+      "loss": 0.4463,
+      "step": 19243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769299820466787e-05,
+      "loss": 0.5738,
+      "step": 19244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769235701461913e-05,
+      "loss": 0.5723,
+      "step": 19245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7691715824570404e-05,
+      "loss": 0.5296,
+      "step": 19246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7691074634521676e-05,
+      "loss": 0.4327,
+      "step": 19247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.769043344447295e-05,
+      "loss": 0.5223,
+      "step": 19248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768979225442421e-05,
+      "loss": 0.4879,
+      "step": 19249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7689151064375484e-05,
+      "loss": 0.5432,
+      "step": 19250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768850987432675e-05,
+      "loss": 0.4894,
+      "step": 19251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768786868427802e-05,
+      "loss": 0.4939,
+      "step": 19252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768722749422929e-05,
+      "loss": 0.3784,
+      "step": 19253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768658630418056e-05,
+      "loss": 0.5384,
+      "step": 19254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768594511413183e-05,
+      "loss": 0.4784,
+      "step": 19255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76853039240831e-05,
+      "loss": 0.5041,
+      "step": 19256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768466273403437e-05,
+      "loss": 0.5017,
+      "step": 19257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768402154398564e-05,
+      "loss": 0.4271,
+      "step": 19258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768338035393691e-05,
+      "loss": 0.4868,
+      "step": 19259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7682739163888174e-05,
+      "loss": 0.4995,
+      "step": 19260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7682097973839446e-05,
+      "loss": 0.4523,
+      "step": 19261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768145678379072e-05,
+      "loss": 0.546,
+      "step": 19262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768081559374199e-05,
+      "loss": 0.6548,
+      "step": 19263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768017440369326e-05,
+      "loss": 0.5019,
+      "step": 19264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7679533213644526e-05,
+      "loss": 0.5038,
+      "step": 19265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76788920235958e-05,
+      "loss": 0.5923,
+      "step": 19266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767825083354706e-05,
+      "loss": 0.5268,
+      "step": 19267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7677609643498334e-05,
+      "loss": 0.5913,
+      "step": 19268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7676968453449606e-05,
+      "loss": 0.5849,
+      "step": 19269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767632726340087e-05,
+      "loss": 0.4914,
+      "step": 19270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767568607335214e-05,
+      "loss": 0.5581,
+      "step": 19271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7675044883303414e-05,
+      "loss": 0.4063,
+      "step": 19272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7674403693254686e-05,
+      "loss": 0.3955,
+      "step": 19273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767376250320595e-05,
+      "loss": 0.4776,
+      "step": 19274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767312131315722e-05,
+      "loss": 0.5222,
+      "step": 19275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767248012310849e-05,
+      "loss": 0.5115,
+      "step": 19276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767183893305976e-05,
+      "loss": 0.4767,
+      "step": 19277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767119774301103e-05,
+      "loss": 0.5087,
+      "step": 19278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76705565529623e-05,
+      "loss": 0.5348,
+      "step": 19279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7669915362913575e-05,
+      "loss": 0.5076,
+      "step": 19280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766927417286484e-05,
+      "loss": 0.5973,
+      "step": 19281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766863298281611e-05,
+      "loss": 0.3936,
+      "step": 19282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7667991792767376e-05,
+      "loss": 0.5584,
+      "step": 19283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766735060271865e-05,
+      "loss": 0.3799,
+      "step": 19284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766670941266991e-05,
+      "loss": 0.6108,
+      "step": 19285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7666068222621185e-05,
+      "loss": 0.4763,
+      "step": 19286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7665427032572456e-05,
+      "loss": 0.5223,
+      "step": 19287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766478584252373e-05,
+      "loss": 0.4574,
+      "step": 19288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7664144652475e-05,
+      "loss": 0.4832,
+      "step": 19289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7663503462426265e-05,
+      "loss": 0.6252,
+      "step": 19290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7662862272377536e-05,
+      "loss": 0.5224,
+      "step": 19291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76622210823288e-05,
+      "loss": 0.5391,
+      "step": 19292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766157989228007e-05,
+      "loss": 0.5674,
+      "step": 19293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766093870223134e-05,
+      "loss": 0.4699,
+      "step": 19294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766029751218261e-05,
+      "loss": 0.6206,
+      "step": 19295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765965632213389e-05,
+      "loss": 0.5023,
+      "step": 19296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765901513208515e-05,
+      "loss": 0.5383,
+      "step": 19297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7658373942036425e-05,
+      "loss": 0.5233,
+      "step": 19298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765773275198769e-05,
+      "loss": 0.5165,
+      "step": 19299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765709156193896e-05,
+      "loss": 0.5411,
+      "step": 19300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7656450371890226e-05,
+      "loss": 0.4835,
+      "step": 19301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76558091818415e-05,
+      "loss": 0.6979,
+      "step": 19302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765516799179277e-05,
+      "loss": 0.521,
+      "step": 19303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765452680174404e-05,
+      "loss": 0.5112,
+      "step": 19304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765388561169531e-05,
+      "loss": 0.5015,
+      "step": 19305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765324442164658e-05,
+      "loss": 0.4739,
+      "step": 19306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765260323159785e-05,
+      "loss": 0.4097,
+      "step": 19307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7651962041549115e-05,
+      "loss": 0.3941,
+      "step": 19308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7651320851500387e-05,
+      "loss": 0.4053,
+      "step": 19309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765067966145165e-05,
+      "loss": 0.4857,
+      "step": 19310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.765003847140292e-05,
+      "loss": 0.443,
+      "step": 19311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7649397281354195e-05,
+      "loss": 0.5567,
+      "step": 19312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764875609130547e-05,
+      "loss": 0.6004,
+      "step": 19313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764811490125674e-05,
+      "loss": 0.548,
+      "step": 19314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7647473711208e-05,
+      "loss": 0.4892,
+      "step": 19315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7646832521159275e-05,
+      "loss": 0.5737,
+      "step": 19316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764619133111054e-05,
+      "loss": 0.5425,
+      "step": 19317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764555014106181e-05,
+      "loss": 0.5141,
+      "step": 19318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7644908951013083e-05,
+      "loss": 0.5288,
+      "step": 19319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7644267760964355e-05,
+      "loss": 0.4437,
+      "step": 19320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764362657091562e-05,
+      "loss": 0.412,
+      "step": 19321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764298538086689e-05,
+      "loss": 0.4522,
+      "step": 19322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7642344190818164e-05,
+      "loss": 0.6571,
+      "step": 19323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.764170300076943e-05,
+      "loss": 0.4319,
+      "step": 19324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76410618107207e-05,
+      "loss": 0.5942,
+      "step": 19325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7640420620671965e-05,
+      "loss": 0.506,
+      "step": 19326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763977943062324e-05,
+      "loss": 0.5052,
+      "step": 19327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763913824057451e-05,
+      "loss": 0.4645,
+      "step": 19328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763849705052578e-05,
+      "loss": 0.508,
+      "step": 19329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7637855860477045e-05,
+      "loss": 0.579,
+      "step": 19330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763721467042832e-05,
+      "loss": 0.5826,
+      "step": 19331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763657348037959e-05,
+      "loss": 0.5413,
+      "step": 19332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7635932290330854e-05,
+      "loss": 0.5523,
+      "step": 19333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7635291100282125e-05,
+      "loss": 0.5035,
+      "step": 19334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76346499102334e-05,
+      "loss": 0.5812,
+      "step": 19335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763400872018467e-05,
+      "loss": 0.4179,
+      "step": 19336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7633367530135934e-05,
+      "loss": 0.4671,
+      "step": 19337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7632726340087205e-05,
+      "loss": 0.5562,
+      "step": 19338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763208515003847e-05,
+      "loss": 0.3817,
+      "step": 19339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763144395998974e-05,
+      "loss": 0.51,
+      "step": 19340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7630802769941014e-05,
+      "loss": 0.5562,
+      "step": 19341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763016157989228e-05,
+      "loss": 0.6611,
+      "step": 19342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762952038984355e-05,
+      "loss": 0.5043,
+      "step": 19343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762887919979482e-05,
+      "loss": 0.4008,
+      "step": 19344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7628238009746094e-05,
+      "loss": 0.5917,
+      "step": 19345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762759681969736e-05,
+      "loss": 0.5047,
+      "step": 19346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762695562964863e-05,
+      "loss": 0.5519,
+      "step": 19347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7626314439599895e-05,
+      "loss": 0.5548,
+      "step": 19348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762567324955117e-05,
+      "loss": 0.5015,
+      "step": 19349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762503205950244e-05,
+      "loss": 0.4958,
+      "step": 19350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762439086945371e-05,
+      "loss": 0.4978,
+      "step": 19351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762374967940498e-05,
+      "loss": 0.5845,
+      "step": 19352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762310848935625e-05,
+      "loss": 0.471,
+      "step": 19353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762246729930752e-05,
+      "loss": 0.592,
+      "step": 19354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7621826109258784e-05,
+      "loss": 0.4892,
+      "step": 19355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7621184919210056e-05,
+      "loss": 0.487,
+      "step": 19356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762054372916132e-05,
+      "loss": 0.5321,
+      "step": 19357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761990253911259e-05,
+      "loss": 0.615,
+      "step": 19358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7619261349063864e-05,
+      "loss": 0.5081,
+      "step": 19359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7618620159015136e-05,
+      "loss": 0.5184,
+      "step": 19360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761797896896641e-05,
+      "loss": 0.5208,
+      "step": 19361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761733777891767e-05,
+      "loss": 0.4596,
+      "step": 19362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7616696588868944e-05,
+      "loss": 0.5623,
+      "step": 19363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761605539882021e-05,
+      "loss": 0.481,
+      "step": 19364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761541420877148e-05,
+      "loss": 0.6175,
+      "step": 19365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7614773018722746e-05,
+      "loss": 0.4431,
+      "step": 19366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761413182867402e-05,
+      "loss": 0.5083,
+      "step": 19367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7613490638625296e-05,
+      "loss": 0.4284,
+      "step": 19368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761284944857656e-05,
+      "loss": 0.4727,
+      "step": 19369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761220825852783e-05,
+      "loss": 0.5921,
+      "step": 19370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76115670684791e-05,
+      "loss": 0.5091,
+      "step": 19371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761092587843037e-05,
+      "loss": 0.4179,
+      "step": 19372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7610284688381634e-05,
+      "loss": 0.4983,
+      "step": 19373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7609643498332906e-05,
+      "loss": 0.4445,
+      "step": 19374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.760900230828418e-05,
+      "loss": 0.4718,
+      "step": 19375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.760836111823545e-05,
+      "loss": 0.4751,
+      "step": 19376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.760771992818672e-05,
+      "loss": 0.3614,
+      "step": 19377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7607078738137986e-05,
+      "loss": 0.5128,
+      "step": 19378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760643754808926e-05,
+      "loss": 0.4614,
+      "step": 19379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760579635804052e-05,
+      "loss": 0.3975,
+      "step": 19380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7605155167991794e-05,
+      "loss": 0.4803,
+      "step": 19381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760451397794306e-05,
+      "loss": 0.6398,
+      "step": 19382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760387278789433e-05,
+      "loss": 0.4697,
+      "step": 19383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.76032315978456e-05,
+      "loss": 0.5509,
+      "step": 19384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7602590407796874e-05,
+      "loss": 0.4563,
+      "step": 19385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7601949217748146e-05,
+      "loss": 0.5602,
+      "step": 19386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760130802769941e-05,
+      "loss": 0.4228,
+      "step": 19387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760066683765068e-05,
+      "loss": 0.4647,
+      "step": 19388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760002564760195e-05,
+      "loss": 0.6664,
+      "step": 19389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759938445755322e-05,
+      "loss": 0.5587,
+      "step": 19390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759874326750449e-05,
+      "loss": 0.556,
+      "step": 19391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759810207745576e-05,
+      "loss": 0.4837,
+      "step": 19392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759746088740703e-05,
+      "loss": 0.5679,
+      "step": 19393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75968196973583e-05,
+      "loss": 0.4788,
+      "step": 19394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759617850730957e-05,
+      "loss": 0.5323,
+      "step": 19395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7595537317260836e-05,
+      "loss": 0.5635,
+      "step": 19396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759489612721211e-05,
+      "loss": 0.5326,
+      "step": 19397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759425493716337e-05,
+      "loss": 0.4347,
+      "step": 19398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7593613747114645e-05,
+      "loss": 0.4792,
+      "step": 19399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7592972557065916e-05,
+      "loss": 0.5861,
+      "step": 19400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759233136701719e-05,
+      "loss": 0.5137,
+      "step": 19401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759169017696845e-05,
+      "loss": 0.5011,
+      "step": 19402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7591048986919725e-05,
+      "loss": 0.5354,
+      "step": 19403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7590407796870996e-05,
+      "loss": 0.4732,
+      "step": 19404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758976660682226e-05,
+      "loss": 0.4378,
+      "step": 19405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758912541677353e-05,
+      "loss": 0.4819,
+      "step": 19406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7588484226724805e-05,
+      "loss": 0.4981,
+      "step": 19407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7587843036676076e-05,
+      "loss": 0.5894,
+      "step": 19408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758720184662734e-05,
+      "loss": 0.4437,
+      "step": 19409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758656065657861e-05,
+      "loss": 0.6012,
+      "step": 19410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758591946652988e-05,
+      "loss": 0.5601,
+      "step": 19411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758527827648115e-05,
+      "loss": 0.4778,
+      "step": 19412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758463708643242e-05,
+      "loss": 0.5642,
+      "step": 19413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7583995896383686e-05,
+      "loss": 0.4828,
+      "step": 19414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758335470633496e-05,
+      "loss": 0.4693,
+      "step": 19415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758271351628623e-05,
+      "loss": 0.453,
+      "step": 19416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75820723262375e-05,
+      "loss": 0.6208,
+      "step": 19417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7581431136188767e-05,
+      "loss": 0.5681,
+      "step": 19418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758078994614004e-05,
+      "loss": 0.5904,
+      "step": 19419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75801487560913e-05,
+      "loss": 0.5301,
+      "step": 19420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7579507566042575e-05,
+      "loss": 0.5534,
+      "step": 19421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757886637599385e-05,
+      "loss": 0.4267,
+      "step": 19422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757822518594512e-05,
+      "loss": 0.5444,
+      "step": 19423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757758399589639e-05,
+      "loss": 0.4439,
+      "step": 19424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7576942805847655e-05,
+      "loss": 0.507,
+      "step": 19425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757630161579893e-05,
+      "loss": 0.5465,
+      "step": 19426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757566042575019e-05,
+      "loss": 0.4959,
+      "step": 19427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757501923570146e-05,
+      "loss": 0.4969,
+      "step": 19428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757437804565273e-05,
+      "loss": 0.5736,
+      "step": 19429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7573736855604e-05,
+      "loss": 0.5732,
+      "step": 19430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757309566555527e-05,
+      "loss": 0.533,
+      "step": 19431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7572454475506543e-05,
+      "loss": 0.5732,
+      "step": 19432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7571813285457815e-05,
+      "loss": 0.4137,
+      "step": 19433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757117209540908e-05,
+      "loss": 0.4572,
+      "step": 19434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757053090536035e-05,
+      "loss": 0.5398,
+      "step": 19435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756988971531162e-05,
+      "loss": 0.5186,
+      "step": 19436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756924852526289e-05,
+      "loss": 0.5884,
+      "step": 19437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7568607335214153e-05,
+      "loss": 0.4737,
+      "step": 19438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756796614516543e-05,
+      "loss": 0.5293,
+      "step": 19439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7567324955116704e-05,
+      "loss": 0.4696,
+      "step": 19440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756668376506797e-05,
+      "loss": 0.5432,
+      "step": 19441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756604257501924e-05,
+      "loss": 0.5013,
+      "step": 19442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7565401384970505e-05,
+      "loss": 0.5637,
+      "step": 19443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756476019492178e-05,
+      "loss": 0.6656,
+      "step": 19444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756411900487304e-05,
+      "loss": 0.4017,
+      "step": 19445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7563477814824314e-05,
+      "loss": 0.4694,
+      "step": 19446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7562836624775585e-05,
+      "loss": 0.6461,
+      "step": 19447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756219543472686e-05,
+      "loss": 0.5159,
+      "step": 19448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756155424467813e-05,
+      "loss": 0.502,
+      "step": 19449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7560913054629394e-05,
+      "loss": 0.5283,
+      "step": 19450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7560271864580665e-05,
+      "loss": 0.5237,
+      "step": 19451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755963067453193e-05,
+      "loss": 0.5616,
+      "step": 19452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75589894844832e-05,
+      "loss": 0.5957,
+      "step": 19453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755834829443447e-05,
+      "loss": 0.4518,
+      "step": 19454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755770710438574e-05,
+      "loss": 0.4107,
+      "step": 19455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755706591433702e-05,
+      "loss": 0.5225,
+      "step": 19456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755642472428828e-05,
+      "loss": 0.5282,
+      "step": 19457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7555783534239554e-05,
+      "loss": 0.5081,
+      "step": 19458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755514234419082e-05,
+      "loss": 0.4523,
+      "step": 19459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755450115414209e-05,
+      "loss": 0.5195,
+      "step": 19460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7553859964093355e-05,
+      "loss": 0.5387,
+      "step": 19461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755321877404463e-05,
+      "loss": 0.5319,
+      "step": 19462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75525775839959e-05,
+      "loss": 0.6581,
+      "step": 19463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755193639394717e-05,
+      "loss": 0.5363,
+      "step": 19464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755129520389844e-05,
+      "loss": 0.6356,
+      "step": 19465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755065401384971e-05,
+      "loss": 0.6205,
+      "step": 19466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755001282380098e-05,
+      "loss": 0.5632,
+      "step": 19467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7549371633752244e-05,
+      "loss": 0.4769,
+      "step": 19468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7548730443703516e-05,
+      "loss": 0.5159,
+      "step": 19469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754808925365478e-05,
+      "loss": 0.5867,
+      "step": 19470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754744806360605e-05,
+      "loss": 0.5241,
+      "step": 19471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7546806873557324e-05,
+      "loss": 0.5914,
+      "step": 19472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7546165683508596e-05,
+      "loss": 0.472,
+      "step": 19473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754552449345987e-05,
+      "loss": 0.4947,
+      "step": 19474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754488330341113e-05,
+      "loss": 0.54,
+      "step": 19475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7544242113362404e-05,
+      "loss": 0.4701,
+      "step": 19476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754360092331367e-05,
+      "loss": 0.4873,
+      "step": 19477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754295973326494e-05,
+      "loss": 0.4646,
+      "step": 19478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754231854321621e-05,
+      "loss": 0.4704,
+      "step": 19479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7541677353167484e-05,
+      "loss": 0.4369,
+      "step": 19480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754103616311875e-05,
+      "loss": 0.5874,
+      "step": 19481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754039497307002e-05,
+      "loss": 0.557,
+      "step": 19482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753975378302129e-05,
+      "loss": 0.6754,
+      "step": 19483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753911259297256e-05,
+      "loss": 0.543,
+      "step": 19484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753847140292383e-05,
+      "loss": 0.4881,
+      "step": 19485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7537830212875094e-05,
+      "loss": 0.5251,
+      "step": 19486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7537189022826366e-05,
+      "loss": 0.509,
+      "step": 19487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753654783277764e-05,
+      "loss": 0.4032,
+      "step": 19488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753590664272891e-05,
+      "loss": 0.5277,
+      "step": 19489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7535265452680174e-05,
+      "loss": 0.5371,
+      "step": 19490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7534624262631446e-05,
+      "loss": 0.4595,
+      "step": 19491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753398307258272e-05,
+      "loss": 0.479,
+      "step": 19492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753334188253398e-05,
+      "loss": 0.5136,
+      "step": 19493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7532700692485254e-05,
+      "loss": 0.5234,
+      "step": 19494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7532059502436526e-05,
+      "loss": 0.4835,
+      "step": 19495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75314183123878e-05,
+      "loss": 0.4884,
+      "step": 19496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753077712233906e-05,
+      "loss": 0.4793,
+      "step": 19497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7530135932290334e-05,
+      "loss": 0.4601,
+      "step": 19498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75294947422416e-05,
+      "loss": 0.4126,
+      "step": 19499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752885355219287e-05,
+      "loss": 0.5577,
+      "step": 19500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752821236214414e-05,
+      "loss": 0.5647,
+      "step": 19501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752757117209541e-05,
+      "loss": 0.5845,
+      "step": 19502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752692998204668e-05,
+      "loss": 0.4226,
+      "step": 19503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752628879199795e-05,
+      "loss": 0.629,
+      "step": 19504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752564760194922e-05,
+      "loss": 0.6077,
+      "step": 19505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752500641190049e-05,
+      "loss": 0.5266,
+      "step": 19506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752436522185176e-05,
+      "loss": 0.5539,
+      "step": 19507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7523724031803025e-05,
+      "loss": 0.5561,
+      "step": 19508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7523082841754296e-05,
+      "loss": 0.5152,
+      "step": 19509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752244165170557e-05,
+      "loss": 0.5365,
+      "step": 19510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752180046165684e-05,
+      "loss": 0.5734,
+      "step": 19511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752115927160811e-05,
+      "loss": 0.4506,
+      "step": 19512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7520518081559376e-05,
+      "loss": 0.5658,
+      "step": 19513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751987689151065e-05,
+      "loss": 0.569,
+      "step": 19514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751923570146191e-05,
+      "loss": 0.5103,
+      "step": 19515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7518594511413185e-05,
+      "loss": 0.5685,
+      "step": 19516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751795332136445e-05,
+      "loss": 0.4776,
+      "step": 19517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751731213131572e-05,
+      "loss": 0.5406,
+      "step": 19518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751667094126699e-05,
+      "loss": 0.4722,
+      "step": 19519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7516029751218265e-05,
+      "loss": 0.5364,
+      "step": 19520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7515388561169537e-05,
+      "loss": 0.4906,
+      "step": 19521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75147473711208e-05,
+      "loss": 0.4181,
+      "step": 19522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751410618107207e-05,
+      "loss": 0.5734,
+      "step": 19523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751346499102334e-05,
+      "loss": 0.5179,
+      "step": 19524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751282380097461e-05,
+      "loss": 0.5331,
+      "step": 19525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7512182610925875e-05,
+      "loss": 0.5635,
+      "step": 19526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7511541420877146e-05,
+      "loss": 0.521,
+      "step": 19527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7510900230828425e-05,
+      "loss": 0.5534,
+      "step": 19528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751025904077969e-05,
+      "loss": 0.4278,
+      "step": 19529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750961785073096e-05,
+      "loss": 0.5294,
+      "step": 19530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7508976660682227e-05,
+      "loss": 0.4944,
+      "step": 19531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75083354706335e-05,
+      "loss": 0.4963,
+      "step": 19532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750769428058476e-05,
+      "loss": 0.5936,
+      "step": 19533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7507053090536035e-05,
+      "loss": 0.5432,
+      "step": 19534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750641190048731e-05,
+      "loss": 0.5088,
+      "step": 19535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750577071043858e-05,
+      "loss": 0.4608,
+      "step": 19536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750512952038985e-05,
+      "loss": 0.4938,
+      "step": 19537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7504488330341115e-05,
+      "loss": 0.4391,
+      "step": 19538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750384714029239e-05,
+      "loss": 0.4591,
+      "step": 19539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750320595024365e-05,
+      "loss": 0.434,
+      "step": 19540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7502564760194923e-05,
+      "loss": 0.5299,
+      "step": 19541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750192357014619e-05,
+      "loss": 0.526,
+      "step": 19542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750128238009746e-05,
+      "loss": 0.5013,
+      "step": 19543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750064119004873e-05,
+      "loss": 0.4705,
+      "step": 19544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.5892,
+      "step": 19545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7499358809951275e-05,
+      "loss": 0.4808,
+      "step": 19546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749871761990254e-05,
+      "loss": 0.6248,
+      "step": 19547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749807642985381e-05,
+      "loss": 0.6032,
+      "step": 19548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749743523980508e-05,
+      "loss": 0.438,
+      "step": 19549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749679404975635e-05,
+      "loss": 0.4366,
+      "step": 19550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749615285970762e-05,
+      "loss": 0.5428,
+      "step": 19551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749551166965889e-05,
+      "loss": 0.6726,
+      "step": 19552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749487047961016e-05,
+      "loss": 0.5066,
+      "step": 19553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749422928956143e-05,
+      "loss": 0.5568,
+      "step": 19554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74935880995127e-05,
+      "loss": 0.4748,
+      "step": 19555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7492946909463965e-05,
+      "loss": 0.6059,
+      "step": 19556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749230571941524e-05,
+      "loss": 0.5162,
+      "step": 19557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74916645293665e-05,
+      "loss": 0.5557,
+      "step": 19558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7491023339317774e-05,
+      "loss": 0.443,
+      "step": 19559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7490382149269045e-05,
+      "loss": 0.4163,
+      "step": 19560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748974095922032e-05,
+      "loss": 0.4154,
+      "step": 19561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748909976917158e-05,
+      "loss": 0.5099,
+      "step": 19562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7488458579122854e-05,
+      "loss": 0.5482,
+      "step": 19563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7487817389074125e-05,
+      "loss": 0.4865,
+      "step": 19564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748717619902539e-05,
+      "loss": 0.4581,
+      "step": 19565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748653500897666e-05,
+      "loss": 0.5711,
+      "step": 19566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7485893818927934e-05,
+      "loss": 0.4975,
+      "step": 19567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7485252628879206e-05,
+      "loss": 0.4589,
+      "step": 19568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748461143883047e-05,
+      "loss": 0.4225,
+      "step": 19569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748397024878174e-05,
+      "loss": 0.4602,
+      "step": 19570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748332905873301e-05,
+      "loss": 0.5015,
+      "step": 19571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748268786868428e-05,
+      "loss": 0.5911,
+      "step": 19572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748204667863555e-05,
+      "loss": 0.5568,
+      "step": 19573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7481405488586816e-05,
+      "loss": 0.5378,
+      "step": 19574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748076429853809e-05,
+      "loss": 0.4502,
+      "step": 19575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748012310848936e-05,
+      "loss": 0.54,
+      "step": 19576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747948191844063e-05,
+      "loss": 0.544,
+      "step": 19577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7478840728391896e-05,
+      "loss": 0.4503,
+      "step": 19578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747819953834317e-05,
+      "loss": 0.5919,
+      "step": 19579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747755834829443e-05,
+      "loss": 0.49,
+      "step": 19580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7476917158245704e-05,
+      "loss": 0.5311,
+      "step": 19581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7476275968196976e-05,
+      "loss": 0.4682,
+      "step": 19582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747563477814825e-05,
+      "loss": 0.612,
+      "step": 19583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747499358809952e-05,
+      "loss": 0.4381,
+      "step": 19584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7474352398050784e-05,
+      "loss": 0.5068,
+      "step": 19585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7473711208002056e-05,
+      "loss": 0.5241,
+      "step": 19586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747307001795332e-05,
+      "loss": 0.5863,
+      "step": 19587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747242882790459e-05,
+      "loss": 0.4157,
+      "step": 19588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747178763785586e-05,
+      "loss": 0.4245,
+      "step": 19589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.747114644780713e-05,
+      "loss": 0.4511,
+      "step": 19590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74705052577584e-05,
+      "loss": 0.566,
+      "step": 19591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746986406770967e-05,
+      "loss": 0.4542,
+      "step": 19592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7469222877660944e-05,
+      "loss": 0.5298,
+      "step": 19593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746858168761221e-05,
+      "loss": 0.468,
+      "step": 19594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746794049756348e-05,
+      "loss": 0.5437,
+      "step": 19595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7467299307514746e-05,
+      "loss": 0.495,
+      "step": 19596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746665811746602e-05,
+      "loss": 0.5798,
+      "step": 19597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746601692741728e-05,
+      "loss": 0.5509,
+      "step": 19598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746537573736856e-05,
+      "loss": 0.5149,
+      "step": 19599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746473454731983e-05,
+      "loss": 0.3982,
+      "step": 19600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74640933572711e-05,
+      "loss": 0.4404,
+      "step": 19601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746345216722237e-05,
+      "loss": 0.4684,
+      "step": 19602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7462810977173634e-05,
+      "loss": 0.4922,
+      "step": 19603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7462169787124906e-05,
+      "loss": 0.4262,
+      "step": 19604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746152859707617e-05,
+      "loss": 0.6152,
+      "step": 19605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746088740702744e-05,
+      "loss": 0.5033,
+      "step": 19606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7460246216978714e-05,
+      "loss": 0.47,
+      "step": 19607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7459605026929986e-05,
+      "loss": 0.4746,
+      "step": 19608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745896383688126e-05,
+      "loss": 0.4193,
+      "step": 19609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745832264683252e-05,
+      "loss": 0.5787,
+      "step": 19610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7457681456783794e-05,
+      "loss": 0.558,
+      "step": 19611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745704026673506e-05,
+      "loss": 0.4703,
+      "step": 19612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745639907668633e-05,
+      "loss": 0.4492,
+      "step": 19613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7455757886637596e-05,
+      "loss": 0.4493,
+      "step": 19614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745511669658887e-05,
+      "loss": 0.5464,
+      "step": 19615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745447550654014e-05,
+      "loss": 0.51,
+      "step": 19616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745383431649141e-05,
+      "loss": 0.5881,
+      "step": 19617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745319312644268e-05,
+      "loss": 0.5752,
+      "step": 19618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745255193639395e-05,
+      "loss": 0.4536,
+      "step": 19619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745191074634522e-05,
+      "loss": 0.5312,
+      "step": 19620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7451269556296485e-05,
+      "loss": 0.4849,
+      "step": 19621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7450628366247756e-05,
+      "loss": 0.5927,
+      "step": 19622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744998717619903e-05,
+      "loss": 0.4108,
+      "step": 19623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74493459861503e-05,
+      "loss": 0.6615,
+      "step": 19624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7448704796101565e-05,
+      "loss": 0.4911,
+      "step": 19625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7448063606052836e-05,
+      "loss": 0.395,
+      "step": 19626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744742241600411e-05,
+      "loss": 0.491,
+      "step": 19627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744678122595537e-05,
+      "loss": 0.6032,
+      "step": 19628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7446140035906645e-05,
+      "loss": 0.5618,
+      "step": 19629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744549884585791e-05,
+      "loss": 0.5056,
+      "step": 19630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744485765580918e-05,
+      "loss": 0.5801,
+      "step": 19631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744421646576045e-05,
+      "loss": 0.4237,
+      "step": 19632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7443575275711725e-05,
+      "loss": 0.4612,
+      "step": 19633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744293408566299e-05,
+      "loss": 0.504,
+      "step": 19634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744229289561426e-05,
+      "loss": 0.5889,
+      "step": 19635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744165170556553e-05,
+      "loss": 0.5193,
+      "step": 19636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74410105155168e-05,
+      "loss": 0.4299,
+      "step": 19637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.744036932546807e-05,
+      "loss": 0.5095,
+      "step": 19638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743972813541934e-05,
+      "loss": 0.5316,
+      "step": 19639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743908694537061e-05,
+      "loss": 0.6207,
+      "step": 19640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743844575532188e-05,
+      "loss": 0.4458,
+      "step": 19641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743780456527315e-05,
+      "loss": 0.5189,
+      "step": 19642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743716337522442e-05,
+      "loss": 0.575,
+      "step": 19643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7436522185175687e-05,
+      "loss": 0.5432,
+      "step": 19644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743588099512696e-05,
+      "loss": 0.4771,
+      "step": 19645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743523980507822e-05,
+      "loss": 0.5634,
+      "step": 19646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7434598615029495e-05,
+      "loss": 0.4664,
+      "step": 19647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743395742498077e-05,
+      "loss": 0.4048,
+      "step": 19648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743331623493204e-05,
+      "loss": 0.4932,
+      "step": 19649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.74326750448833e-05,
+      "loss": 0.5574,
+      "step": 19650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7432033854834575e-05,
+      "loss": 0.5701,
+      "step": 19651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743139266478585e-05,
+      "loss": 0.4362,
+      "step": 19652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743075147473711e-05,
+      "loss": 0.4587,
+      "step": 19653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7430110284688383e-05,
+      "loss": 0.6597,
+      "step": 19654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7429469094639655e-05,
+      "loss": 0.4645,
+      "step": 19655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742882790459093e-05,
+      "loss": 0.5143,
+      "step": 19656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742818671454219e-05,
+      "loss": 0.5305,
+      "step": 19657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7427545524493464e-05,
+      "loss": 0.4723,
+      "step": 19658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742690433444473e-05,
+      "loss": 0.5453,
+      "step": 19659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7426263144396e-05,
+      "loss": 0.6165,
+      "step": 19660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742562195434727e-05,
+      "loss": 0.5081,
+      "step": 19661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742498076429854e-05,
+      "loss": 0.516,
+      "step": 19662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742433957424981e-05,
+      "loss": 0.6186,
+      "step": 19663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742369838420108e-05,
+      "loss": 0.5242,
+      "step": 19664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742305719415235e-05,
+      "loss": 0.5611,
+      "step": 19665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742241600410362e-05,
+      "loss": 0.4102,
+      "step": 19666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742177481405489e-05,
+      "loss": 0.6507,
+      "step": 19667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7421133624006154e-05,
+      "loss": 0.5047,
+      "step": 19668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7420492433957425e-05,
+      "loss": 0.4658,
+      "step": 19669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.74198512439087e-05,
+      "loss": 0.5989,
+      "step": 19670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741921005385997e-05,
+      "loss": 0.5245,
+      "step": 19671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741856886381124e-05,
+      "loss": 0.532,
+      "step": 19672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7417927673762505e-05,
+      "loss": 0.5521,
+      "step": 19673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741728648371378e-05,
+      "loss": 0.5279,
+      "step": 19674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741664529366504e-05,
+      "loss": 0.568,
+      "step": 19675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7416004103616314e-05,
+      "loss": 0.5192,
+      "step": 19676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741536291356758e-05,
+      "loss": 0.5082,
+      "step": 19677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741472172351885e-05,
+      "loss": 0.4844,
+      "step": 19678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741408053347012e-05,
+      "loss": 0.5143,
+      "step": 19679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7413439343421394e-05,
+      "loss": 0.4519,
+      "step": 19680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7412798153372666e-05,
+      "loss": 0.4397,
+      "step": 19681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741215696332393e-05,
+      "loss": 0.5847,
+      "step": 19682
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.74115157732752e-05,
+      "loss": 0.5387,
+      "step": 19683
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741087458322647e-05,
+      "loss": 0.5976,
+      "step": 19684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741023339317774e-05,
+      "loss": 0.4192,
+      "step": 19685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7409592203129004e-05,
+      "loss": 0.4918,
+      "step": 19686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740895101308028e-05,
+      "loss": 0.5389,
+      "step": 19687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7408309823031554e-05,
+      "loss": 0.5591,
+      "step": 19688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740766863298282e-05,
+      "loss": 0.4564,
+      "step": 19689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740702744293409e-05,
+      "loss": 0.4407,
+      "step": 19690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7406386252885356e-05,
+      "loss": 0.4477,
+      "step": 19691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740574506283663e-05,
+      "loss": 0.5441,
+      "step": 19692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740510387278789e-05,
+      "loss": 0.4914,
+      "step": 19693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7404462682739164e-05,
+      "loss": 0.4048,
+      "step": 19694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7403821492690436e-05,
+      "loss": 0.4777,
+      "step": 19695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740318030264171e-05,
+      "loss": 0.5253,
+      "step": 19696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740253911259298e-05,
+      "loss": 0.4838,
+      "step": 19697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7401897922544244e-05,
+      "loss": 0.4442,
+      "step": 19698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7401256732495516e-05,
+      "loss": 0.5363,
+      "step": 19699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740061554244678e-05,
+      "loss": 0.5075,
+      "step": 19700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739997435239805e-05,
+      "loss": 0.43,
+      "step": 19701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739933316234932e-05,
+      "loss": 0.4709,
+      "step": 19702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739869197230059e-05,
+      "loss": 0.4098,
+      "step": 19703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739805078225186e-05,
+      "loss": 0.571,
+      "step": 19704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739740959220313e-05,
+      "loss": 0.5873,
+      "step": 19705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7396768402154404e-05,
+      "loss": 0.4522,
+      "step": 19706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739612721210567e-05,
+      "loss": 0.6624,
+      "step": 19707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739548602205694e-05,
+      "loss": 0.4467,
+      "step": 19708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7394844832008206e-05,
+      "loss": 0.445,
+      "step": 19709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739420364195948e-05,
+      "loss": 0.5665,
+      "step": 19710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739356245191075e-05,
+      "loss": 0.4652,
+      "step": 19711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739292126186202e-05,
+      "loss": 0.5818,
+      "step": 19712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7392280071813286e-05,
+      "loss": 0.492,
+      "step": 19713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739163888176456e-05,
+      "loss": 0.4284,
+      "step": 19714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739099769171583e-05,
+      "loss": 0.5325,
+      "step": 19715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7390356501667094e-05,
+      "loss": 0.3885,
+      "step": 19716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7389715311618366e-05,
+      "loss": 0.544,
+      "step": 19717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738907412156963e-05,
+      "loss": 0.4031,
+      "step": 19718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73884329315209e-05,
+      "loss": 0.4801,
+      "step": 19719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7387791741472174e-05,
+      "loss": 0.427,
+      "step": 19720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7387150551423446e-05,
+      "loss": 0.641,
+      "step": 19721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738650936137471e-05,
+      "loss": 0.4441,
+      "step": 19722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738586817132598e-05,
+      "loss": 0.4787,
+      "step": 19723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7385226981277255e-05,
+      "loss": 0.6555,
+      "step": 19724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738458579122852e-05,
+      "loss": 0.4904,
+      "step": 19725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738394460117979e-05,
+      "loss": 0.5052,
+      "step": 19726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738330341113106e-05,
+      "loss": 0.5374,
+      "step": 19727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7382662221082335e-05,
+      "loss": 0.4933,
+      "step": 19728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73820210310336e-05,
+      "loss": 0.5587,
+      "step": 19729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738137984098487e-05,
+      "loss": 0.4702,
+      "step": 19730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7380738650936136e-05,
+      "loss": 0.5145,
+      "step": 19731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738009746088741e-05,
+      "loss": 0.481,
+      "step": 19732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737945627083868e-05,
+      "loss": 0.511,
+      "step": 19733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7378815080789945e-05,
+      "loss": 0.5609,
+      "step": 19734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7378173890741216e-05,
+      "loss": 0.497,
+      "step": 19735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737753270069249e-05,
+      "loss": 0.4145,
+      "step": 19736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737689151064376e-05,
+      "loss": 0.3814,
+      "step": 19737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7376250320595025e-05,
+      "loss": 0.5568,
+      "step": 19738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7375609130546296e-05,
+      "loss": 0.4815,
+      "step": 19739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737496794049756e-05,
+      "loss": 0.5695,
+      "step": 19740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737432675044883e-05,
+      "loss": 0.5137,
+      "step": 19741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7373685560400105e-05,
+      "loss": 0.496,
+      "step": 19742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7373044370351376e-05,
+      "loss": 0.4244,
+      "step": 19743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737240318030265e-05,
+      "loss": 0.5355,
+      "step": 19744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737176199025391e-05,
+      "loss": 0.4232,
+      "step": 19745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7371120800205185e-05,
+      "loss": 0.4914,
+      "step": 19746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737047961015645e-05,
+      "loss": 0.5079,
+      "step": 19747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736983842010772e-05,
+      "loss": 0.4998,
+      "step": 19748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7369197230058986e-05,
+      "loss": 0.4776,
+      "step": 19749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736855604001026e-05,
+      "loss": 0.439,
+      "step": 19750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736791484996153e-05,
+      "loss": 0.5784,
+      "step": 19751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73672736599128e-05,
+      "loss": 0.4119,
+      "step": 19752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736663246986407e-05,
+      "loss": 0.3826,
+      "step": 19753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736599127981534e-05,
+      "loss": 0.4433,
+      "step": 19754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736535008976661e-05,
+      "loss": 0.6012,
+      "step": 19755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7364708899717875e-05,
+      "loss": 0.4036,
+      "step": 19756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736406770966915e-05,
+      "loss": 0.4244,
+      "step": 19757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736342651962041e-05,
+      "loss": 0.47,
+      "step": 19758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736278532957169e-05,
+      "loss": 0.4912,
+      "step": 19759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736214413952296e-05,
+      "loss": 0.4456,
+      "step": 19760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736150294947423e-05,
+      "loss": 0.4872,
+      "step": 19761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73608617594255e-05,
+      "loss": 0.5274,
+      "step": 19762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736022056937676e-05,
+      "loss": 0.3844,
+      "step": 19763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7359579379328035e-05,
+      "loss": 0.6617,
+      "step": 19764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73589381892793e-05,
+      "loss": 0.6832,
+      "step": 19765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735829699923057e-05,
+      "loss": 0.5646,
+      "step": 19766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7357655809181843e-05,
+      "loss": 0.4467,
+      "step": 19767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7357014619133115e-05,
+      "loss": 0.5008,
+      "step": 19768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735637342908439e-05,
+      "loss": 0.5734,
+      "step": 19769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735573223903565e-05,
+      "loss": 0.4635,
+      "step": 19770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7355091048986924e-05,
+      "loss": 0.6942,
+      "step": 19771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735444985893819e-05,
+      "loss": 0.5551,
+      "step": 19772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735380866888946e-05,
+      "loss": 0.49,
+      "step": 19773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7353167478840725e-05,
+      "loss": 0.5156,
+      "step": 19774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7352526288792e-05,
+      "loss": 0.5283,
+      "step": 19775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735188509874327e-05,
+      "loss": 0.4052,
+      "step": 19776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735124390869454e-05,
+      "loss": 0.5264,
+      "step": 19777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735060271864581e-05,
+      "loss": 0.5046,
+      "step": 19778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734996152859708e-05,
+      "loss": 0.501,
+      "step": 19779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734932033854835e-05,
+      "loss": 0.4878,
+      "step": 19780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7348679148499614e-05,
+      "loss": 0.6089,
+      "step": 19781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7348037958450885e-05,
+      "loss": 0.5883,
+      "step": 19782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734739676840216e-05,
+      "loss": 0.5254,
+      "step": 19783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734675557835343e-05,
+      "loss": 0.5556,
+      "step": 19784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7346114388304694e-05,
+      "loss": 0.5637,
+      "step": 19785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7345473198255965e-05,
+      "loss": 0.4875,
+      "step": 19786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734483200820724e-05,
+      "loss": 0.4675,
+      "step": 19787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73441908181585e-05,
+      "loss": 0.5692,
+      "step": 19788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7343549628109774e-05,
+      "loss": 0.5221,
+      "step": 19789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734290843806104e-05,
+      "loss": 0.5346,
+      "step": 19790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734226724801231e-05,
+      "loss": 0.505,
+      "step": 19791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734162605796358e-05,
+      "loss": 0.5564,
+      "step": 19792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7340984867914854e-05,
+      "loss": 0.5676,
+      "step": 19793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734034367786612e-05,
+      "loss": 0.4982,
+      "step": 19794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733970248781739e-05,
+      "loss": 0.4465,
+      "step": 19795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733906129776866e-05,
+      "loss": 0.522,
+      "step": 19796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733842010771993e-05,
+      "loss": 0.4064,
+      "step": 19797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73377789176712e-05,
+      "loss": 0.4554,
+      "step": 19798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733713772762247e-05,
+      "loss": 0.4515,
+      "step": 19799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733649653757374e-05,
+      "loss": 0.5619,
+      "step": 19800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733585534752501e-05,
+      "loss": 0.4552,
+      "step": 19801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733521415747628e-05,
+      "loss": 0.5517,
+      "step": 19802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7334572967427544e-05,
+      "loss": 0.4076,
+      "step": 19803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7333931777378816e-05,
+      "loss": 0.4823,
+      "step": 19804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733329058733009e-05,
+      "loss": 0.6618,
+      "step": 19805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733264939728135e-05,
+      "loss": 0.5465,
+      "step": 19806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7332008207232624e-05,
+      "loss": 0.5078,
+      "step": 19807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7331367017183896e-05,
+      "loss": 0.4651,
+      "step": 19808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733072582713517e-05,
+      "loss": 0.4374,
+      "step": 19809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733008463708643e-05,
+      "loss": 0.4718,
+      "step": 19810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7329443447037704e-05,
+      "loss": 0.4922,
+      "step": 19811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732880225698897e-05,
+      "loss": 0.5275,
+      "step": 19812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732816106694024e-05,
+      "loss": 0.5339,
+      "step": 19813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732751987689151e-05,
+      "loss": 0.4949,
+      "step": 19814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7326878686842784e-05,
+      "loss": 0.4618,
+      "step": 19815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7326237496794056e-05,
+      "loss": 0.4834,
+      "step": 19816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732559630674532e-05,
+      "loss": 0.4582,
+      "step": 19817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732495511669659e-05,
+      "loss": 0.4368,
+      "step": 19818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732431392664786e-05,
+      "loss": 0.5448,
+      "step": 19819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732367273659913e-05,
+      "loss": 0.4125,
+      "step": 19820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7323031546550394e-05,
+      "loss": 0.5046,
+      "step": 19821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7322390356501666e-05,
+      "loss": 0.5361,
+      "step": 19822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732174916645294e-05,
+      "loss": 0.4488,
+      "step": 19823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732110797640421e-05,
+      "loss": 0.4967,
+      "step": 19824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732046678635548e-05,
+      "loss": 0.5332,
+      "step": 19825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7319825596306746e-05,
+      "loss": 0.4465,
+      "step": 19826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731918440625802e-05,
+      "loss": 0.3846,
+      "step": 19827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731854321620928e-05,
+      "loss": 0.4724,
+      "step": 19828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7317902026160554e-05,
+      "loss": 0.4581,
+      "step": 19829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7317260836111826e-05,
+      "loss": 0.5,
+      "step": 19830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73166196460631e-05,
+      "loss": 0.4556,
+      "step": 19831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731597845601437e-05,
+      "loss": 0.4512,
+      "step": 19832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7315337265965634e-05,
+      "loss": 0.4452,
+      "step": 19833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7314696075916906e-05,
+      "loss": 0.4974,
+      "step": 19834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731405488586817e-05,
+      "loss": 0.5558,
+      "step": 19835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731341369581944e-05,
+      "loss": 0.4338,
+      "step": 19836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731277250577071e-05,
+      "loss": 0.468,
+      "step": 19837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731213131572198e-05,
+      "loss": 0.489,
+      "step": 19838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731149012567325e-05,
+      "loss": 0.5111,
+      "step": 19839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731084893562452e-05,
+      "loss": 0.5364,
+      "step": 19840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7310207745575795e-05,
+      "loss": 0.5547,
+      "step": 19841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730956655552706e-05,
+      "loss": 0.4433,
+      "step": 19842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730892536547833e-05,
+      "loss": 0.5384,
+      "step": 19843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7308284175429596e-05,
+      "loss": 0.4531,
+      "step": 19844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730764298538087e-05,
+      "loss": 0.5682,
+      "step": 19845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730700179533213e-05,
+      "loss": 0.621,
+      "step": 19846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730636060528341e-05,
+      "loss": 0.4833,
+      "step": 19847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730571941523468e-05,
+      "loss": 0.52,
+      "step": 19848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730507822518595e-05,
+      "loss": 0.6008,
+      "step": 19849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730443703513722e-05,
+      "loss": 0.5094,
+      "step": 19850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7303795845088485e-05,
+      "loss": 0.5911,
+      "step": 19851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7303154655039756e-05,
+      "loss": 0.4915,
+      "step": 19852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730251346499102e-05,
+      "loss": 0.4299,
+      "step": 19853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730187227494229e-05,
+      "loss": 0.396,
+      "step": 19854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7301231084893565e-05,
+      "loss": 0.6205,
+      "step": 19855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7300589894844837e-05,
+      "loss": 0.4358,
+      "step": 19856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729994870479611e-05,
+      "loss": 0.4117,
+      "step": 19857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729930751474737e-05,
+      "loss": 0.4538,
+      "step": 19858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7298666324698645e-05,
+      "loss": 0.4962,
+      "step": 19859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729802513464991e-05,
+      "loss": 0.445,
+      "step": 19860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729738394460118e-05,
+      "loss": 0.4788,
+      "step": 19861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7296742754552446e-05,
+      "loss": 0.4873,
+      "step": 19862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729610156450372e-05,
+      "loss": 0.425,
+      "step": 19863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729546037445499e-05,
+      "loss": 0.4475,
+      "step": 19864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729481918440626e-05,
+      "loss": 0.5916,
+      "step": 19865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729417799435753e-05,
+      "loss": 0.3873,
+      "step": 19866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.72935368043088e-05,
+      "loss": 0.5662,
+      "step": 19867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729289561426007e-05,
+      "loss": 0.5504,
+      "step": 19868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7292254424211335e-05,
+      "loss": 0.566,
+      "step": 19869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729161323416261e-05,
+      "loss": 0.581,
+      "step": 19870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729097204411388e-05,
+      "loss": 0.5868,
+      "step": 19871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729033085406515e-05,
+      "loss": 0.4349,
+      "step": 19872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7289689664016415e-05,
+      "loss": 0.3933,
+      "step": 19873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728904847396769e-05,
+      "loss": 0.4434,
+      "step": 19874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728840728391896e-05,
+      "loss": 0.5533,
+      "step": 19875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7287766093870223e-05,
+      "loss": 0.4229,
+      "step": 19876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7287124903821495e-05,
+      "loss": 0.5056,
+      "step": 19877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728648371377276e-05,
+      "loss": 0.5789,
+      "step": 19878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728584252372403e-05,
+      "loss": 0.5694,
+      "step": 19879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7285201333675303e-05,
+      "loss": 0.5586,
+      "step": 19880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7284560143626575e-05,
+      "loss": 0.555,
+      "step": 19881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728391895357784e-05,
+      "loss": 0.484,
+      "step": 19882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728327776352911e-05,
+      "loss": 0.4545,
+      "step": 19883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7282636573480384e-05,
+      "loss": 0.5024,
+      "step": 19884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728199538343165e-05,
+      "loss": 0.5486,
+      "step": 19885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728135419338292e-05,
+      "loss": 0.5239,
+      "step": 19886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728071300333419e-05,
+      "loss": 0.4512,
+      "step": 19887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7280071813285464e-05,
+      "loss": 0.3787,
+      "step": 19888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.727943062323673e-05,
+      "loss": 0.5471,
+      "step": 19889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7278789433188e-05,
+      "loss": 0.5514,
+      "step": 19890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7278148243139265e-05,
+      "loss": 0.4267,
+      "step": 19891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.727750705309054e-05,
+      "loss": 0.5441,
+      "step": 19892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.727686586304181e-05,
+      "loss": 0.4776,
+      "step": 19893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7276224672993074e-05,
+      "loss": 0.5344,
+      "step": 19894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7275583482944345e-05,
+      "loss": 0.5306,
+      "step": 19895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.727494229289562e-05,
+      "loss": 0.4536,
+      "step": 19896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.727430110284689e-05,
+      "loss": 0.4787,
+      "step": 19897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7273659912798154e-05,
+      "loss": 0.451,
+      "step": 19898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7273018722749425e-05,
+      "loss": 0.575,
+      "step": 19899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.727237753270069e-05,
+      "loss": 0.5525,
+      "step": 19900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.727173634265196e-05,
+      "loss": 0.4533,
+      "step": 19901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7271095152603234e-05,
+      "loss": 0.4863,
+      "step": 19902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7270453962554506e-05,
+      "loss": 0.4289,
+      "step": 19903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726981277250578e-05,
+      "loss": 0.5596,
+      "step": 19904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726917158245704e-05,
+      "loss": 0.5665,
+      "step": 19905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7268530392408314e-05,
+      "loss": 0.5164,
+      "step": 19906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726788920235958e-05,
+      "loss": 0.5026,
+      "step": 19907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726724801231085e-05,
+      "loss": 0.5725,
+      "step": 19908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7266606822262116e-05,
+      "loss": 0.5348,
+      "step": 19909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726596563221339e-05,
+      "loss": 0.4885,
+      "step": 19910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726532444216466e-05,
+      "loss": 0.525,
+      "step": 19911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726468325211593e-05,
+      "loss": 0.5719,
+      "step": 19912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72640420620672e-05,
+      "loss": 0.4539,
+      "step": 19913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726340087201847e-05,
+      "loss": 0.4927,
+      "step": 19914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726275968196974e-05,
+      "loss": 0.5326,
+      "step": 19915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7262118491921004e-05,
+      "loss": 0.532,
+      "step": 19916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7261477301872276e-05,
+      "loss": 0.5071,
+      "step": 19917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726083611182354e-05,
+      "loss": 0.53,
+      "step": 19918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726019492177482e-05,
+      "loss": 0.5208,
+      "step": 19919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725955373172609e-05,
+      "loss": 0.4805,
+      "step": 19920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7258912541677356e-05,
+      "loss": 0.4775,
+      "step": 19921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725827135162863e-05,
+      "loss": 0.5998,
+      "step": 19922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725763016157989e-05,
+      "loss": 0.4767,
+      "step": 19923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7256988971531164e-05,
+      "loss": 0.507,
+      "step": 19924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725634778148243e-05,
+      "loss": 0.5176,
+      "step": 19925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72557065914337e-05,
+      "loss": 0.5889,
+      "step": 19926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725506540138497e-05,
+      "loss": 0.5137,
+      "step": 19927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7254424211336244e-05,
+      "loss": 0.4986,
+      "step": 19928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7253783021287516e-05,
+      "loss": 0.4496,
+      "step": 19929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725314183123878e-05,
+      "loss": 0.5669,
+      "step": 19930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725250064119005e-05,
+      "loss": 0.5296,
+      "step": 19931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725185945114132e-05,
+      "loss": 0.3692,
+      "step": 19932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725121826109259e-05,
+      "loss": 0.4278,
+      "step": 19933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7250577071043854e-05,
+      "loss": 0.4403,
+      "step": 19934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7249935880995126e-05,
+      "loss": 0.4052,
+      "step": 19935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72492946909464e-05,
+      "loss": 0.512,
+      "step": 19936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724865350089767e-05,
+      "loss": 0.4674,
+      "step": 19937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724801231084894e-05,
+      "loss": 0.4454,
+      "step": 19938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7247371120800206e-05,
+      "loss": 0.4164,
+      "step": 19939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724672993075148e-05,
+      "loss": 0.5336,
+      "step": 19940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724608874070274e-05,
+      "loss": 0.5744,
+      "step": 19941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7245447550654014e-05,
+      "loss": 0.5346,
+      "step": 19942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7244806360605286e-05,
+      "loss": 0.4888,
+      "step": 19943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724416517055656e-05,
+      "loss": 0.4365,
+      "step": 19944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724352398050782e-05,
+      "loss": 0.5186,
+      "step": 19945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7242882790459094e-05,
+      "loss": 0.3841,
+      "step": 19946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7242241600410366e-05,
+      "loss": 0.547,
+      "step": 19947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724160041036163e-05,
+      "loss": 0.4208,
+      "step": 19948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72409592203129e-05,
+      "loss": 0.6177,
+      "step": 19949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724031803026417e-05,
+      "loss": 0.5155,
+      "step": 19950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723967684021544e-05,
+      "loss": 0.4248,
+      "step": 19951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723903565016671e-05,
+      "loss": 0.624,
+      "step": 19952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723839446011798e-05,
+      "loss": 0.4127,
+      "step": 19953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723775327006925e-05,
+      "loss": 0.4909,
+      "step": 19954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723711208002052e-05,
+      "loss": 0.4906,
+      "step": 19955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723647088997179e-05,
+      "loss": 0.4928,
+      "step": 19956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7235829699923056e-05,
+      "loss": 0.3846,
+      "step": 19957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723518850987433e-05,
+      "loss": 0.4423,
+      "step": 19958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72345473198256e-05,
+      "loss": 0.5094,
+      "step": 19959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723390612977687e-05,
+      "loss": 0.3993,
+      "step": 19960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7233264939728136e-05,
+      "loss": 0.6069,
+      "step": 19961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723262374967941e-05,
+      "loss": 0.5289,
+      "step": 19962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723198255963067e-05,
+      "loss": 0.5269,
+      "step": 19963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7231341369581945e-05,
+      "loss": 0.5788,
+      "step": 19964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7230700179533216e-05,
+      "loss": 0.5572,
+      "step": 19965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723005898948448e-05,
+      "loss": 0.6157,
+      "step": 19966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722941779943575e-05,
+      "loss": 0.5183,
+      "step": 19967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7228776609387025e-05,
+      "loss": 0.4886,
+      "step": 19968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7228135419338297e-05,
+      "loss": 0.6488,
+      "step": 19969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722749422928956e-05,
+      "loss": 0.418,
+      "step": 19970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722685303924083e-05,
+      "loss": 0.4962,
+      "step": 19971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72262118491921e-05,
+      "loss": 0.4568,
+      "step": 19972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722557065914337e-05,
+      "loss": 0.4945,
+      "step": 19973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722492946909464e-05,
+      "loss": 0.515,
+      "step": 19974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722428827904591e-05,
+      "loss": 0.5032,
+      "step": 19975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7223647088997185e-05,
+      "loss": 0.4068,
+      "step": 19976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722300589894845e-05,
+      "loss": 0.565,
+      "step": 19977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722236470889972e-05,
+      "loss": 0.6558,
+      "step": 19978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7221723518850987e-05,
+      "loss": 0.5732,
+      "step": 19979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722108232880226e-05,
+      "loss": 0.5087,
+      "step": 19980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722044113875352e-05,
+      "loss": 0.4566,
+      "step": 19981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7219799948704795e-05,
+      "loss": 0.4701,
+      "step": 19982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721915875865607e-05,
+      "loss": 0.4825,
+      "step": 19983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721851756860734e-05,
+      "loss": 0.5302,
+      "step": 19984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721787637855861e-05,
+      "loss": 0.5416,
+      "step": 19985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7217235188509875e-05,
+      "loss": 0.5723,
+      "step": 19986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721659399846115e-05,
+      "loss": 0.4677,
+      "step": 19987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721595280841241e-05,
+      "loss": 0.5722,
+      "step": 19988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7215311618363683e-05,
+      "loss": 0.6627,
+      "step": 19989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721467042831495e-05,
+      "loss": 0.601,
+      "step": 19990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721402923826623e-05,
+      "loss": 0.4721,
+      "step": 19991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72133880482175e-05,
+      "loss": 0.5487,
+      "step": 19992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7212746858168764e-05,
+      "loss": 0.4497,
+      "step": 19993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7212105668120035e-05,
+      "loss": 0.4212,
+      "step": 19994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72114644780713e-05,
+      "loss": 0.5466,
+      "step": 19995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721082328802257e-05,
+      "loss": 0.5169,
+      "step": 19996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721018209797384e-05,
+      "loss": 0.5285,
+      "step": 19997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720954090792511e-05,
+      "loss": 0.5086,
+      "step": 19998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720889971787638e-05,
+      "loss": 0.541,
+      "step": 19999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720825852782765e-05,
+      "loss": 0.5193,
+      "step": 20000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7207617337778924e-05,
+      "loss": 0.5951,
+      "step": 20001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720697614773019e-05,
+      "loss": 0.5549,
+      "step": 20002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720633495768146e-05,
+      "loss": 0.564,
+      "step": 20003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7205693767632725e-05,
+      "loss": 0.557,
+      "step": 20004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7205052577584e-05,
+      "loss": 0.4955,
+      "step": 20005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720441138753526e-05,
+      "loss": 0.4952,
+      "step": 20006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720377019748654e-05,
+      "loss": 0.5043,
+      "step": 20007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7203129007437805e-05,
+      "loss": 0.6426,
+      "step": 20008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720248781738908e-05,
+      "loss": 0.459,
+      "step": 20009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720184662734035e-05,
+      "loss": 0.4796,
+      "step": 20010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7201205437291614e-05,
+      "loss": 0.4894,
+      "step": 20011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7200564247242885e-05,
+      "loss": 0.5129,
+      "step": 20012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719992305719415e-05,
+      "loss": 0.4606,
+      "step": 20013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719928186714542e-05,
+      "loss": 0.5633,
+      "step": 20014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7198640677096694e-05,
+      "loss": 0.5825,
+      "step": 20015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7197999487047966e-05,
+      "loss": 0.5078,
+      "step": 20016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719735829699923e-05,
+      "loss": 0.5226,
+      "step": 20017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71967171069505e-05,
+      "loss": 0.5327,
+      "step": 20018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7196075916901774e-05,
+      "loss": 0.386,
+      "step": 20019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719543472685304e-05,
+      "loss": 0.5438,
+      "step": 20020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719479353680431e-05,
+      "loss": 0.5493,
+      "step": 20021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7194152346755576e-05,
+      "loss": 0.4648,
+      "step": 20022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719351115670685e-05,
+      "loss": 0.5078,
+      "step": 20023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719286996665812e-05,
+      "loss": 0.4553,
+      "step": 20024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719222877660939e-05,
+      "loss": 0.4768,
+      "step": 20025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719158758656066e-05,
+      "loss": 0.5444,
+      "step": 20026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719094639651193e-05,
+      "loss": 0.4913,
+      "step": 20027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71903052064632e-05,
+      "loss": 0.4435,
+      "step": 20028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7189664016414464e-05,
+      "loss": 0.508,
+      "step": 20029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7189022826365736e-05,
+      "loss": 0.5573,
+      "step": 20030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718838163631701e-05,
+      "loss": 0.4706,
+      "step": 20031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718774044626828e-05,
+      "loss": 0.5957,
+      "step": 20032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7187099256219544e-05,
+      "loss": 0.4888,
+      "step": 20033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7186458066170816e-05,
+      "loss": 0.5932,
+      "step": 20034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718581687612209e-05,
+      "loss": 0.4279,
+      "step": 20035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718517568607335e-05,
+      "loss": 0.5442,
+      "step": 20036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7184534496024624e-05,
+      "loss": 0.3398,
+      "step": 20037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718389330597589e-05,
+      "loss": 0.5084,
+      "step": 20038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718325211592716e-05,
+      "loss": 0.4044,
+      "step": 20039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718261092587843e-05,
+      "loss": 0.5552,
+      "step": 20040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7181969735829704e-05,
+      "loss": 0.4619,
+      "step": 20041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718132854578097e-05,
+      "loss": 0.4113,
+      "step": 20042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718068735573224e-05,
+      "loss": 0.484,
+      "step": 20043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718004616568351e-05,
+      "loss": 0.3618,
+      "step": 20044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717940497563478e-05,
+      "loss": 0.324,
+      "step": 20045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717876378558605e-05,
+      "loss": 0.5341,
+      "step": 20046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717812259553732e-05,
+      "loss": 0.5326,
+      "step": 20047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717748140548859e-05,
+      "loss": 0.4871,
+      "step": 20048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717684021543986e-05,
+      "loss": 0.4301,
+      "step": 20049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717619902539113e-05,
+      "loss": 0.645,
+      "step": 20050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7175557835342394e-05,
+      "loss": 0.588,
+      "step": 20051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7174916645293666e-05,
+      "loss": 0.5251,
+      "step": 20052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717427545524494e-05,
+      "loss": 0.479,
+      "step": 20053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71736342651962e-05,
+      "loss": 0.3765,
+      "step": 20054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7172993075147474e-05,
+      "loss": 0.5242,
+      "step": 20055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7172351885098746e-05,
+      "loss": 0.5987,
+      "step": 20056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717171069505002e-05,
+      "loss": 0.5915,
+      "step": 20057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717106950500128e-05,
+      "loss": 0.5296,
+      "step": 20058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7170428314952555e-05,
+      "loss": 0.5208,
+      "step": 20059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716978712490382e-05,
+      "loss": 0.5162,
+      "step": 20060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716914593485509e-05,
+      "loss": 0.5079,
+      "step": 20061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716850474480636e-05,
+      "loss": 0.5643,
+      "step": 20062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7167863554757635e-05,
+      "loss": 0.5741,
+      "step": 20063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7167222364708906e-05,
+      "loss": 0.5754,
+      "step": 20064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716658117466017e-05,
+      "loss": 0.6107,
+      "step": 20065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716593998461144e-05,
+      "loss": 0.5219,
+      "step": 20066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716529879456271e-05,
+      "loss": 0.5009,
+      "step": 20067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716465760451398e-05,
+      "loss": 0.5387,
+      "step": 20068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7164016414465245e-05,
+      "loss": 0.5045,
+      "step": 20069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7163375224416516e-05,
+      "loss": 0.5016,
+      "step": 20070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716273403436779e-05,
+      "loss": 0.6426,
+      "step": 20071
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716209284431906e-05,
+      "loss": 0.4761,
+      "step": 20072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716145165427033e-05,
+      "loss": 0.5416,
+      "step": 20073
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7160810464221596e-05,
+      "loss": 0.5294,
+      "step": 20074
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716016927417287e-05,
+      "loss": 0.4387,
+      "step": 20075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715952808412413e-05,
+      "loss": 0.5698,
+      "step": 20076
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7158886894075405e-05,
+      "loss": 0.512,
+      "step": 20077
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715824570402667e-05,
+      "loss": 0.5518,
+      "step": 20078
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715760451397795e-05,
+      "loss": 0.4542,
+      "step": 20079
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715696332392922e-05,
+      "loss": 0.4229,
+      "step": 20080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7156322133880485e-05,
+      "loss": 0.502,
+      "step": 20081
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7155680943831757e-05,
+      "loss": 0.4218,
+      "step": 20082
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715503975378302e-05,
+      "loss": 0.5636,
+      "step": 20083
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715439856373429e-05,
+      "loss": 0.4507,
+      "step": 20084
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715375737368556e-05,
+      "loss": 0.4931,
+      "step": 20085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715311618363683e-05,
+      "loss": 0.5863,
+      "step": 20086
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71524749935881e-05,
+      "loss": 0.5919,
+      "step": 20087
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715183380353937e-05,
+      "loss": 0.5502,
+      "step": 20088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7151192613490645e-05,
+      "loss": 0.4708,
+      "step": 20089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715055142344191e-05,
+      "loss": 0.4963,
+      "step": 20090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714991023339318e-05,
+      "loss": 0.5443,
+      "step": 20091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714926904334445e-05,
+      "loss": 0.491,
+      "step": 20092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714862785329572e-05,
+      "loss": 0.4633,
+      "step": 20093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714798666324698e-05,
+      "loss": 0.3824,
+      "step": 20094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714734547319826e-05,
+      "loss": 0.4856,
+      "step": 20095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714670428314953e-05,
+      "loss": 0.5295,
+      "step": 20096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71460630931008e-05,
+      "loss": 0.5131,
+      "step": 20097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714542190305207e-05,
+      "loss": 0.5777,
+      "step": 20098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7144780713003335e-05,
+      "loss": 0.4558,
+      "step": 20099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714413952295461e-05,
+      "loss": 0.3795,
+      "step": 20100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714349833290587e-05,
+      "loss": 0.4497,
+      "step": 20101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7142857142857143e-05,
+      "loss": 0.4713,
+      "step": 20102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7142215952808415e-05,
+      "loss": 0.5463,
+      "step": 20103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714157476275969e-05,
+      "loss": 0.5548,
+      "step": 20104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.714093357271095e-05,
+      "loss": 0.428,
+      "step": 20105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7140292382662224e-05,
+      "loss": 0.4982,
+      "step": 20106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7139651192613495e-05,
+      "loss": 0.5586,
+      "step": 20107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713901000256476e-05,
+      "loss": 0.5256,
+      "step": 20108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713836881251603e-05,
+      "loss": 0.4383,
+      "step": 20109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71377276224673e-05,
+      "loss": 0.532,
+      "step": 20110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713708643241857e-05,
+      "loss": 0.4655,
+      "step": 20111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713644524236984e-05,
+      "loss": 0.4134,
+      "step": 20112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713580405232111e-05,
+      "loss": 0.4961,
+      "step": 20113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713516286227238e-05,
+      "loss": 0.5117,
+      "step": 20114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713452167222365e-05,
+      "loss": 0.5153,
+      "step": 20115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713388048217492e-05,
+      "loss": 0.578,
+      "step": 20116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7133239292126185e-05,
+      "loss": 0.5208,
+      "step": 20117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713259810207746e-05,
+      "loss": 0.5443,
+      "step": 20118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713195691202873e-05,
+      "loss": 0.5259,
+      "step": 20119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713131572198e-05,
+      "loss": 0.4069,
+      "step": 20120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7130674531931265e-05,
+      "loss": 0.4717,
+      "step": 20121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713003334188254e-05,
+      "loss": 0.466,
+      "step": 20122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71293921518338e-05,
+      "loss": 0.4455,
+      "step": 20123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7128750961785074e-05,
+      "loss": 0.5922,
+      "step": 20124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7128109771736346e-05,
+      "loss": 0.6085,
+      "step": 20125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712746858168761e-05,
+      "loss": 0.51,
+      "step": 20126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712682739163888e-05,
+      "loss": 0.4633,
+      "step": 20127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7126186201590154e-05,
+      "loss": 0.5321,
+      "step": 20128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7125545011541426e-05,
+      "loss": 0.4338,
+      "step": 20129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712490382149269e-05,
+      "loss": 0.4209,
+      "step": 20130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712426263144396e-05,
+      "loss": 0.6276,
+      "step": 20131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712362144139523e-05,
+      "loss": 0.4405,
+      "step": 20132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71229802513465e-05,
+      "loss": 0.4317,
+      "step": 20133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712233906129777e-05,
+      "loss": 0.4837,
+      "step": 20134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712169787124904e-05,
+      "loss": 0.4043,
+      "step": 20135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7121056681200314e-05,
+      "loss": 0.4773,
+      "step": 20136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.712041549115158e-05,
+      "loss": 0.4605,
+      "step": 20137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711977430110285e-05,
+      "loss": 0.4119,
+      "step": 20138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7119133111054116e-05,
+      "loss": 0.5491,
+      "step": 20139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711849192100539e-05,
+      "loss": 0.4581,
+      "step": 20140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711785073095665e-05,
+      "loss": 0.4964,
+      "step": 20141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7117209540907924e-05,
+      "loss": 0.5649,
+      "step": 20142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7116568350859196e-05,
+      "loss": 0.4212,
+      "step": 20143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711592716081047e-05,
+      "loss": 0.4517,
+      "step": 20144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711528597076174e-05,
+      "loss": 0.4815,
+      "step": 20145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7114644780713004e-05,
+      "loss": 0.5501,
+      "step": 20146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7114003590664276e-05,
+      "loss": 0.5461,
+      "step": 20147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711336240061554e-05,
+      "loss": 0.361,
+      "step": 20148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711272121056681e-05,
+      "loss": 0.4065,
+      "step": 20149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711208002051808e-05,
+      "loss": 0.4941,
+      "step": 20150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7111438830469356e-05,
+      "loss": 0.5833,
+      "step": 20151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711079764042063e-05,
+      "loss": 0.5519,
+      "step": 20152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711015645037189e-05,
+      "loss": 0.4235,
+      "step": 20153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7109515260323164e-05,
+      "loss": 0.5427,
+      "step": 20154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.710887407027443e-05,
+      "loss": 0.4727,
+      "step": 20155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.71082328802257e-05,
+      "loss": 0.5461,
+      "step": 20156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7107591690176966e-05,
+      "loss": 0.4636,
+      "step": 20157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710695050012824e-05,
+      "loss": 0.5774,
+      "step": 20158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710630931007951e-05,
+      "loss": 0.5949,
+      "step": 20159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710566812003078e-05,
+      "loss": 0.5731,
+      "step": 20160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710502692998205e-05,
+      "loss": 0.5925,
+      "step": 20161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710438573993332e-05,
+      "loss": 0.4778,
+      "step": 20162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710374454988459e-05,
+      "loss": 0.5071,
+      "step": 20163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7103103359835854e-05,
+      "loss": 0.404,
+      "step": 20164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7102462169787126e-05,
+      "loss": 0.4068,
+      "step": 20165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710182097973839e-05,
+      "loss": 0.4282,
+      "step": 20166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710117978968967e-05,
+      "loss": 0.6162,
+      "step": 20167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7100538599640934e-05,
+      "loss": 0.4927,
+      "step": 20168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7099897409592206e-05,
+      "loss": 0.4412,
+      "step": 20169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709925621954348e-05,
+      "loss": 0.561,
+      "step": 20170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709861502949474e-05,
+      "loss": 0.6604,
+      "step": 20171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7097973839446015e-05,
+      "loss": 0.5728,
+      "step": 20172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709733264939728e-05,
+      "loss": 0.4575,
+      "step": 20173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709669145934855e-05,
+      "loss": 0.5445,
+      "step": 20174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709605026929982e-05,
+      "loss": 0.4618,
+      "step": 20175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7095409079251095e-05,
+      "loss": 0.4286,
+      "step": 20176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709476788920236e-05,
+      "loss": 0.6221,
+      "step": 20177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709412669915363e-05,
+      "loss": 0.5518,
+      "step": 20178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70934855091049e-05,
+      "loss": 0.6153,
+      "step": 20179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709284431905617e-05,
+      "loss": 0.4571,
+      "step": 20180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709220312900744e-05,
+      "loss": 0.4562,
+      "step": 20181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7091561938958705e-05,
+      "loss": 0.5368,
+      "step": 20182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7090920748909976e-05,
+      "loss": 0.5442,
+      "step": 20183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709027955886125e-05,
+      "loss": 0.5134,
+      "step": 20184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708963836881252e-05,
+      "loss": 0.4663,
+      "step": 20185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7088997178763785e-05,
+      "loss": 0.4895,
+      "step": 20186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7088355988715056e-05,
+      "loss": 0.5552,
+      "step": 20187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708771479866633e-05,
+      "loss": 0.4696,
+      "step": 20188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708707360861759e-05,
+      "loss": 0.424,
+      "step": 20189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7086432418568865e-05,
+      "loss": 0.4856,
+      "step": 20190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7085791228520137e-05,
+      "loss": 0.5504,
+      "step": 20191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708515003847141e-05,
+      "loss": 0.4546,
+      "step": 20192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708450884842267e-05,
+      "loss": 0.4784,
+      "step": 20193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7083867658373945e-05,
+      "loss": 0.4797,
+      "step": 20194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708322646832521e-05,
+      "loss": 0.4661,
+      "step": 20195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708258527827648e-05,
+      "loss": 0.5332,
+      "step": 20196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708194408822775e-05,
+      "loss": 0.6679,
+      "step": 20197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708130289817902e-05,
+      "loss": 0.4403,
+      "step": 20198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708066170813029e-05,
+      "loss": 0.3859,
+      "step": 20199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708002051808156e-05,
+      "loss": 0.525,
+      "step": 20200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707937932803283e-05,
+      "loss": 0.5041,
+      "step": 20201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70787381379841e-05,
+      "loss": 0.6059,
+      "step": 20202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707809694793537e-05,
+      "loss": 0.4545,
+      "step": 20203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7077455757886635e-05,
+      "loss": 0.5247,
+      "step": 20204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707681456783791e-05,
+      "loss": 0.4688,
+      "step": 20205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707617337778918e-05,
+      "loss": 0.5259,
+      "step": 20206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707553218774045e-05,
+      "loss": 0.701,
+      "step": 20207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707489099769172e-05,
+      "loss": 0.545,
+      "step": 20208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707424980764299e-05,
+      "loss": 0.4978,
+      "step": 20209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707360861759426e-05,
+      "loss": 0.45,
+      "step": 20210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7072967427545523e-05,
+      "loss": 0.4494,
+      "step": 20211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7072326237496795e-05,
+      "loss": 0.5479,
+      "step": 20212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707168504744807e-05,
+      "loss": 0.4264,
+      "step": 20213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707104385739933e-05,
+      "loss": 0.3715,
+      "step": 20214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7070402667350604e-05,
+      "loss": 0.4554,
+      "step": 20215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7069761477301875e-05,
+      "loss": 0.4928,
+      "step": 20216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706912028725315e-05,
+      "loss": 0.5198,
+      "step": 20217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706847909720441e-05,
+      "loss": 0.5615,
+      "step": 20218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7067837907155684e-05,
+      "loss": 0.4872,
+      "step": 20219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706719671710695e-05,
+      "loss": 0.3993,
+      "step": 20220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706655552705822e-05,
+      "loss": 0.6266,
+      "step": 20221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706591433700949e-05,
+      "loss": 0.5464,
+      "step": 20222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7065273146960764e-05,
+      "loss": 0.3829,
+      "step": 20223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7064631956912035e-05,
+      "loss": 0.6988,
+      "step": 20224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70639907668633e-05,
+      "loss": 0.3936,
+      "step": 20225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706334957681457e-05,
+      "loss": 0.5308,
+      "step": 20226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706270838676584e-05,
+      "loss": 0.3717,
+      "step": 20227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706206719671711e-05,
+      "loss": 0.4797,
+      "step": 20228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7061426006668374e-05,
+      "loss": 0.4816,
+      "step": 20229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7060784816619645e-05,
+      "loss": 0.4982,
+      "step": 20230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706014362657092e-05,
+      "loss": 0.4922,
+      "step": 20231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705950243652219e-05,
+      "loss": 0.5609,
+      "step": 20232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705886124647346e-05,
+      "loss": 0.4946,
+      "step": 20233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7058220056424725e-05,
+      "loss": 0.5884,
+      "step": 20234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7057578866376e-05,
+      "loss": 0.5551,
+      "step": 20235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705693767632726e-05,
+      "loss": 0.5696,
+      "step": 20236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7056296486278534e-05,
+      "loss": 0.4654,
+      "step": 20237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70556552962298e-05,
+      "loss": 0.4439,
+      "step": 20238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705501410618108e-05,
+      "loss": 0.4637,
+      "step": 20239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705437291613235e-05,
+      "loss": 0.4365,
+      "step": 20240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7053731726083614e-05,
+      "loss": 0.5327,
+      "step": 20241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7053090536034886e-05,
+      "loss": 0.4885,
+      "step": 20242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705244934598615e-05,
+      "loss": 0.5037,
+      "step": 20243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705180815593742e-05,
+      "loss": 0.5554,
+      "step": 20244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705116696588869e-05,
+      "loss": 0.4471,
+      "step": 20245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705052577583996e-05,
+      "loss": 0.4648,
+      "step": 20246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704988458579123e-05,
+      "loss": 0.5184,
+      "step": 20247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70492433957425e-05,
+      "loss": 0.5741,
+      "step": 20248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7048602205693774e-05,
+      "loss": 0.5719,
+      "step": 20249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704796101564504e-05,
+      "loss": 0.4584,
+      "step": 20250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704731982559631e-05,
+      "loss": 0.4241,
+      "step": 20251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7046678635547576e-05,
+      "loss": 0.4442,
+      "step": 20252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704603744549885e-05,
+      "loss": 0.523,
+      "step": 20253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704539625545011e-05,
+      "loss": 0.5348,
+      "step": 20254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704475506540139e-05,
+      "loss": 0.502,
+      "step": 20255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7044113875352656e-05,
+      "loss": 0.6095,
+      "step": 20256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704347268530393e-05,
+      "loss": 0.5478,
+      "step": 20257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70428314952552e-05,
+      "loss": 0.5348,
+      "step": 20258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7042190305206464e-05,
+      "loss": 0.5274,
+      "step": 20259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7041549115157736e-05,
+      "loss": 0.5142,
+      "step": 20260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7040907925109e-05,
+      "loss": 0.3954,
+      "step": 20261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704026673506027e-05,
+      "loss": 0.5645,
+      "step": 20262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7039625545011544e-05,
+      "loss": 0.5467,
+      "step": 20263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7038984354962816e-05,
+      "loss": 0.4577,
+      "step": 20264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703834316491408e-05,
+      "loss": 0.6004,
+      "step": 20265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703770197486535e-05,
+      "loss": 0.4911,
+      "step": 20266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7037060784816624e-05,
+      "loss": 0.5261,
+      "step": 20267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703641959476789e-05,
+      "loss": 0.4577,
+      "step": 20268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703577840471916e-05,
+      "loss": 0.5023,
+      "step": 20269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7035137214670426e-05,
+      "loss": 0.5325,
+      "step": 20270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70344960246217e-05,
+      "loss": 0.6051,
+      "step": 20271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703385483457297e-05,
+      "loss": 0.4656,
+      "step": 20272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703321364452424e-05,
+      "loss": 0.4858,
+      "step": 20273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7032572454475506e-05,
+      "loss": 0.5489,
+      "step": 20274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703193126442678e-05,
+      "loss": 0.4568,
+      "step": 20275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703129007437805e-05,
+      "loss": 0.4162,
+      "step": 20276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7030648884329314e-05,
+      "loss": 0.4431,
+      "step": 20277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7030007694280586e-05,
+      "loss": 0.4539,
+      "step": 20278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702936650423186e-05,
+      "loss": 0.5441,
+      "step": 20279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702872531418313e-05,
+      "loss": 0.4741,
+      "step": 20280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7028084124134395e-05,
+      "loss": 0.5486,
+      "step": 20281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7027442934085666e-05,
+      "loss": 0.644,
+      "step": 20282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702680174403693e-05,
+      "loss": 0.475,
+      "step": 20283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70261605539882e-05,
+      "loss": 0.5835,
+      "step": 20284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7025519363939475e-05,
+      "loss": 0.4723,
+      "step": 20285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702487817389074e-05,
+      "loss": 0.6499,
+      "step": 20286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702423698384201e-05,
+      "loss": 0.4918,
+      "step": 20287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702359579379328e-05,
+      "loss": 0.4724,
+      "step": 20288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7022954603744555e-05,
+      "loss": 0.4044,
+      "step": 20289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702231341369582e-05,
+      "loss": 0.5197,
+      "step": 20290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702167222364709e-05,
+      "loss": 0.5416,
+      "step": 20291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7021031033598356e-05,
+      "loss": 0.5329,
+      "step": 20292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702038984354963e-05,
+      "loss": 0.5753,
+      "step": 20293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70197486535009e-05,
+      "loss": 0.5714,
+      "step": 20294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701910746345217e-05,
+      "loss": 0.597,
+      "step": 20295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701846627340344e-05,
+      "loss": 0.5067,
+      "step": 20296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701782508335471e-05,
+      "loss": 0.5362,
+      "step": 20297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701718389330598e-05,
+      "loss": 0.5675,
+      "step": 20298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7016542703257245e-05,
+      "loss": 0.543,
+      "step": 20299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7015901513208516e-05,
+      "loss": 0.4209,
+      "step": 20300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701526032315978e-05,
+      "loss": 0.5399,
+      "step": 20301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701461913311105e-05,
+      "loss": 0.5588,
+      "step": 20302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7013977943062325e-05,
+      "loss": 0.452,
+      "step": 20303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7013336753013597e-05,
+      "loss": 0.5088,
+      "step": 20304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701269556296487e-05,
+      "loss": 0.4975,
+      "step": 20305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701205437291613e-05,
+      "loss": 0.4097,
+      "step": 20306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7011413182867405e-05,
+      "loss": 0.5348,
+      "step": 20307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701077199281867e-05,
+      "loss": 0.4716,
+      "step": 20308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701013080276994e-05,
+      "loss": 0.4764,
+      "step": 20309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7009489612721207e-05,
+      "loss": 0.5454,
+      "step": 20310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7008848422672485e-05,
+      "loss": 0.739,
+      "step": 20311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700820723262376e-05,
+      "loss": 0.4795,
+      "step": 20312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700756604257502e-05,
+      "loss": 0.5044,
+      "step": 20313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7006924852526293e-05,
+      "loss": 0.4911,
+      "step": 20314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700628366247756e-05,
+      "loss": 0.5199,
+      "step": 20315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700564247242883e-05,
+      "loss": 0.7126,
+      "step": 20316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7005001282380095e-05,
+      "loss": 0.372,
+      "step": 20317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700436009233137e-05,
+      "loss": 0.4098,
+      "step": 20318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700371890228264e-05,
+      "loss": 0.4157,
+      "step": 20319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700307771223391e-05,
+      "loss": 0.4858,
+      "step": 20320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700243652218518e-05,
+      "loss": 0.3747,
+      "step": 20321
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700179533213645e-05,
+      "loss": 0.5451,
+      "step": 20322
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700115414208772e-05,
+      "loss": 0.4943,
+      "step": 20323
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7000512952038983e-05,
+      "loss": 0.5474,
+      "step": 20324
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6999871761990255e-05,
+      "loss": 0.6053,
+      "step": 20325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699923057194152e-05,
+      "loss": 0.6076,
+      "step": 20326
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69985893818928e-05,
+      "loss": 0.5303,
+      "step": 20327
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6997948191844064e-05,
+      "loss": 0.4962,
+      "step": 20328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6997307001795335e-05,
+      "loss": 0.4239,
+      "step": 20329
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699666581174661e-05,
+      "loss": 0.5834,
+      "step": 20330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699602462169787e-05,
+      "loss": 0.4674,
+      "step": 20331
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6995383431649144e-05,
+      "loss": 0.5815,
+      "step": 20332
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699474224160041e-05,
+      "loss": 0.5648,
+      "step": 20333
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699410105155168e-05,
+      "loss": 0.4462,
+      "step": 20334
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699345986150295e-05,
+      "loss": 0.5258,
+      "step": 20335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6992818671454224e-05,
+      "loss": 0.5658,
+      "step": 20336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699217748140549e-05,
+      "loss": 0.3995,
+      "step": 20337
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699153629135676e-05,
+      "loss": 0.5455,
+      "step": 20338
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.699089510130803e-05,
+      "loss": 0.585,
+      "step": 20339
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69902539112593e-05,
+      "loss": 0.4367,
+      "step": 20340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.698961272121057e-05,
+      "loss": 0.578,
+      "step": 20341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6988971531161834e-05,
+      "loss": 0.5389,
+      "step": 20342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6988330341113105e-05,
+      "loss": 0.5173,
+      "step": 20343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.698768915106438e-05,
+      "loss": 0.5538,
+      "step": 20344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.698704796101565e-05,
+      "loss": 0.3962,
+      "step": 20345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6986406770966914e-05,
+      "loss": 0.5844,
+      "step": 20346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6985765580918185e-05,
+      "loss": 0.4038,
+      "step": 20347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.698512439086946e-05,
+      "loss": 0.4923,
+      "step": 20348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.698448320082072e-05,
+      "loss": 0.4611,
+      "step": 20349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6983842010771994e-05,
+      "loss": 0.5875,
+      "step": 20350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6983200820723266e-05,
+      "loss": 0.5056,
+      "step": 20351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.698255963067454e-05,
+      "loss": 0.5078,
+      "step": 20352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69819184406258e-05,
+      "loss": 0.4208,
+      "step": 20353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6981277250577074e-05,
+      "loss": 0.5338,
+      "step": 20354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.698063606052834e-05,
+      "loss": 0.4015,
+      "step": 20355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697999487047961e-05,
+      "loss": 0.5843,
+      "step": 20356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697935368043088e-05,
+      "loss": 0.4752,
+      "step": 20357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697871249038215e-05,
+      "loss": 0.4617,
+      "step": 20358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697807130033342e-05,
+      "loss": 0.4942,
+      "step": 20359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697743011028469e-05,
+      "loss": 0.6092,
+      "step": 20360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697678892023596e-05,
+      "loss": 0.4207,
+      "step": 20361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697614773018723e-05,
+      "loss": 0.5197,
+      "step": 20362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69755065401385e-05,
+      "loss": 0.5674,
+      "step": 20363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6974865350089764e-05,
+      "loss": 0.4314,
+      "step": 20364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6974224160041036e-05,
+      "loss": 0.4866,
+      "step": 20365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697358296999231e-05,
+      "loss": 0.5786,
+      "step": 20366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697294177994358e-05,
+      "loss": 0.4984,
+      "step": 20367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697230058989485e-05,
+      "loss": 0.4966,
+      "step": 20368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6971659399846116e-05,
+      "loss": 0.4994,
+      "step": 20369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697101820979739e-05,
+      "loss": 0.4695,
+      "step": 20370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697037701974865e-05,
+      "loss": 0.5256,
+      "step": 20371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6969735829699924e-05,
+      "loss": 0.4322,
+      "step": 20372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696909463965119e-05,
+      "loss": 0.6301,
+      "step": 20373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696845344960246e-05,
+      "loss": 0.4033,
+      "step": 20374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696781225955373e-05,
+      "loss": 0.4675,
+      "step": 20375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6967171069505004e-05,
+      "loss": 0.5919,
+      "step": 20376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6966529879456276e-05,
+      "loss": 0.5368,
+      "step": 20377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696588868940754e-05,
+      "loss": 0.4726,
+      "step": 20378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696524749935881e-05,
+      "loss": 0.4368,
+      "step": 20379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696460630931008e-05,
+      "loss": 0.4287,
+      "step": 20380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696396511926135e-05,
+      "loss": 0.5287,
+      "step": 20381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696332392921262e-05,
+      "loss": 0.4614,
+      "step": 20382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696268273916389e-05,
+      "loss": 0.6159,
+      "step": 20383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6962041549115164e-05,
+      "loss": 0.5477,
+      "step": 20384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696140035906643e-05,
+      "loss": 0.5677,
+      "step": 20385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69607591690177e-05,
+      "loss": 0.569,
+      "step": 20386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6960117978968966e-05,
+      "loss": 0.6159,
+      "step": 20387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695947678892024e-05,
+      "loss": 0.5495,
+      "step": 20388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69588355988715e-05,
+      "loss": 0.4441,
+      "step": 20389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6958194408822774e-05,
+      "loss": 0.546,
+      "step": 20390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6957553218774046e-05,
+      "loss": 0.4797,
+      "step": 20391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695691202872532e-05,
+      "loss": 0.6183,
+      "step": 20392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695627083867659e-05,
+      "loss": 0.5519,
+      "step": 20393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6955629648627855e-05,
+      "loss": 0.6194,
+      "step": 20394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6954988458579126e-05,
+      "loss": 0.4793,
+      "step": 20395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695434726853039e-05,
+      "loss": 0.4419,
+      "step": 20396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695370607848166e-05,
+      "loss": 0.5322,
+      "step": 20397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695306488843293e-05,
+      "loss": 0.7176,
+      "step": 20398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6952423698384206e-05,
+      "loss": 0.5937,
+      "step": 20399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695178250833548e-05,
+      "loss": 0.6023,
+      "step": 20400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695114131828674e-05,
+      "loss": 0.4867,
+      "step": 20401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6950500128238015e-05,
+      "loss": 0.4979,
+      "step": 20402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694985893818928e-05,
+      "loss": 0.4901,
+      "step": 20403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694921774814055e-05,
+      "loss": 0.4216,
+      "step": 20404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6948576558091816e-05,
+      "loss": 0.506,
+      "step": 20405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694793536804309e-05,
+      "loss": 0.6106,
+      "step": 20406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694729417799436e-05,
+      "loss": 0.5753,
+      "step": 20407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694665298794563e-05,
+      "loss": 0.5304,
+      "step": 20408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69460117978969e-05,
+      "loss": 0.5385,
+      "step": 20409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694537060784817e-05,
+      "loss": 0.5007,
+      "step": 20410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694472941779944e-05,
+      "loss": 0.5356,
+      "step": 20411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6944088227750705e-05,
+      "loss": 0.5128,
+      "step": 20412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6943447037701976e-05,
+      "loss": 0.4461,
+      "step": 20413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694280584765324e-05,
+      "loss": 0.5136,
+      "step": 20414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694216465760452e-05,
+      "loss": 0.4685,
+      "step": 20415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6941523467555785e-05,
+      "loss": 0.5019,
+      "step": 20416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6940882277507057e-05,
+      "loss": 0.4913,
+      "step": 20417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.694024108745833e-05,
+      "loss": 0.5008,
+      "step": 20418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693959989740959e-05,
+      "loss": 0.545,
+      "step": 20419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6938958707360865e-05,
+      "loss": 0.4543,
+      "step": 20420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693831751731213e-05,
+      "loss": 0.5096,
+      "step": 20421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.69376763272634e-05,
+      "loss": 0.4341,
+      "step": 20422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693703513721467e-05,
+      "loss": 0.5031,
+      "step": 20423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6936393947165945e-05,
+      "loss": 0.575,
+      "step": 20424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693575275711721e-05,
+      "loss": 0.5052,
+      "step": 20425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693511156706848e-05,
+      "loss": 0.5584,
+      "step": 20426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6934470377019753e-05,
+      "loss": 0.5299,
+      "step": 20427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693382918697102e-05,
+      "loss": 0.4848,
+      "step": 20428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693318799692229e-05,
+      "loss": 0.6429,
+      "step": 20429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6932546806873555e-05,
+      "loss": 0.5582,
+      "step": 20430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693190561682483e-05,
+      "loss": 0.5142,
+      "step": 20431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.69312644267761e-05,
+      "loss": 0.5401,
+      "step": 20432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693062323672737e-05,
+      "loss": 0.469,
+      "step": 20433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6929982046678635e-05,
+      "loss": 0.4272,
+      "step": 20434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692934085662991e-05,
+      "loss": 0.5163,
+      "step": 20435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692869966658118e-05,
+      "loss": 0.5252,
+      "step": 20436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6928058476532443e-05,
+      "loss": 0.4297,
+      "step": 20437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6927417286483715e-05,
+      "loss": 0.4481,
+      "step": 20438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692677609643499e-05,
+      "loss": 0.5205,
+      "step": 20439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692613490638626e-05,
+      "loss": 0.5263,
+      "step": 20440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6925493716337524e-05,
+      "loss": 0.4479,
+      "step": 20441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6924852526288795e-05,
+      "loss": 0.459,
+      "step": 20442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692421133624006e-05,
+      "loss": 0.4405,
+      "step": 20443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692357014619133e-05,
+      "loss": 0.5365,
+      "step": 20444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6922928956142604e-05,
+      "loss": 0.4412,
+      "step": 20445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692228776609387e-05,
+      "loss": 0.53,
+      "step": 20446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692164657604514e-05,
+      "loss": 0.5014,
+      "step": 20447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692100538599641e-05,
+      "loss": 0.5078,
+      "step": 20448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6920364195947684e-05,
+      "loss": 0.4961,
+      "step": 20449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691972300589895e-05,
+      "loss": 0.5834,
+      "step": 20450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691908181585022e-05,
+      "loss": 0.5464,
+      "step": 20451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6918440625801485e-05,
+      "loss": 0.5538,
+      "step": 20452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691779943575276e-05,
+      "loss": 0.5348,
+      "step": 20453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691715824570403e-05,
+      "loss": 0.4784,
+      "step": 20454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.69165170556553e-05,
+      "loss": 0.3686,
+      "step": 20455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691587586560657e-05,
+      "loss": 0.5945,
+      "step": 20456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691523467555784e-05,
+      "loss": 0.6523,
+      "step": 20457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691459348550911e-05,
+      "loss": 0.5985,
+      "step": 20458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6913952295460374e-05,
+      "loss": 0.5448,
+      "step": 20459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6913311105411646e-05,
+      "loss": 0.4724,
+      "step": 20460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691266991536291e-05,
+      "loss": 0.4715,
+      "step": 20461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691202872531418e-05,
+      "loss": 0.3482,
+      "step": 20462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6911387535265454e-05,
+      "loss": 0.5273,
+      "step": 20463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6910746345216726e-05,
+      "loss": 0.4776,
+      "step": 20464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6910105155168e-05,
+      "loss": 0.5099,
+      "step": 20465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690946396511926e-05,
+      "loss": 0.5342,
+      "step": 20466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6908822775070534e-05,
+      "loss": 0.5442,
+      "step": 20467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.69081815850218e-05,
+      "loss": 0.3782,
+      "step": 20468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690754039497307e-05,
+      "loss": 0.4854,
+      "step": 20469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6906899204924336e-05,
+      "loss": 0.5679,
+      "step": 20470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6906258014875614e-05,
+      "loss": 0.4693,
+      "step": 20471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6905616824826886e-05,
+      "loss": 0.7024,
+      "step": 20472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690497563477815e-05,
+      "loss": 0.4765,
+      "step": 20473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690433444472942e-05,
+      "loss": 0.5539,
+      "step": 20474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690369325468069e-05,
+      "loss": 0.4747,
+      "step": 20475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690305206463196e-05,
+      "loss": 0.5149,
+      "step": 20476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6902410874583224e-05,
+      "loss": 0.6203,
+      "step": 20477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6901769684534496e-05,
+      "loss": 0.4685,
+      "step": 20478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690112849448577e-05,
+      "loss": 0.5004,
+      "step": 20479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690048730443704e-05,
+      "loss": 0.4223,
+      "step": 20480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689984611438831e-05,
+      "loss": 0.5298,
+      "step": 20481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6899204924339576e-05,
+      "loss": 0.5191,
+      "step": 20482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689856373429085e-05,
+      "loss": 0.4701,
+      "step": 20483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689792254424211e-05,
+      "loss": 0.5325,
+      "step": 20484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6897281354193384e-05,
+      "loss": 0.4951,
+      "step": 20485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689664016414465e-05,
+      "loss": 0.4064,
+      "step": 20486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689599897409593e-05,
+      "loss": 0.5014,
+      "step": 20487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689535778404719e-05,
+      "loss": 0.4447,
+      "step": 20488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6894716593998464e-05,
+      "loss": 0.4889,
+      "step": 20489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6894075403949736e-05,
+      "loss": 0.457,
+      "step": 20490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6893434213901e-05,
+      "loss": 0.494,
+      "step": 20491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689279302385227e-05,
+      "loss": 0.4089,
+      "step": 20492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689215183380354e-05,
+      "loss": 0.4936,
+      "step": 20493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689151064375481e-05,
+      "loss": 0.4785,
+      "step": 20494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689086945370608e-05,
+      "loss": 0.6149,
+      "step": 20495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689022826365735e-05,
+      "loss": 0.529,
+      "step": 20496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688958707360862e-05,
+      "loss": 0.4819,
+      "step": 20497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688894588355989e-05,
+      "loss": 0.5027,
+      "step": 20498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688830469351116e-05,
+      "loss": 0.4669,
+      "step": 20499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6887663503462426e-05,
+      "loss": 0.5368,
+      "step": 20500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68870223134137e-05,
+      "loss": 0.6153,
+      "step": 20501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688638112336496e-05,
+      "loss": 0.5073,
+      "step": 20502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688573993331624e-05,
+      "loss": 0.6027,
+      "step": 20503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6885098743267506e-05,
+      "loss": 0.5926,
+      "step": 20504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688445755321878e-05,
+      "loss": 0.5453,
+      "step": 20505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688381636317004e-05,
+      "loss": 0.496,
+      "step": 20506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6883175173121315e-05,
+      "loss": 0.5017,
+      "step": 20507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6882533983072586e-05,
+      "loss": 0.5455,
+      "step": 20508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688189279302385e-05,
+      "loss": 0.5322,
+      "step": 20509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688125160297512e-05,
+      "loss": 0.643,
+      "step": 20510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6880610412926395e-05,
+      "loss": 0.3427,
+      "step": 20511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6879969222877666e-05,
+      "loss": 0.5366,
+      "step": 20512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687932803282893e-05,
+      "loss": 0.5536,
+      "step": 20513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68786868427802e-05,
+      "loss": 0.4954,
+      "step": 20514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687804565273147e-05,
+      "loss": 0.5197,
+      "step": 20515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687740446268274e-05,
+      "loss": 0.5349,
+      "step": 20516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687676327263401e-05,
+      "loss": 0.526,
+      "step": 20517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6876122082585276e-05,
+      "loss": 0.5176,
+      "step": 20518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687548089253655e-05,
+      "loss": 0.5679,
+      "step": 20519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687483970248782e-05,
+      "loss": 0.5943,
+      "step": 20520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687419851243909e-05,
+      "loss": 0.5011,
+      "step": 20521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6873557322390356e-05,
+      "loss": 0.4169,
+      "step": 20522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687291613234163e-05,
+      "loss": 0.4722,
+      "step": 20523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687227494229289e-05,
+      "loss": 0.6015,
+      "step": 20524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6871633752244165e-05,
+      "loss": 0.4589,
+      "step": 20525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6870992562195437e-05,
+      "loss": 0.5692,
+      "step": 20526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687035137214671e-05,
+      "loss": 0.5462,
+      "step": 20527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686971018209798e-05,
+      "loss": 0.5503,
+      "step": 20528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6869068992049245e-05,
+      "loss": 0.5125,
+      "step": 20529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686842780200052e-05,
+      "loss": 0.5449,
+      "step": 20530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686778661195178e-05,
+      "loss": 0.4321,
+      "step": 20531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686714542190305e-05,
+      "loss": 0.4583,
+      "step": 20532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686650423185432e-05,
+      "loss": 0.4372,
+      "step": 20533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686586304180559e-05,
+      "loss": 0.4879,
+      "step": 20534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686522185175686e-05,
+      "loss": 0.4053,
+      "step": 20535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686458066170813e-05,
+      "loss": 0.5645,
+      "step": 20536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6863939471659405e-05,
+      "loss": 0.5071,
+      "step": 20537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686329828161067e-05,
+      "loss": 0.4321,
+      "step": 20538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686265709156194e-05,
+      "loss": 0.4573,
+      "step": 20539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686201590151321e-05,
+      "loss": 0.4597,
+      "step": 20540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686137471146448e-05,
+      "loss": 0.4651,
+      "step": 20541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686073352141575e-05,
+      "loss": 0.5199,
+      "step": 20542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686009233136702e-05,
+      "loss": 0.4939,
+      "step": 20543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6859451141318294e-05,
+      "loss": 0.4809,
+      "step": 20544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685880995126956e-05,
+      "loss": 0.5312,
+      "step": 20545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685816876122083e-05,
+      "loss": 0.5693,
+      "step": 20546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6857527571172095e-05,
+      "loss": 0.4041,
+      "step": 20547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685688638112337e-05,
+      "loss": 0.4757,
+      "step": 20548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685624519107463e-05,
+      "loss": 0.508,
+      "step": 20549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6855604001025904e-05,
+      "loss": 0.5237,
+      "step": 20550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6854962810977175e-05,
+      "loss": 0.5039,
+      "step": 20551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685432162092845e-05,
+      "loss": 0.4912,
+      "step": 20552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685368043087972e-05,
+      "loss": 0.423,
+      "step": 20553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6853039240830984e-05,
+      "loss": 0.5145,
+      "step": 20554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6852398050782255e-05,
+      "loss": 0.557,
+      "step": 20555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685175686073352e-05,
+      "loss": 0.5406,
+      "step": 20556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685111567068479e-05,
+      "loss": 0.5082,
+      "step": 20557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685047448063606e-05,
+      "loss": 0.4858,
+      "step": 20558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6849833290587335e-05,
+      "loss": 0.5038,
+      "step": 20559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68491921005386e-05,
+      "loss": 0.5375,
+      "step": 20560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684855091048987e-05,
+      "loss": 0.568,
+      "step": 20561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6847909720441144e-05,
+      "loss": 0.4557,
+      "step": 20562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684726853039241e-05,
+      "loss": 0.5441,
+      "step": 20563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684662734034368e-05,
+      "loss": 0.6126,
+      "step": 20564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6845986150294945e-05,
+      "loss": 0.4789,
+      "step": 20565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684534496024622e-05,
+      "loss": 0.5016,
+      "step": 20566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684470377019749e-05,
+      "loss": 0.4384,
+      "step": 20567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684406258014876e-05,
+      "loss": 0.3581,
+      "step": 20568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6843421390100025e-05,
+      "loss": 0.5035,
+      "step": 20569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68427802000513e-05,
+      "loss": 0.4867,
+      "step": 20570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684213901000257e-05,
+      "loss": 0.5664,
+      "step": 20571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6841497819953834e-05,
+      "loss": 0.4675,
+      "step": 20572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6840856629905106e-05,
+      "loss": 0.4548,
+      "step": 20573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684021543985637e-05,
+      "loss": 0.5507,
+      "step": 20574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683957424980765e-05,
+      "loss": 0.4903,
+      "step": 20575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6838933059758914e-05,
+      "loss": 0.6058,
+      "step": 20576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6838291869710186e-05,
+      "loss": 0.4725,
+      "step": 20577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683765067966145e-05,
+      "loss": 0.5923,
+      "step": 20578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683700948961272e-05,
+      "loss": 0.5751,
+      "step": 20579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6836368299563994e-05,
+      "loss": 0.4434,
+      "step": 20580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683572710951526e-05,
+      "loss": 0.5458,
+      "step": 20581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683508591946653e-05,
+      "loss": 0.4874,
+      "step": 20582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68344447294178e-05,
+      "loss": 0.4638,
+      "step": 20583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6833803539369074e-05,
+      "loss": 0.5061,
+      "step": 20584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683316234932034e-05,
+      "loss": 0.5385,
+      "step": 20585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683252115927161e-05,
+      "loss": 0.6627,
+      "step": 20586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683187996922288e-05,
+      "loss": 0.5454,
+      "step": 20587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683123877917415e-05,
+      "loss": 0.4817,
+      "step": 20588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683059758912542e-05,
+      "loss": 0.5606,
+      "step": 20589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6829956399076684e-05,
+      "loss": 0.4919,
+      "step": 20590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6829315209027956e-05,
+      "loss": 0.4411,
+      "step": 20591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682867401897923e-05,
+      "loss": 0.6231,
+      "step": 20592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68280328289305e-05,
+      "loss": 0.3937,
+      "step": 20593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6827391638881764e-05,
+      "loss": 0.4405,
+      "step": 20594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6826750448833036e-05,
+      "loss": 0.4403,
+      "step": 20595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682610925878431e-05,
+      "loss": 0.6379,
+      "step": 20596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682546806873557e-05,
+      "loss": 0.5205,
+      "step": 20597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6824826878686844e-05,
+      "loss": 0.3659,
+      "step": 20598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6824185688638116e-05,
+      "loss": 0.5581,
+      "step": 20599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682354449858939e-05,
+      "loss": 0.6051,
+      "step": 20600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682290330854065e-05,
+      "loss": 0.467,
+      "step": 20601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6822262118491924e-05,
+      "loss": 0.5514,
+      "step": 20602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682162092844319e-05,
+      "loss": 0.4453,
+      "step": 20603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682097973839446e-05,
+      "loss": 0.5005,
+      "step": 20604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682033854834573e-05,
+      "loss": 0.6112,
+      "step": 20605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6819697358297e-05,
+      "loss": 0.5692,
+      "step": 20606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681905616824827e-05,
+      "loss": 0.537,
+      "step": 20607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681841497819954e-05,
+      "loss": 0.5051,
+      "step": 20608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681777378815081e-05,
+      "loss": 0.5695,
+      "step": 20609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681713259810208e-05,
+      "loss": 0.5238,
+      "step": 20610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681649140805335e-05,
+      "loss": 0.5108,
+      "step": 20611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6815850218004614e-05,
+      "loss": 0.4562,
+      "step": 20612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6815209027955886e-05,
+      "loss": 0.5561,
+      "step": 20613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681456783790716e-05,
+      "loss": 0.5316,
+      "step": 20614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681392664785843e-05,
+      "loss": 0.4647,
+      "step": 20615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68132854578097e-05,
+      "loss": 0.5679,
+      "step": 20616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6812644267760966e-05,
+      "loss": 0.4188,
+      "step": 20617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681200307771224e-05,
+      "loss": 0.4925,
+      "step": 20618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68113618876635e-05,
+      "loss": 0.5502,
+      "step": 20619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6810720697614775e-05,
+      "loss": 0.4749,
+      "step": 20620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681007950756604e-05,
+      "loss": 0.5232,
+      "step": 20621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680943831751731e-05,
+      "loss": 0.5612,
+      "step": 20622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680879712746858e-05,
+      "loss": 0.4915,
+      "step": 20623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6808155937419855e-05,
+      "loss": 0.4892,
+      "step": 20624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6807514747371126e-05,
+      "loss": 0.5238,
+      "step": 20625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680687355732239e-05,
+      "loss": 0.4539,
+      "step": 20626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680623236727366e-05,
+      "loss": 0.4976,
+      "step": 20627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680559117722493e-05,
+      "loss": 0.4939,
+      "step": 20628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.68049499871762e-05,
+      "loss": 0.541,
+      "step": 20629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6804308797127465e-05,
+      "loss": 0.5512,
+      "step": 20630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680366760707874e-05,
+      "loss": 0.4898,
+      "step": 20631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6803026417030015e-05,
+      "loss": 0.4869,
+      "step": 20632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680238522698128e-05,
+      "loss": 0.4071,
+      "step": 20633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680174403693255e-05,
+      "loss": 0.449,
+      "step": 20634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6801102846883816e-05,
+      "loss": 0.5361,
+      "step": 20635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680046165683509e-05,
+      "loss": 0.5486,
+      "step": 20636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679982046678635e-05,
+      "loss": 0.5862,
+      "step": 20637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6799179276737625e-05,
+      "loss": 0.5803,
+      "step": 20638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6798538086688897e-05,
+      "loss": 0.5868,
+      "step": 20639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679789689664017e-05,
+      "loss": 0.5262,
+      "step": 20640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679725570659144e-05,
+      "loss": 0.4402,
+      "step": 20641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6796614516542705e-05,
+      "loss": 0.5854,
+      "step": 20642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679597332649398e-05,
+      "loss": 0.494,
+      "step": 20643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679533213644524e-05,
+      "loss": 0.4852,
+      "step": 20644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679469094639651e-05,
+      "loss": 0.4953,
+      "step": 20645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679404975634778e-05,
+      "loss": 0.4677,
+      "step": 20646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679340856629906e-05,
+      "loss": 0.5287,
+      "step": 20647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679276737625032e-05,
+      "loss": 0.4795,
+      "step": 20648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6792126186201593e-05,
+      "loss": 0.4791,
+      "step": 20649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6791484996152865e-05,
+      "loss": 0.5299,
+      "step": 20650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679084380610413e-05,
+      "loss": 0.3982,
+      "step": 20651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.67902026160554e-05,
+      "loss": 0.486,
+      "step": 20652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678956142600667e-05,
+      "loss": 0.4979,
+      "step": 20653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678892023595794e-05,
+      "loss": 0.4407,
+      "step": 20654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678827904590921e-05,
+      "loss": 0.452,
+      "step": 20655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678763785586048e-05,
+      "loss": 0.4887,
+      "step": 20656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678699666581175e-05,
+      "loss": 0.4896,
+      "step": 20657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678635547576302e-05,
+      "loss": 0.4549,
+      "step": 20658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678571428571429e-05,
+      "loss": 0.5571,
+      "step": 20659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6785073095665555e-05,
+      "loss": 0.4722,
+      "step": 20660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678443190561683e-05,
+      "loss": 0.4854,
+      "step": 20661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678379071556809e-05,
+      "loss": 0.6293,
+      "step": 20662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678314952551937e-05,
+      "loss": 0.5512,
+      "step": 20663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6782508335470635e-05,
+      "loss": 0.558,
+      "step": 20664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678186714542191e-05,
+      "loss": 0.4705,
+      "step": 20665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678122595537317e-05,
+      "loss": 0.5616,
+      "step": 20666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6780584765324444e-05,
+      "loss": 0.5337,
+      "step": 20667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6779943575275715e-05,
+      "loss": 0.4918,
+      "step": 20668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.677930238522698e-05,
+      "loss": 0.4926,
+      "step": 20669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.677866119517825e-05,
+      "loss": 0.4587,
+      "step": 20670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6778020005129524e-05,
+      "loss": 0.5537,
+      "step": 20671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6777378815080795e-05,
+      "loss": 0.4803,
+      "step": 20672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.677673762503206e-05,
+      "loss": 0.5497,
+      "step": 20673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.677609643498333e-05,
+      "loss": 0.4895,
+      "step": 20674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.67754552449346e-05,
+      "loss": 0.5164,
+      "step": 20675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.677481405488587e-05,
+      "loss": 0.5245,
+      "step": 20676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.677417286483714e-05,
+      "loss": 0.5137,
+      "step": 20677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6773531674788405e-05,
+      "loss": 0.448,
+      "step": 20678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.677289048473968e-05,
+      "loss": 0.4658,
+      "step": 20679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.677224929469095e-05,
+      "loss": 0.543,
+      "step": 20680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.677160810464222e-05,
+      "loss": 0.518,
+      "step": 20681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6770966914593486e-05,
+      "loss": 0.5564,
+      "step": 20682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.677032572454476e-05,
+      "loss": 0.4432,
+      "step": 20683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676968453449602e-05,
+      "loss": 0.5999,
+      "step": 20684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6769043344447294e-05,
+      "loss": 0.4781,
+      "step": 20685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6768402154398566e-05,
+      "loss": 0.6413,
+      "step": 20686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676776096434984e-05,
+      "loss": 0.5175,
+      "step": 20687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676711977430111e-05,
+      "loss": 0.4456,
+      "step": 20688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6766478584252374e-05,
+      "loss": 0.5,
+      "step": 20689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6765837394203646e-05,
+      "loss": 0.6563,
+      "step": 20690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676519620415491e-05,
+      "loss": 0.6049,
+      "step": 20691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676455501410618e-05,
+      "loss": 0.5471,
+      "step": 20692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676391382405745e-05,
+      "loss": 0.554,
+      "step": 20693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676327263400872e-05,
+      "loss": 0.5858,
+      "step": 20694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676263144395999e-05,
+      "loss": 0.5347,
+      "step": 20695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676199025391126e-05,
+      "loss": 0.5249,
+      "step": 20696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6761349063862534e-05,
+      "loss": 0.492,
+      "step": 20697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67607078738138e-05,
+      "loss": 0.4748,
+      "step": 20698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676006668376507e-05,
+      "loss": 0.5837,
+      "step": 20699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6759425493716336e-05,
+      "loss": 0.4791,
+      "step": 20700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675878430366761e-05,
+      "loss": 0.4888,
+      "step": 20701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675814311361888e-05,
+      "loss": 0.5887,
+      "step": 20702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675750192357015e-05,
+      "loss": 0.6174,
+      "step": 20703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675686073352142e-05,
+      "loss": 0.4281,
+      "step": 20704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675621954347269e-05,
+      "loss": 0.473,
+      "step": 20705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675557835342396e-05,
+      "loss": 0.4931,
+      "step": 20706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6754937163375224e-05,
+      "loss": 0.4765,
+      "step": 20707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6754295973326496e-05,
+      "loss": 0.4329,
+      "step": 20708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675365478327776e-05,
+      "loss": 0.5596,
+      "step": 20709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675301359322903e-05,
+      "loss": 0.4583,
+      "step": 20710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6752372403180304e-05,
+      "loss": 0.52,
+      "step": 20711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6751731213131576e-05,
+      "loss": 0.5702,
+      "step": 20712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675109002308285e-05,
+      "loss": 0.5363,
+      "step": 20713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675044883303411e-05,
+      "loss": 0.5496,
+      "step": 20714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6749807642985384e-05,
+      "loss": 0.4065,
+      "step": 20715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674916645293665e-05,
+      "loss": 0.4175,
+      "step": 20716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674852526288792e-05,
+      "loss": 0.5531,
+      "step": 20717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6747884072839186e-05,
+      "loss": 0.5,
+      "step": 20718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6747242882790464e-05,
+      "loss": 0.5177,
+      "step": 20719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674660169274173e-05,
+      "loss": 0.5758,
+      "step": 20720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6745960502693e-05,
+      "loss": 0.6627,
+      "step": 20721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674531931264427e-05,
+      "loss": 0.4997,
+      "step": 20722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674467812259554e-05,
+      "loss": 0.531,
+      "step": 20723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674403693254681e-05,
+      "loss": 0.5244,
+      "step": 20724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6743395742498074e-05,
+      "loss": 0.4957,
+      "step": 20725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6742754552449346e-05,
+      "loss": 0.4536,
+      "step": 20726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674211336240062e-05,
+      "loss": 0.4314,
+      "step": 20727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674147217235189e-05,
+      "loss": 0.464,
+      "step": 20728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6740830982303155e-05,
+      "loss": 0.6002,
+      "step": 20729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6740189792254426e-05,
+      "loss": 0.4681,
+      "step": 20730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67395486022057e-05,
+      "loss": 0.4583,
+      "step": 20731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673890741215696e-05,
+      "loss": 0.5159,
+      "step": 20732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6738266222108235e-05,
+      "loss": 0.4686,
+      "step": 20733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67376250320595e-05,
+      "loss": 0.4815,
+      "step": 20734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673698384201078e-05,
+      "loss": 0.4636,
+      "step": 20735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673634265196204e-05,
+      "loss": 0.4654,
+      "step": 20736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6735701461913315e-05,
+      "loss": 0.4204,
+      "step": 20737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673506027186458e-05,
+      "loss": 0.5262,
+      "step": 20738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673441908181585e-05,
+      "loss": 0.4383,
+      "step": 20739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673377789176712e-05,
+      "loss": 0.4991,
+      "step": 20740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673313670171839e-05,
+      "loss": 0.5036,
+      "step": 20741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673249551166966e-05,
+      "loss": 0.4489,
+      "step": 20742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673185432162093e-05,
+      "loss": 0.4964,
+      "step": 20743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67312131315722e-05,
+      "loss": 0.5346,
+      "step": 20744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673057194152347e-05,
+      "loss": 0.4205,
+      "step": 20745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672993075147474e-05,
+      "loss": 0.4973,
+      "step": 20746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6729289561426005e-05,
+      "loss": 0.4338,
+      "step": 20747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6728648371377277e-05,
+      "loss": 0.435,
+      "step": 20748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672800718132855e-05,
+      "loss": 0.4892,
+      "step": 20749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672736599127981e-05,
+      "loss": 0.5062,
+      "step": 20750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672672480123109e-05,
+      "loss": 0.5928,
+      "step": 20751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6726083611182357e-05,
+      "loss": 0.4595,
+      "step": 20752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672544242113363e-05,
+      "loss": 0.5915,
+      "step": 20753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672480123108489e-05,
+      "loss": 0.416,
+      "step": 20754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6724160041036165e-05,
+      "loss": 0.4814,
+      "step": 20755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672351885098743e-05,
+      "loss": 0.4805,
+      "step": 20756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67228776609387e-05,
+      "loss": 0.6825,
+      "step": 20757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672223647088997e-05,
+      "loss": 0.4357,
+      "step": 20758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6721595280841245e-05,
+      "loss": 0.4267,
+      "step": 20759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672095409079252e-05,
+      "loss": 0.443,
+      "step": 20760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672031290074378e-05,
+      "loss": 0.457,
+      "step": 20761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6719671710695053e-05,
+      "loss": 0.6225,
+      "step": 20762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671903052064632e-05,
+      "loss": 0.5934,
+      "step": 20763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671838933059759e-05,
+      "loss": 0.4713,
+      "step": 20764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6717748140548855e-05,
+      "loss": 0.6159,
+      "step": 20765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671710695050013e-05,
+      "loss": 0.5531,
+      "step": 20766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67164657604514e-05,
+      "loss": 0.4831,
+      "step": 20767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671582457040267e-05,
+      "loss": 0.5393,
+      "step": 20768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671518338035394e-05,
+      "loss": 0.453,
+      "step": 20769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671454219030521e-05,
+      "loss": 0.4482,
+      "step": 20770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671390100025648e-05,
+      "loss": 0.6134,
+      "step": 20771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6713259810207743e-05,
+      "loss": 0.5178,
+      "step": 20772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6712618620159015e-05,
+      "loss": 0.4769,
+      "step": 20773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671197743011029e-05,
+      "loss": 0.47,
+      "step": 20774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671133624006156e-05,
+      "loss": 0.5311,
+      "step": 20775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671069505001283e-05,
+      "loss": 0.575,
+      "step": 20776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6710053859964095e-05,
+      "loss": 0.5262,
+      "step": 20777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670941266991537e-05,
+      "loss": 0.4563,
+      "step": 20778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670877147986663e-05,
+      "loss": 0.4179,
+      "step": 20779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6708130289817904e-05,
+      "loss": 0.5243,
+      "step": 20780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670748909976917e-05,
+      "loss": 0.5322,
+      "step": 20781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670684790972044e-05,
+      "loss": 0.5204,
+      "step": 20782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670620671967171e-05,
+      "loss": 0.5411,
+      "step": 20783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6705565529622984e-05,
+      "loss": 0.4423,
+      "step": 20784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6704924339574255e-05,
+      "loss": 0.5142,
+      "step": 20785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670428314952552e-05,
+      "loss": 0.4327,
+      "step": 20786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670364195947679e-05,
+      "loss": 0.4788,
+      "step": 20787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670300076942806e-05,
+      "loss": 0.4128,
+      "step": 20788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670235957937933e-05,
+      "loss": 0.511,
+      "step": 20789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67017183893306e-05,
+      "loss": 0.4313,
+      "step": 20790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670107719928187e-05,
+      "loss": 0.4837,
+      "step": 20791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6700436009233144e-05,
+      "loss": 0.5795,
+      "step": 20792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669979481918441e-05,
+      "loss": 0.4756,
+      "step": 20793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669915362913568e-05,
+      "loss": 0.4405,
+      "step": 20794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6698512439086946e-05,
+      "loss": 0.6275,
+      "step": 20795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669787124903822e-05,
+      "loss": 0.544,
+      "step": 20796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669723005898948e-05,
+      "loss": 0.3817,
+      "step": 20797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6696588868940754e-05,
+      "loss": 0.5993,
+      "step": 20798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6695947678892026e-05,
+      "loss": 0.5478,
+      "step": 20799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66953064888433e-05,
+      "loss": 0.4581,
+      "step": 20800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669466529879457e-05,
+      "loss": 0.439,
+      "step": 20801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6694024108745834e-05,
+      "loss": 0.4358,
+      "step": 20802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6693382918697106e-05,
+      "loss": 0.5624,
+      "step": 20803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669274172864837e-05,
+      "loss": 0.4098,
+      "step": 20804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669210053859964e-05,
+      "loss": 0.6263,
+      "step": 20805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669145934855091e-05,
+      "loss": 0.6269,
+      "step": 20806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6690818158502186e-05,
+      "loss": 0.4076,
+      "step": 20807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669017696845345e-05,
+      "loss": 0.589,
+      "step": 20808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668953577840472e-05,
+      "loss": 0.4277,
+      "step": 20809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6688894588355994e-05,
+      "loss": 0.5778,
+      "step": 20810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668825339830726e-05,
+      "loss": 0.5077,
+      "step": 20811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668761220825853e-05,
+      "loss": 0.4792,
+      "step": 20812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6686971018209796e-05,
+      "loss": 0.62,
+      "step": 20813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668632982816107e-05,
+      "loss": 0.5242,
+      "step": 20814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668568863811234e-05,
+      "loss": 0.5214,
+      "step": 20815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668504744806361e-05,
+      "loss": 0.4773,
+      "step": 20816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6684406258014876e-05,
+      "loss": 0.5008,
+      "step": 20817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668376506796615e-05,
+      "loss": 0.5412,
+      "step": 20818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668312387791742e-05,
+      "loss": 0.4099,
+      "step": 20819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6682482687868684e-05,
+      "loss": 0.5601,
+      "step": 20820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6681841497819956e-05,
+      "loss": 0.4862,
+      "step": 20821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668120030777122e-05,
+      "loss": 0.5012,
+      "step": 20822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66805591177225e-05,
+      "loss": 0.5241,
+      "step": 20823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6679917927673764e-05,
+      "loss": 0.4955,
+      "step": 20824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6679276737625036e-05,
+      "loss": 0.5528,
+      "step": 20825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66786355475763e-05,
+      "loss": 0.4706,
+      "step": 20826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667799435752757e-05,
+      "loss": 0.4707,
+      "step": 20827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6677353167478844e-05,
+      "loss": 0.4649,
+      "step": 20828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667671197743011e-05,
+      "loss": 0.4944,
+      "step": 20829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667607078738138e-05,
+      "loss": 0.4299,
+      "step": 20830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667542959733265e-05,
+      "loss": 0.5888,
+      "step": 20831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6674788407283925e-05,
+      "loss": 0.4762,
+      "step": 20832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667414721723519e-05,
+      "loss": 0.4459,
+      "step": 20833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667350602718646e-05,
+      "loss": 0.5241,
+      "step": 20834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6672864837137726e-05,
+      "loss": 0.5305,
+      "step": 20835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6672223647089e-05,
+      "loss": 0.436,
+      "step": 20836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667158245704027e-05,
+      "loss": 0.4753,
+      "step": 20837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6670941266991534e-05,
+      "loss": 0.4385,
+      "step": 20838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6670300076942806e-05,
+      "loss": 0.5471,
+      "step": 20839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666965888689408e-05,
+      "loss": 0.4905,
+      "step": 20840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666901769684535e-05,
+      "loss": 0.5197,
+      "step": 20841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6668376506796615e-05,
+      "loss": 0.6123,
+      "step": 20842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6667735316747886e-05,
+      "loss": 0.4216,
+      "step": 20843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666709412669915e-05,
+      "loss": 0.5928,
+      "step": 20844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666645293665042e-05,
+      "loss": 0.608,
+      "step": 20845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6665811746601695e-05,
+      "loss": 0.4658,
+      "step": 20846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6665170556552966e-05,
+      "loss": 0.4753,
+      "step": 20847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666452936650424e-05,
+      "loss": 0.5105,
+      "step": 20848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66638881764555e-05,
+      "loss": 0.5289,
+      "step": 20849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6663246986406775e-05,
+      "loss": 0.4938,
+      "step": 20850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666260579635804e-05,
+      "loss": 0.4708,
+      "step": 20851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666196460630931e-05,
+      "loss": 0.5323,
+      "step": 20852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6661323416260576e-05,
+      "loss": 0.3787,
+      "step": 20853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666068222621185e-05,
+      "loss": 0.4819,
+      "step": 20854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666004103616312e-05,
+      "loss": 0.5453,
+      "step": 20855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665939984611439e-05,
+      "loss": 0.5156,
+      "step": 20856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665875865606566e-05,
+      "loss": 0.5134,
+      "step": 20857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665811746601693e-05,
+      "loss": 0.5569,
+      "step": 20858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66574762759682e-05,
+      "loss": 0.372,
+      "step": 20859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6656835085919465e-05,
+      "loss": 0.5287,
+      "step": 20860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6656193895870737e-05,
+      "loss": 0.5276,
+      "step": 20861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665555270582201e-05,
+      "loss": 0.4597,
+      "step": 20862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665491151577328e-05,
+      "loss": 0.5824,
+      "step": 20863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665427032572455e-05,
+      "loss": 0.5201,
+      "step": 20864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665362913567582e-05,
+      "loss": 0.5379,
+      "step": 20865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665298794562709e-05,
+      "loss": 0.5621,
+      "step": 20866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665234675557835e-05,
+      "loss": 0.4799,
+      "step": 20867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6651705565529625e-05,
+      "loss": 0.5669,
+      "step": 20868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665106437548089e-05,
+      "loss": 0.4754,
+      "step": 20869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665042318543216e-05,
+      "loss": 0.5477,
+      "step": 20870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664978199538343e-05,
+      "loss": 0.5229,
+      "step": 20871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6649140805334705e-05,
+      "loss": 0.474,
+      "step": 20872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664849961528598e-05,
+      "loss": 0.4685,
+      "step": 20873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664785842523724e-05,
+      "loss": 0.5197,
+      "step": 20874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6647217235188513e-05,
+      "loss": 0.493,
+      "step": 20875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664657604513978e-05,
+      "loss": 0.425,
+      "step": 20876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664593485509105e-05,
+      "loss": 0.4782,
+      "step": 20877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6645293665042315e-05,
+      "loss": 0.5574,
+      "step": 20878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6644652474993594e-05,
+      "loss": 0.5562,
+      "step": 20879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664401128494486e-05,
+      "loss": 0.4441,
+      "step": 20880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664337009489613e-05,
+      "loss": 0.456,
+      "step": 20881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66427289048474e-05,
+      "loss": 0.4802,
+      "step": 20882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664208771479867e-05,
+      "loss": 0.4556,
+      "step": 20883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664144652474994e-05,
+      "loss": 0.517,
+      "step": 20884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6640805334701204e-05,
+      "loss": 0.4931,
+      "step": 20885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6640164144652475e-05,
+      "loss": 0.5181,
+      "step": 20886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663952295460375e-05,
+      "loss": 0.5854,
+      "step": 20887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663888176455502e-05,
+      "loss": 0.5977,
+      "step": 20888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6638240574506284e-05,
+      "loss": 0.4875,
+      "step": 20889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6637599384457555e-05,
+      "loss": 0.4706,
+      "step": 20890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663695819440883e-05,
+      "loss": 0.4896,
+      "step": 20891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663631700436009e-05,
+      "loss": 0.5255,
+      "step": 20892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6635675814311364e-05,
+      "loss": 0.5114,
+      "step": 20893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663503462426263e-05,
+      "loss": 0.4288,
+      "step": 20894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663439343421391e-05,
+      "loss": 0.5667,
+      "step": 20895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663375224416517e-05,
+      "loss": 0.572,
+      "step": 20896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6633111054116444e-05,
+      "loss": 0.5375,
+      "step": 20897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663246986406771e-05,
+      "loss": 0.4612,
+      "step": 20898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663182867401898e-05,
+      "loss": 0.4678,
+      "step": 20899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663118748397025e-05,
+      "loss": 0.4836,
+      "step": 20900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663054629392152e-05,
+      "loss": 0.5218,
+      "step": 20901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662990510387279e-05,
+      "loss": 0.4405,
+      "step": 20902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662926391382406e-05,
+      "loss": 0.4919,
+      "step": 20903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662862272377533e-05,
+      "loss": 0.3759,
+      "step": 20904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66279815337266e-05,
+      "loss": 0.5262,
+      "step": 20905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662734034367787e-05,
+      "loss": 0.5769,
+      "step": 20906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6626699153629134e-05,
+      "loss": 0.4468,
+      "step": 20907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6626057963580406e-05,
+      "loss": 0.5024,
+      "step": 20908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662541677353168e-05,
+      "loss": 0.4656,
+      "step": 20909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662477558348294e-05,
+      "loss": 0.5094,
+      "step": 20910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662413439343422e-05,
+      "loss": 0.4117,
+      "step": 20911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6623493203385486e-05,
+      "loss": 0.5086,
+      "step": 20912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662285201333676e-05,
+      "loss": 0.4647,
+      "step": 20913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662221082328802e-05,
+      "loss": 0.4894,
+      "step": 20914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6621569633239294e-05,
+      "loss": 0.3399,
+      "step": 20915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662092844319056e-05,
+      "loss": 0.5258,
+      "step": 20916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662028725314183e-05,
+      "loss": 0.5232,
+      "step": 20917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66196460630931e-05,
+      "loss": 0.4362,
+      "step": 20918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6619004873044374e-05,
+      "loss": 0.5765,
+      "step": 20919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6618363682995646e-05,
+      "loss": 0.6774,
+      "step": 20920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661772249294691e-05,
+      "loss": 0.4702,
+      "step": 20921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661708130289818e-05,
+      "loss": 0.5662,
+      "step": 20922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661644011284945e-05,
+      "loss": 0.4948,
+      "step": 20923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661579892280072e-05,
+      "loss": 0.4782,
+      "step": 20924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6615157732751984e-05,
+      "loss": 0.696,
+      "step": 20925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6614516542703256e-05,
+      "loss": 0.5293,
+      "step": 20926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661387535265453e-05,
+      "loss": 0.4799,
+      "step": 20927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66132341626058e-05,
+      "loss": 0.4354,
+      "step": 20928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661259297255707e-05,
+      "loss": 0.4284,
+      "step": 20929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6611951782508336e-05,
+      "loss": 0.6269,
+      "step": 20930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661131059245961e-05,
+      "loss": 0.5636,
+      "step": 20931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661066940241087e-05,
+      "loss": 0.5135,
+      "step": 20932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6610028212362144e-05,
+      "loss": 0.5014,
+      "step": 20933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6609387022313416e-05,
+      "loss": 0.4957,
+      "step": 20934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.660874583226469e-05,
+      "loss": 0.456,
+      "step": 20935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.660810464221596e-05,
+      "loss": 0.4696,
+      "step": 20936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6607463452167224e-05,
+      "loss": 0.498,
+      "step": 20937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6606822262118496e-05,
+      "loss": 0.3849,
+      "step": 20938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660618107206976e-05,
+      "loss": 0.4205,
+      "step": 20939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660553988202103e-05,
+      "loss": 0.4683,
+      "step": 20940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66048986919723e-05,
+      "loss": 0.4583,
+      "step": 20941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660425750192357e-05,
+      "loss": 0.4556,
+      "step": 20942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660361631187484e-05,
+      "loss": 0.5478,
+      "step": 20943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660297512182611e-05,
+      "loss": 0.4735,
+      "step": 20944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6602333931777385e-05,
+      "loss": 0.6629,
+      "step": 20945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660169274172865e-05,
+      "loss": 0.4996,
+      "step": 20946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660105155167992e-05,
+      "loss": 0.509,
+      "step": 20947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6600410361631186e-05,
+      "loss": 0.43,
+      "step": 20948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659976917158246e-05,
+      "loss": 0.5075,
+      "step": 20949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659912798153373e-05,
+      "loss": 0.4697,
+      "step": 20950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6598486791485e-05,
+      "loss": 0.5164,
+      "step": 20951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6597845601436266e-05,
+      "loss": 0.4835,
+      "step": 20952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659720441138754e-05,
+      "loss": 0.6664,
+      "step": 20953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659656322133881e-05,
+      "loss": 0.5388,
+      "step": 20954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6595922031290075e-05,
+      "loss": 0.4946,
+      "step": 20955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6595280841241346e-05,
+      "loss": 0.6053,
+      "step": 20956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659463965119261e-05,
+      "loss": 0.5272,
+      "step": 20957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659399846114388e-05,
+      "loss": 0.6516,
+      "step": 20958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6593357271095155e-05,
+      "loss": 0.4735,
+      "step": 20959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6592716081046426e-05,
+      "loss": 0.4024,
+      "step": 20960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659207489099769e-05,
+      "loss": 0.5011,
+      "step": 20961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659143370094896e-05,
+      "loss": 0.5511,
+      "step": 20962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6590792510900235e-05,
+      "loss": 0.5017,
+      "step": 20963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65901513208515e-05,
+      "loss": 0.4641,
+      "step": 20964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658951013080277e-05,
+      "loss": 0.5155,
+      "step": 20965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6588868940754036e-05,
+      "loss": 0.5023,
+      "step": 20966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6588227750705315e-05,
+      "loss": 0.5558,
+      "step": 20967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658758656065658e-05,
+      "loss": 0.6723,
+      "step": 20968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658694537060785e-05,
+      "loss": 0.4098,
+      "step": 20969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658630418055912e-05,
+      "loss": 0.5759,
+      "step": 20970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658566299051039e-05,
+      "loss": 0.488,
+      "step": 20971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658502180046166e-05,
+      "loss": 0.5584,
+      "step": 20972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6584380610412925e-05,
+      "loss": 0.5037,
+      "step": 20973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6583739420364197e-05,
+      "loss": 0.3427,
+      "step": 20974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658309823031547e-05,
+      "loss": 0.563,
+      "step": 20975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658245704026674e-05,
+      "loss": 0.5393,
+      "step": 20976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6581815850218005e-05,
+      "loss": 0.5728,
+      "step": 20977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658117466016928e-05,
+      "loss": 0.4489,
+      "step": 20978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658053347012055e-05,
+      "loss": 0.5272,
+      "step": 20979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657989228007181e-05,
+      "loss": 0.5655,
+      "step": 20980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6579251090023085e-05,
+      "loss": 0.5497,
+      "step": 20981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657860989997435e-05,
+      "loss": 0.6573,
+      "step": 20982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657796870992563e-05,
+      "loss": 0.4722,
+      "step": 20983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6577327519876893e-05,
+      "loss": 0.5174,
+      "step": 20984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6576686329828165e-05,
+      "loss": 0.4227,
+      "step": 20985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657604513977943e-05,
+      "loss": 0.5725,
+      "step": 20986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65754039497307e-05,
+      "loss": 0.4905,
+      "step": 20987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6574762759681973e-05,
+      "loss": 0.5262,
+      "step": 20988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657412156963324e-05,
+      "loss": 0.5078,
+      "step": 20989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657348037958451e-05,
+      "loss": 0.6219,
+      "step": 20990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657283918953578e-05,
+      "loss": 0.4544,
+      "step": 20991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6572197999487054e-05,
+      "loss": 0.5032,
+      "step": 20992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657155680943832e-05,
+      "loss": 0.6141,
+      "step": 20993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657091561938959e-05,
+      "loss": 0.5019,
+      "step": 20994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6570274429340855e-05,
+      "loss": 0.6399,
+      "step": 20995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656963323929213e-05,
+      "loss": 0.5159,
+      "step": 20996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65689920492434e-05,
+      "loss": 0.4658,
+      "step": 20997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6568350859194664e-05,
+      "loss": 0.5432,
+      "step": 20998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6567709669145935e-05,
+      "loss": 0.4871,
+      "step": 20999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656706847909721e-05,
+      "loss": 0.4721,
+      "step": 21000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656642728904848e-05,
+      "loss": 0.4711,
+      "step": 21001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6565786098999744e-05,
+      "loss": 0.566,
+      "step": 21002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6565144908951015e-05,
+      "loss": 0.4451,
+      "step": 21003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656450371890228e-05,
+      "loss": 0.5013,
+      "step": 21004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656386252885355e-05,
+      "loss": 0.5524,
+      "step": 21005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6563221338804824e-05,
+      "loss": 0.5569,
+      "step": 21006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6562580148756095e-05,
+      "loss": 0.5157,
+      "step": 21007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656193895870737e-05,
+      "loss": 0.4863,
+      "step": 21008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656129776865863e-05,
+      "loss": 0.3977,
+      "step": 21009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6560656578609904e-05,
+      "loss": 0.468,
+      "step": 21010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656001538856117e-05,
+      "loss": 0.4308,
+      "step": 21011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655937419851244e-05,
+      "loss": 0.5331,
+      "step": 21012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6558733008463705e-05,
+      "loss": 0.3891,
+      "step": 21013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655809181841498e-05,
+      "loss": 0.4181,
+      "step": 21014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655745062836625e-05,
+      "loss": 0.4609,
+      "step": 21015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655680943831752e-05,
+      "loss": 0.7033,
+      "step": 21016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655616824826879e-05,
+      "loss": 0.413,
+      "step": 21017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655552705822006e-05,
+      "loss": 0.4344,
+      "step": 21018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655488586817133e-05,
+      "loss": 0.5355,
+      "step": 21019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6554244678122594e-05,
+      "loss": 0.4724,
+      "step": 21020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6553603488073866e-05,
+      "loss": 0.6142,
+      "step": 21021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655296229802514e-05,
+      "loss": 0.5381,
+      "step": 21022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655232110797641e-05,
+      "loss": 0.4515,
+      "step": 21023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655167991792768e-05,
+      "loss": 0.5217,
+      "step": 21024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6551038727878946e-05,
+      "loss": 0.4475,
+      "step": 21025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655039753783022e-05,
+      "loss": 0.6222,
+      "step": 21026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654975634778148e-05,
+      "loss": 0.6144,
+      "step": 21027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6549115157732754e-05,
+      "loss": 0.4782,
+      "step": 21028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654847396768402e-05,
+      "loss": 0.4492,
+      "step": 21029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654783277763529e-05,
+      "loss": 0.4603,
+      "step": 21030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654719158758656e-05,
+      "loss": 0.6045,
+      "step": 21031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6546550397537834e-05,
+      "loss": 0.5112,
+      "step": 21032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6545909207489106e-05,
+      "loss": 0.3952,
+      "step": 21033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654526801744037e-05,
+      "loss": 0.5273,
+      "step": 21034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654462682739164e-05,
+      "loss": 0.3758,
+      "step": 21035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654398563734291e-05,
+      "loss": 0.5223,
+      "step": 21036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654334444729418e-05,
+      "loss": 0.4877,
+      "step": 21037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654270325724545e-05,
+      "loss": 0.4562,
+      "step": 21038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654206206719672e-05,
+      "loss": 0.4767,
+      "step": 21039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654142087714799e-05,
+      "loss": 0.5185,
+      "step": 21040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654077968709926e-05,
+      "loss": 0.4409,
+      "step": 21041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654013849705053e-05,
+      "loss": 0.5255,
+      "step": 21042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6539497307001796e-05,
+      "loss": 0.5706,
+      "step": 21043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653885611695307e-05,
+      "loss": 0.4845,
+      "step": 21044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653821492690433e-05,
+      "loss": 0.4706,
+      "step": 21045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6537573736855604e-05,
+      "loss": 0.5734,
+      "step": 21046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6536932546806876e-05,
+      "loss": 0.4451,
+      "step": 21047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653629135675815e-05,
+      "loss": 0.5485,
+      "step": 21048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653565016670941e-05,
+      "loss": 0.4845,
+      "step": 21049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6535008976660684e-05,
+      "loss": 0.3635,
+      "step": 21050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6534367786611956e-05,
+      "loss": 0.47,
+      "step": 21051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653372659656322e-05,
+      "loss": 0.5354,
+      "step": 21052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653308540651449e-05,
+      "loss": 0.4554,
+      "step": 21053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653244421646576e-05,
+      "loss": 0.4829,
+      "step": 21054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6531803026417036e-05,
+      "loss": 0.4679,
+      "step": 21055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65311618363683e-05,
+      "loss": 0.5078,
+      "step": 21056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653052064631957e-05,
+      "loss": 0.528,
+      "step": 21057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652987945627084e-05,
+      "loss": 0.5009,
+      "step": 21058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652923826622211e-05,
+      "loss": 0.5237,
+      "step": 21059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652859707617338e-05,
+      "loss": 0.4181,
+      "step": 21060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6527955886124646e-05,
+      "loss": 0.4376,
+      "step": 21061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652731469607592e-05,
+      "loss": 0.5488,
+      "step": 21062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652667350602719e-05,
+      "loss": 0.5076,
+      "step": 21063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652603231597846e-05,
+      "loss": 0.6773,
+      "step": 21064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6525391125929726e-05,
+      "loss": 0.4767,
+      "step": 21065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6524749935881e-05,
+      "loss": 0.4761,
+      "step": 21066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652410874583226e-05,
+      "loss": 0.4099,
+      "step": 21067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6523467555783535e-05,
+      "loss": 0.4918,
+      "step": 21068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6522826365734806e-05,
+      "loss": 0.4868,
+      "step": 21069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652218517568607e-05,
+      "loss": 0.5656,
+      "step": 21070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652154398563735e-05,
+      "loss": 0.5064,
+      "step": 21071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6520902795588615e-05,
+      "loss": 0.4237,
+      "step": 21072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6520261605539886e-05,
+      "loss": 0.4284,
+      "step": 21073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651962041549115e-05,
+      "loss": 0.4191,
+      "step": 21074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651897922544242e-05,
+      "loss": 0.5406,
+      "step": 21075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651833803539369e-05,
+      "loss": 0.4981,
+      "step": 21076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651769684534496e-05,
+      "loss": 0.5233,
+      "step": 21077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651705565529623e-05,
+      "loss": 0.4693,
+      "step": 21078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65164144652475e-05,
+      "loss": 0.3778,
+      "step": 21079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6515773275198775e-05,
+      "loss": 0.3055,
+      "step": 21080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651513208515004e-05,
+      "loss": 0.535,
+      "step": 21081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651449089510131e-05,
+      "loss": 0.6207,
+      "step": 21082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6513849705052577e-05,
+      "loss": 0.6576,
+      "step": 21083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651320851500385e-05,
+      "loss": 0.5923,
+      "step": 21084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651256732495511e-05,
+      "loss": 0.4279,
+      "step": 21085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6511926134906385e-05,
+      "loss": 0.4467,
+      "step": 21086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6511284944857657e-05,
+      "loss": 0.5267,
+      "step": 21087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651064375480893e-05,
+      "loss": 0.5536,
+      "step": 21088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65100025647602e-05,
+      "loss": 0.5004,
+      "step": 21089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6509361374711465e-05,
+      "loss": 0.4588,
+      "step": 21090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650872018466274e-05,
+      "loss": 0.6669,
+      "step": 21091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6508078994614e-05,
+      "loss": 0.4604,
+      "step": 21092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650743780456527e-05,
+      "loss": 0.5107,
+      "step": 21093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6506796614516545e-05,
+      "loss": 0.4602,
+      "step": 21094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650615542446782e-05,
+      "loss": 0.3994,
+      "step": 21095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650551423441909e-05,
+      "loss": 0.519,
+      "step": 21096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6504873044370353e-05,
+      "loss": 0.4432,
+      "step": 21097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6504231854321625e-05,
+      "loss": 0.5045,
+      "step": 21098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650359066427289e-05,
+      "loss": 0.4954,
+      "step": 21099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650294947422416e-05,
+      "loss": 0.5464,
+      "step": 21100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650230828417543e-05,
+      "loss": 0.5078,
+      "step": 21101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65016670941267e-05,
+      "loss": 0.467,
+      "step": 21102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650102590407797e-05,
+      "loss": 0.5862,
+      "step": 21103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650038471402924e-05,
+      "loss": 0.4366,
+      "step": 21104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6499743523980514e-05,
+      "loss": 0.4978,
+      "step": 21105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649910233393178e-05,
+      "loss": 0.5823,
+      "step": 21106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649846114388305e-05,
+      "loss": 0.4957,
+      "step": 21107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6497819953834315e-05,
+      "loss": 0.546,
+      "step": 21108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649717876378559e-05,
+      "loss": 0.4148,
+      "step": 21109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649653757373686e-05,
+      "loss": 0.4491,
+      "step": 21110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649589638368813e-05,
+      "loss": 0.4857,
+      "step": 21111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6495255193639395e-05,
+      "loss": 0.5755,
+      "step": 21112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649461400359067e-05,
+      "loss": 0.4463,
+      "step": 21113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649397281354194e-05,
+      "loss": 0.4693,
+      "step": 21114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6493331623493204e-05,
+      "loss": 0.5164,
+      "step": 21115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6492690433444475e-05,
+      "loss": 0.53,
+      "step": 21116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649204924339574e-05,
+      "loss": 0.5679,
+      "step": 21117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649140805334701e-05,
+      "loss": 0.4734,
+      "step": 21118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6490766863298284e-05,
+      "loss": 0.5064,
+      "step": 21119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6490125673249555e-05,
+      "loss": 0.5405,
+      "step": 21120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648948448320082e-05,
+      "loss": 0.465,
+      "step": 21121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648884329315209e-05,
+      "loss": 0.4327,
+      "step": 21122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6488202103103364e-05,
+      "loss": 0.5769,
+      "step": 21123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648756091305463e-05,
+      "loss": 0.5865,
+      "step": 21124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.64869197230059e-05,
+      "loss": 0.4804,
+      "step": 21125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6486278532957165e-05,
+      "loss": 0.4617,
+      "step": 21126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6485637342908444e-05,
+      "loss": 0.6427,
+      "step": 21127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648499615285971e-05,
+      "loss": 0.4148,
+      "step": 21128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648435496281098e-05,
+      "loss": 0.4505,
+      "step": 21129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6483713772762246e-05,
+      "loss": 0.4559,
+      "step": 21130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648307258271352e-05,
+      "loss": 0.403,
+      "step": 21131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.648243139266479e-05,
+      "loss": 0.4758,
+      "step": 21132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6481790202616054e-05,
+      "loss": 0.4856,
+      "step": 21133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6481149012567326e-05,
+      "loss": 0.4694,
+      "step": 21134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.64805078225186e-05,
+      "loss": 0.5233,
+      "step": 21135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647986663246987e-05,
+      "loss": 0.5232,
+      "step": 21136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6479225442421134e-05,
+      "loss": 0.7223,
+      "step": 21137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6478584252372406e-05,
+      "loss": 0.3198,
+      "step": 21138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647794306232367e-05,
+      "loss": 0.4952,
+      "step": 21139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647730187227494e-05,
+      "loss": 0.501,
+      "step": 21140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6476660682226214e-05,
+      "loss": 0.5204,
+      "step": 21141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647601949217748e-05,
+      "loss": 0.5256,
+      "step": 21142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647537830212876e-05,
+      "loss": 0.4484,
+      "step": 21143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647473711208002e-05,
+      "loss": 0.5026,
+      "step": 21144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6474095922031294e-05,
+      "loss": 0.4265,
+      "step": 21145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647345473198256e-05,
+      "loss": 0.3949,
+      "step": 21146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647281354193383e-05,
+      "loss": 0.4285,
+      "step": 21147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6472172351885096e-05,
+      "loss": 0.4726,
+      "step": 21148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647153116183637e-05,
+      "loss": 0.5445,
+      "step": 21149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647088997178764e-05,
+      "loss": 0.5919,
+      "step": 21150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647024878173891e-05,
+      "loss": 0.619,
+      "step": 21151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646960759169018e-05,
+      "loss": 0.4186,
+      "step": 21152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646896640164145e-05,
+      "loss": 0.5818,
+      "step": 21153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646832521159272e-05,
+      "loss": 0.4795,
+      "step": 21154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6467684021543984e-05,
+      "loss": 0.5009,
+      "step": 21155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6467042831495256e-05,
+      "loss": 0.474,
+      "step": 21156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646640164144653e-05,
+      "loss": 0.4528,
+      "step": 21157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646576045139779e-05,
+      "loss": 0.4144,
+      "step": 21158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646511926134907e-05,
+      "loss": 0.5571,
+      "step": 21159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6464478071300336e-05,
+      "loss": 0.5234,
+      "step": 21160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646383688125161e-05,
+      "loss": 0.6214,
+      "step": 21161
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646319569120287e-05,
+      "loss": 0.4679,
+      "step": 21162
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6462554501154144e-05,
+      "loss": 0.4768,
+      "step": 21163
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646191331110541e-05,
+      "loss": 0.468,
+      "step": 21164
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646127212105668e-05,
+      "loss": 0.5692,
+      "step": 21165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646063093100795e-05,
+      "loss": 0.6387,
+      "step": 21166
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6459989740959225e-05,
+      "loss": 0.5026,
+      "step": 21167
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6459348550910496e-05,
+      "loss": 0.6306,
+      "step": 21168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645870736086176e-05,
+      "loss": 0.5095,
+      "step": 21169
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645806617081303e-05,
+      "loss": 0.4994,
+      "step": 21170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.64574249807643e-05,
+      "loss": 0.5904,
+      "step": 21171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645678379071557e-05,
+      "loss": 0.5922,
+      "step": 21172
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6456142600666834e-05,
+      "loss": 0.4125,
+      "step": 21173
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6455501410618106e-05,
+      "loss": 0.5994,
+      "step": 21174
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645486022056938e-05,
+      "loss": 0.5657,
+      "step": 21175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645421903052065e-05,
+      "loss": 0.5825,
+      "step": 21176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645357784047192e-05,
+      "loss": 0.5213,
+      "step": 21177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6452936650423186e-05,
+      "loss": 0.4199,
+      "step": 21178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645229546037446e-05,
+      "loss": 0.576,
+      "step": 21179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.645165427032572e-05,
+      "loss": 0.5239,
+      "step": 21180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6451013080276995e-05,
+      "loss": 0.6281,
+      "step": 21181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6450371890228266e-05,
+      "loss": 0.5312,
+      "step": 21182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644973070017954e-05,
+      "loss": 0.4697,
+      "step": 21183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644908951013081e-05,
+      "loss": 0.3999,
+      "step": 21184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6448448320082075e-05,
+      "loss": 0.4753,
+      "step": 21185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6447807130033346e-05,
+      "loss": 0.5295,
+      "step": 21186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644716593998461e-05,
+      "loss": 0.5282,
+      "step": 21187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644652474993588e-05,
+      "loss": 0.4985,
+      "step": 21188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644588355988715e-05,
+      "loss": 0.4874,
+      "step": 21189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644524236983842e-05,
+      "loss": 0.5568,
+      "step": 21190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644460117978969e-05,
+      "loss": 0.5765,
+      "step": 21191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644395998974096e-05,
+      "loss": 0.4523,
+      "step": 21192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6443318799692235e-05,
+      "loss": 0.4967,
+      "step": 21193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.64426776096435e-05,
+      "loss": 0.4925,
+      "step": 21194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644203641959477e-05,
+      "loss": 0.4554,
+      "step": 21195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6441395229546037e-05,
+      "loss": 0.606,
+      "step": 21196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644075403949731e-05,
+      "loss": 0.5247,
+      "step": 21197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644011284944858e-05,
+      "loss": 0.5385,
+      "step": 21198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643947165939985e-05,
+      "loss": 0.4652,
+      "step": 21199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643883046935112e-05,
+      "loss": 0.6022,
+      "step": 21200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643818927930239e-05,
+      "loss": 0.4768,
+      "step": 21201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643754808925366e-05,
+      "loss": 0.4578,
+      "step": 21202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6436906899204925e-05,
+      "loss": 0.4731,
+      "step": 21203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64362657091562e-05,
+      "loss": 0.4126,
+      "step": 21204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643562451910746e-05,
+      "loss": 0.5366,
+      "step": 21205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643498332905873e-05,
+      "loss": 0.5233,
+      "step": 21206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6434342139010005e-05,
+      "loss": 0.5896,
+      "step": 21207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643370094896128e-05,
+      "loss": 0.3822,
+      "step": 21208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643305975891254e-05,
+      "loss": 0.4421,
+      "step": 21209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6432418568863813e-05,
+      "loss": 0.5013,
+      "step": 21210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6431777378815085e-05,
+      "loss": 0.5032,
+      "step": 21211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643113618876635e-05,
+      "loss": 0.4031,
+      "step": 21212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643049499871762e-05,
+      "loss": 0.5013,
+      "step": 21213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642985380866889e-05,
+      "loss": 0.4674,
+      "step": 21214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6429212618620165e-05,
+      "loss": 0.4777,
+      "step": 21215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642857142857143e-05,
+      "loss": 0.5388,
+      "step": 21216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64279302385227e-05,
+      "loss": 0.5159,
+      "step": 21217
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642728904847397e-05,
+      "loss": 0.5049,
+      "step": 21218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642664785842524e-05,
+      "loss": 0.4691,
+      "step": 21219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642600666837651e-05,
+      "loss": 0.4375,
+      "step": 21220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6425365478327775e-05,
+      "loss": 0.4781,
+      "step": 21221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642472428827905e-05,
+      "loss": 0.5945,
+      "step": 21222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642408309823032e-05,
+      "loss": 0.4421,
+      "step": 21223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642344190818159e-05,
+      "loss": 0.4851,
+      "step": 21224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6422800718132855e-05,
+      "loss": 0.5876,
+      "step": 21225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642215952808413e-05,
+      "loss": 0.6373,
+      "step": 21226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642151833803539e-05,
+      "loss": 0.5682,
+      "step": 21227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6420877147986664e-05,
+      "loss": 0.4753,
+      "step": 21228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6420235957937935e-05,
+      "loss": 0.4714,
+      "step": 21229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64195947678892e-05,
+      "loss": 0.4918,
+      "step": 21230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641895357784048e-05,
+      "loss": 0.5962,
+      "step": 21231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6418312387791744e-05,
+      "loss": 0.5606,
+      "step": 21232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6417671197743016e-05,
+      "loss": 0.4884,
+      "step": 21233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641703000769428e-05,
+      "loss": 0.5505,
+      "step": 21234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641638881764555e-05,
+      "loss": 0.5275,
+      "step": 21235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641574762759682e-05,
+      "loss": 0.4796,
+      "step": 21236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641510643754809e-05,
+      "loss": 0.4066,
+      "step": 21237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641446524749936e-05,
+      "loss": 0.4401,
+      "step": 21238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641382405745063e-05,
+      "loss": 0.4845,
+      "step": 21239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6413182867401904e-05,
+      "loss": 0.5201,
+      "step": 21240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641254167735317e-05,
+      "loss": 0.4754,
+      "step": 21241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641190048730444e-05,
+      "loss": 0.4764,
+      "step": 21242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6411259297255706e-05,
+      "loss": 0.5524,
+      "step": 21243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641061810720698e-05,
+      "loss": 0.4531,
+      "step": 21244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640997691715824e-05,
+      "loss": 0.4682,
+      "step": 21245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6409335727109514e-05,
+      "loss": 0.5502,
+      "step": 21246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6408694537060786e-05,
+      "loss": 0.4941,
+      "step": 21247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640805334701206e-05,
+      "loss": 0.494,
+      "step": 21248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640741215696333e-05,
+      "loss": 0.5037,
+      "step": 21249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6406770966914594e-05,
+      "loss": 0.627,
+      "step": 21250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6406129776865866e-05,
+      "loss": 0.4336,
+      "step": 21251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640548858681713e-05,
+      "loss": 0.5805,
+      "step": 21252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64048473967684e-05,
+      "loss": 0.5405,
+      "step": 21253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6404206206719674e-05,
+      "loss": 0.6005,
+      "step": 21254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6403565016670946e-05,
+      "loss": 0.4348,
+      "step": 21255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640292382662222e-05,
+      "loss": 0.4751,
+      "step": 21256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640228263657348e-05,
+      "loss": 0.5367,
+      "step": 21257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6401641446524754e-05,
+      "loss": 0.4562,
+      "step": 21258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640100025647602e-05,
+      "loss": 0.5924,
+      "step": 21259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640035906642729e-05,
+      "loss": 0.3893,
+      "step": 21260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6399717876378556e-05,
+      "loss": 0.5289,
+      "step": 21261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639907668632983e-05,
+      "loss": 0.4124,
+      "step": 21262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63984354962811e-05,
+      "loss": 0.4334,
+      "step": 21263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639779430623237e-05,
+      "loss": 0.4517,
+      "step": 21264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639715311618364e-05,
+      "loss": 0.4868,
+      "step": 21265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639651192613491e-05,
+      "loss": 0.5154,
+      "step": 21266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639587073608618e-05,
+      "loss": 0.5721,
+      "step": 21267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6395229546037444e-05,
+      "loss": 0.4224,
+      "step": 21268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6394588355988716e-05,
+      "loss": 0.4789,
+      "step": 21269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639394716593999e-05,
+      "loss": 0.5674,
+      "step": 21270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639330597589126e-05,
+      "loss": 0.607,
+      "step": 21271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6392664785842524e-05,
+      "loss": 0.4678,
+      "step": 21272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6392023595793796e-05,
+      "loss": 0.6064,
+      "step": 21273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639138240574507e-05,
+      "loss": 0.4693,
+      "step": 21274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639074121569633e-05,
+      "loss": 0.4816,
+      "step": 21275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6390100025647604e-05,
+      "loss": 0.4292,
+      "step": 21276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638945883559887e-05,
+      "loss": 0.4231,
+      "step": 21277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638881764555014e-05,
+      "loss": 0.5333,
+      "step": 21278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638817645550141e-05,
+      "loss": 0.3957,
+      "step": 21279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6387535265452685e-05,
+      "loss": 0.374,
+      "step": 21280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638689407540395e-05,
+      "loss": 0.577,
+      "step": 21281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638625288535522e-05,
+      "loss": 0.5367,
+      "step": 21282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638561169530649e-05,
+      "loss": 0.4627,
+      "step": 21283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638497050525776e-05,
+      "loss": 0.5354,
+      "step": 21284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638432931520903e-05,
+      "loss": 0.5803,
+      "step": 21285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6383688125160295e-05,
+      "loss": 0.681,
+      "step": 21286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638304693511157e-05,
+      "loss": 0.5358,
+      "step": 21287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638240574506284e-05,
+      "loss": 0.4977,
+      "step": 21288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638176455501411e-05,
+      "loss": 0.5679,
+      "step": 21289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6381123364965375e-05,
+      "loss": 0.5845,
+      "step": 21290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6380482174916646e-05,
+      "loss": 0.6409,
+      "step": 21291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637984098486792e-05,
+      "loss": 0.6395,
+      "step": 21292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637919979481918e-05,
+      "loss": 0.5203,
+      "step": 21293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6378558604770455e-05,
+      "loss": 0.5118,
+      "step": 21294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6377917414721726e-05,
+      "loss": 0.5482,
+      "step": 21295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6377276224673e-05,
+      "loss": 0.4821,
+      "step": 21296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637663503462426e-05,
+      "loss": 0.6018,
+      "step": 21297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6375993844575535e-05,
+      "loss": 0.4268,
+      "step": 21298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63753526545268e-05,
+      "loss": 0.6697,
+      "step": 21299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637471146447807e-05,
+      "loss": 0.5105,
+      "step": 21300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637407027442934e-05,
+      "loss": 0.463,
+      "step": 21301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637342908438061e-05,
+      "loss": 0.5288,
+      "step": 21302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6372787894331887e-05,
+      "loss": 0.5785,
+      "step": 21303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637214670428315e-05,
+      "loss": 0.447,
+      "step": 21304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637150551423442e-05,
+      "loss": 0.4968,
+      "step": 21305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637086432418569e-05,
+      "loss": 0.4217,
+      "step": 21306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637022313413696e-05,
+      "loss": 0.5919,
+      "step": 21307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6369581944088225e-05,
+      "loss": 0.4667,
+      "step": 21308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6368940754039497e-05,
+      "loss": 0.4071,
+      "step": 21309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636829956399077e-05,
+      "loss": 0.4471,
+      "step": 21310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636765837394204e-05,
+      "loss": 0.5029,
+      "step": 21311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636701718389331e-05,
+      "loss": 0.4676,
+      "step": 21312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636637599384458e-05,
+      "loss": 0.6039,
+      "step": 21313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636573480379585e-05,
+      "loss": 0.598,
+      "step": 21314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636509361374711e-05,
+      "loss": 0.4986,
+      "step": 21315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6364452423698385e-05,
+      "loss": 0.5845,
+      "step": 21316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636381123364965e-05,
+      "loss": 0.5045,
+      "step": 21317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636317004360092e-05,
+      "loss": 0.5342,
+      "step": 21318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63625288535522e-05,
+      "loss": 0.5635,
+      "step": 21319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6361887663503465e-05,
+      "loss": 0.4905,
+      "step": 21320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636124647345474e-05,
+      "loss": 0.5265,
+      "step": 21321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6360605283406e-05,
+      "loss": 0.5362,
+      "step": 21322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6359964093357274e-05,
+      "loss": 0.6086,
+      "step": 21323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635932290330854e-05,
+      "loss": 0.495,
+      "step": 21324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635868171325981e-05,
+      "loss": 0.399,
+      "step": 21325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635804052321108e-05,
+      "loss": 0.5252,
+      "step": 21326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6357399333162354e-05,
+      "loss": 0.5451,
+      "step": 21327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6356758143113625e-05,
+      "loss": 0.5975,
+      "step": 21328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635611695306489e-05,
+      "loss": 0.5656,
+      "step": 21329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635547576301616e-05,
+      "loss": 0.4758,
+      "step": 21330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635483457296743e-05,
+      "loss": 0.4176,
+      "step": 21331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63541933829187e-05,
+      "loss": 0.4831,
+      "step": 21332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6353552192869964e-05,
+      "loss": 0.4673,
+      "step": 21333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6352911002821235e-05,
+      "loss": 0.5907,
+      "step": 21334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635226981277251e-05,
+      "loss": 0.4632,
+      "step": 21335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635162862272378e-05,
+      "loss": 0.4933,
+      "step": 21336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635098743267505e-05,
+      "loss": 0.5551,
+      "step": 21337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6350346242626315e-05,
+      "loss": 0.4958,
+      "step": 21338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634970505257759e-05,
+      "loss": 0.4641,
+      "step": 21339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634906386252885e-05,
+      "loss": 0.4145,
+      "step": 21340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6348422672480124e-05,
+      "loss": 0.4871,
+      "step": 21341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6347781482431395e-05,
+      "loss": 0.5924,
+      "step": 21342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634714029238267e-05,
+      "loss": 0.4414,
+      "step": 21343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634649910233394e-05,
+      "loss": 0.3702,
+      "step": 21344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6345857912285204e-05,
+      "loss": 0.4449,
+      "step": 21345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6345216722236476e-05,
+      "loss": 0.4281,
+      "step": 21346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634457553218774e-05,
+      "loss": 0.5529,
+      "step": 21347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634393434213901e-05,
+      "loss": 0.5074,
+      "step": 21348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634329315209028e-05,
+      "loss": 0.5577,
+      "step": 21349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634265196204155e-05,
+      "loss": 0.5139,
+      "step": 21350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634201077199282e-05,
+      "loss": 0.5787,
+      "step": 21351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634136958194409e-05,
+      "loss": 0.6032,
+      "step": 21352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6340728391895364e-05,
+      "loss": 0.5806,
+      "step": 21353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634008720184663e-05,
+      "loss": 0.513,
+      "step": 21354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63394460117979e-05,
+      "loss": 0.4677,
+      "step": 21355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6338804821749166e-05,
+      "loss": 0.6009,
+      "step": 21356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633816363170044e-05,
+      "loss": 0.4944,
+      "step": 21357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633752244165171e-05,
+      "loss": 0.4789,
+      "step": 21358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633688125160298e-05,
+      "loss": 0.4488,
+      "step": 21359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6336240061554246e-05,
+      "loss": 0.508,
+      "step": 21360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633559887150552e-05,
+      "loss": 0.4924,
+      "step": 21361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633495768145679e-05,
+      "loss": 0.5083,
+      "step": 21362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6334316491408054e-05,
+      "loss": 0.4475,
+      "step": 21363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6333675301359326e-05,
+      "loss": 0.5102,
+      "step": 21364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633303411131059e-05,
+      "loss": 0.4011,
+      "step": 21365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633239292126186e-05,
+      "loss": 0.4537,
+      "step": 21366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6331751731213134e-05,
+      "loss": 0.4647,
+      "step": 21367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6331110541164406e-05,
+      "loss": 0.5478,
+      "step": 21368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633046935111567e-05,
+      "loss": 0.4281,
+      "step": 21369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632982816106694e-05,
+      "loss": 0.4661,
+      "step": 21370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6329186971018214e-05,
+      "loss": 0.4817,
+      "step": 21371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632854578096948e-05,
+      "loss": 0.433,
+      "step": 21372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632790459092075e-05,
+      "loss": 0.5247,
+      "step": 21373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6327263400872016e-05,
+      "loss": 0.4914,
+      "step": 21374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6326622210823294e-05,
+      "loss": 0.4769,
+      "step": 21375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632598102077456e-05,
+      "loss": 0.457,
+      "step": 21376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632533983072583e-05,
+      "loss": 0.4021,
+      "step": 21377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6324698640677096e-05,
+      "loss": 0.443,
+      "step": 21378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632405745062837e-05,
+      "loss": 0.4965,
+      "step": 21379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632341626057964e-05,
+      "loss": 0.5367,
+      "step": 21380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6322775070530904e-05,
+      "loss": 0.4386,
+      "step": 21381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6322133880482176e-05,
+      "loss": 0.461,
+      "step": 21382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632149269043345e-05,
+      "loss": 0.4876,
+      "step": 21383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632085150038472e-05,
+      "loss": 0.3662,
+      "step": 21384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6320210310335984e-05,
+      "loss": 0.5042,
+      "step": 21385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6319569120287256e-05,
+      "loss": 0.4486,
+      "step": 21386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631892793023852e-05,
+      "loss": 0.5418,
+      "step": 21387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631828674018979e-05,
+      "loss": 0.5932,
+      "step": 21388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6317645550141065e-05,
+      "loss": 0.5545,
+      "step": 21389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631700436009233e-05,
+      "loss": 0.4457,
+      "step": 21390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631636317004361e-05,
+      "loss": 0.7139,
+      "step": 21391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631572197999487e-05,
+      "loss": 0.4742,
+      "step": 21392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6315080789946145e-05,
+      "loss": 0.5928,
+      "step": 21393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631443959989741e-05,
+      "loss": 0.5793,
+      "step": 21394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631379840984868e-05,
+      "loss": 0.6123,
+      "step": 21395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6313157219799946e-05,
+      "loss": 0.5104,
+      "step": 21396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631251602975122e-05,
+      "loss": 0.5381,
+      "step": 21397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631187483970249e-05,
+      "loss": 0.5061,
+      "step": 21398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631123364965376e-05,
+      "loss": 0.589,
+      "step": 21399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631059245960503e-05,
+      "loss": 0.5237,
+      "step": 21400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63099512695563e-05,
+      "loss": 0.5872,
+      "step": 21401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630931007950757e-05,
+      "loss": 0.5408,
+      "step": 21402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6308668889458835e-05,
+      "loss": 0.5092,
+      "step": 21403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6308027699410106e-05,
+      "loss": 0.491,
+      "step": 21404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630738650936137e-05,
+      "loss": 0.5849,
+      "step": 21405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630674531931264e-05,
+      "loss": 0.5147,
+      "step": 21406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6306104129263915e-05,
+      "loss": 0.471,
+      "step": 21407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6305462939215186e-05,
+      "loss": 0.514,
+      "step": 21408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630482174916646e-05,
+      "loss": 0.4831,
+      "step": 21409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630418055911772e-05,
+      "loss": 0.5592,
+      "step": 21410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6303539369068995e-05,
+      "loss": 0.4583,
+      "step": 21411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630289817902026e-05,
+      "loss": 0.5195,
+      "step": 21412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630225698897153e-05,
+      "loss": 0.5227,
+      "step": 21413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63016157989228e-05,
+      "loss": 0.5123,
+      "step": 21414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6300974608874075e-05,
+      "loss": 0.5467,
+      "step": 21415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630033341882535e-05,
+      "loss": 0.4083,
+      "step": 21416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629969222877661e-05,
+      "loss": 0.4378,
+      "step": 21417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629905103872788e-05,
+      "loss": 0.4546,
+      "step": 21418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629840984867915e-05,
+      "loss": 0.5236,
+      "step": 21419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629776865863042e-05,
+      "loss": 0.3941,
+      "step": 21420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6297127468581685e-05,
+      "loss": 0.4329,
+      "step": 21421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6296486278532957e-05,
+      "loss": 0.5659,
+      "step": 21422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629584508848423e-05,
+      "loss": 0.5395,
+      "step": 21423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.62952038984355e-05,
+      "loss": 0.47,
+      "step": 21424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629456270838677e-05,
+      "loss": 0.5414,
+      "step": 21425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629392151833804e-05,
+      "loss": 0.5949,
+      "step": 21426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629328032828931e-05,
+      "loss": 0.5205,
+      "step": 21427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629263913824057e-05,
+      "loss": 0.4355,
+      "step": 21428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6291997948191845e-05,
+      "loss": 0.5024,
+      "step": 21429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629135675814312e-05,
+      "loss": 0.5067,
+      "step": 21430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629071556809439e-05,
+      "loss": 0.5353,
+      "step": 21431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6290074378045653e-05,
+      "loss": 0.5324,
+      "step": 21432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6289433187996925e-05,
+      "loss": 0.515,
+      "step": 21433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.62887919979482e-05,
+      "loss": 0.526,
+      "step": 21434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628815080789946e-05,
+      "loss": 0.4719,
+      "step": 21435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6287509617850734e-05,
+      "loss": 0.4199,
+      "step": 21436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6286868427802e-05,
+      "loss": 0.5191,
+      "step": 21437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628622723775327e-05,
+      "loss": 0.4197,
+      "step": 21438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628558604770454e-05,
+      "loss": 0.5624,
+      "step": 21439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6284944857655814e-05,
+      "loss": 0.4061,
+      "step": 21440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628430366760708e-05,
+      "loss": 0.5054,
+      "step": 21441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628366247755835e-05,
+      "loss": 0.5757,
+      "step": 21442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628302128750962e-05,
+      "loss": 0.5925,
+      "step": 21443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628238009746089e-05,
+      "loss": 0.4584,
+      "step": 21444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628173890741216e-05,
+      "loss": 0.6803,
+      "step": 21445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628109771736343e-05,
+      "loss": 0.4636,
+      "step": 21446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.62804565273147e-05,
+      "loss": 0.562,
+      "step": 21447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627981533726597e-05,
+      "loss": 0.59,
+      "step": 21448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627917414721724e-05,
+      "loss": 0.5914,
+      "step": 21449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6278532957168504e-05,
+      "loss": 0.4236,
+      "step": 21450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6277891767119775e-05,
+      "loss": 0.489,
+      "step": 21451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627725057707105e-05,
+      "loss": 0.4849,
+      "step": 21452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627660938702231e-05,
+      "loss": 0.5433,
+      "step": 21453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6275968196973584e-05,
+      "loss": 0.5628,
+      "step": 21454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6275327006924855e-05,
+      "loss": 0.5396,
+      "step": 21455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627468581687613e-05,
+      "loss": 0.4594,
+      "step": 21456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627404462682739e-05,
+      "loss": 0.5928,
+      "step": 21457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6273403436778664e-05,
+      "loss": 0.4764,
+      "step": 21458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627276224672993e-05,
+      "loss": 0.5267,
+      "step": 21459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62721210566812e-05,
+      "loss": 0.6371,
+      "step": 21460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627147986663247e-05,
+      "loss": 0.5489,
+      "step": 21461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627083867658374e-05,
+      "loss": 0.4193,
+      "step": 21462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6270197486535016e-05,
+      "loss": 0.5362,
+      "step": 21463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626955629648628e-05,
+      "loss": 0.4529,
+      "step": 21464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626891510643755e-05,
+      "loss": 0.4425,
+      "step": 21465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626827391638882e-05,
+      "loss": 0.474,
+      "step": 21466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626763272634009e-05,
+      "loss": 0.4375,
+      "step": 21467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6266991536291354e-05,
+      "loss": 0.5047,
+      "step": 21468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6266350346242626e-05,
+      "loss": 0.5742,
+      "step": 21469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62657091561939e-05,
+      "loss": 0.5168,
+      "step": 21470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626506796614517e-05,
+      "loss": 0.5781,
+      "step": 21471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626442677609644e-05,
+      "loss": 0.4619,
+      "step": 21472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6263785586047706e-05,
+      "loss": 0.4883,
+      "step": 21473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626314439599898e-05,
+      "loss": 0.5607,
+      "step": 21474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626250320595024e-05,
+      "loss": 0.5228,
+      "step": 21475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6261862015901514e-05,
+      "loss": 0.5868,
+      "step": 21476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626122082585278e-05,
+      "loss": 0.5002,
+      "step": 21477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626057963580405e-05,
+      "loss": 0.4843,
+      "step": 21478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625993844575533e-05,
+      "loss": 0.4826,
+      "step": 21479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6259297255706594e-05,
+      "loss": 0.4834,
+      "step": 21480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6258656065657866e-05,
+      "loss": 0.3979,
+      "step": 21481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625801487560913e-05,
+      "loss": 0.4056,
+      "step": 21482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62573736855604e-05,
+      "loss": 0.5353,
+      "step": 21483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625673249551167e-05,
+      "loss": 0.5623,
+      "step": 21484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625609130546294e-05,
+      "loss": 0.4835,
+      "step": 21485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625545011541421e-05,
+      "loss": 0.4896,
+      "step": 21486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625480892536548e-05,
+      "loss": 0.6278,
+      "step": 21487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6254167735316754e-05,
+      "loss": 0.5536,
+      "step": 21488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625352654526802e-05,
+      "loss": 0.4682,
+      "step": 21489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625288535521929e-05,
+      "loss": 0.4862,
+      "step": 21490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6252244165170556e-05,
+      "loss": 0.5422,
+      "step": 21491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625160297512183e-05,
+      "loss": 0.3402,
+      "step": 21492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625096178507309e-05,
+      "loss": 0.4271,
+      "step": 21493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6250320595024364e-05,
+      "loss": 0.6256,
+      "step": 21494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6249679404975636e-05,
+      "loss": 0.4155,
+      "step": 21495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624903821492691e-05,
+      "loss": 0.4427,
+      "step": 21496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624839702487818e-05,
+      "loss": 0.2692,
+      "step": 21497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6247755834829444e-05,
+      "loss": 0.5003,
+      "step": 21498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6247114644780716e-05,
+      "loss": 0.5066,
+      "step": 21499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624647345473198e-05,
+      "loss": 0.5729,
+      "step": 21500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624583226468325e-05,
+      "loss": 0.5088,
+      "step": 21501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6245191074634525e-05,
+      "loss": 0.5086,
+      "step": 21502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6244549884585796e-05,
+      "loss": 0.6588,
+      "step": 21503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624390869453706e-05,
+      "loss": 0.3945,
+      "step": 21504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624326750448833e-05,
+      "loss": 0.3742,
+      "step": 21505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6242626314439605e-05,
+      "loss": 0.4754,
+      "step": 21506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624198512439087e-05,
+      "loss": 0.5571,
+      "step": 21507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624134393434214e-05,
+      "loss": 0.5991,
+      "step": 21508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6240702744293406e-05,
+      "loss": 0.5281,
+      "step": 21509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624006155424468e-05,
+      "loss": 0.521,
+      "step": 21510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623942036419595e-05,
+      "loss": 0.3713,
+      "step": 21511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623877917414722e-05,
+      "loss": 0.4834,
+      "step": 21512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6238137984098486e-05,
+      "loss": 0.3851,
+      "step": 21513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623749679404976e-05,
+      "loss": 0.4486,
+      "step": 21514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623685560400103e-05,
+      "loss": 0.5805,
+      "step": 21515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6236214413952295e-05,
+      "loss": 0.536,
+      "step": 21516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6235573223903566e-05,
+      "loss": 0.4691,
+      "step": 21517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623493203385484e-05,
+      "loss": 0.4455,
+      "step": 21518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623429084380611e-05,
+      "loss": 0.5156,
+      "step": 21519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6233649653757375e-05,
+      "loss": 0.4573,
+      "step": 21520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6233008463708646e-05,
+      "loss": 0.5058,
+      "step": 21521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623236727365991e-05,
+      "loss": 0.483,
+      "step": 21522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623172608361118e-05,
+      "loss": 0.5835,
+      "step": 21523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6231084893562455e-05,
+      "loss": 0.6338,
+      "step": 21524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623044370351372e-05,
+      "loss": 0.4113,
+      "step": 21525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622980251346499e-05,
+      "loss": 0.5649,
+      "step": 21526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622916132341626e-05,
+      "loss": 0.4438,
+      "step": 21527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6228520133367535e-05,
+      "loss": 0.3891,
+      "step": 21528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62278789433188e-05,
+      "loss": 0.5431,
+      "step": 21529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622723775327007e-05,
+      "loss": 0.4514,
+      "step": 21530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6226596563221337e-05,
+      "loss": 0.5032,
+      "step": 21531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622595537317261e-05,
+      "loss": 0.4404,
+      "step": 21532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622531418312388e-05,
+      "loss": 0.502,
+      "step": 21533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6224672993075145e-05,
+      "loss": 0.4513,
+      "step": 21534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6224031803026423e-05,
+      "loss": 0.5104,
+      "step": 21535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622339061297769e-05,
+      "loss": 0.5524,
+      "step": 21536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622274942292896e-05,
+      "loss": 0.4936,
+      "step": 21537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6222108232880225e-05,
+      "loss": 0.4996,
+      "step": 21538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62214670428315e-05,
+      "loss": 0.4919,
+      "step": 21539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622082585278277e-05,
+      "loss": 0.4158,
+      "step": 21540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6220184662734033e-05,
+      "loss": 0.5038,
+      "step": 21541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6219543472685305e-05,
+      "loss": 0.6079,
+      "step": 21542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621890228263658e-05,
+      "loss": 0.5415,
+      "step": 21543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621826109258785e-05,
+      "loss": 0.5475,
+      "step": 21544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6217619902539113e-05,
+      "loss": 0.4772,
+      "step": 21545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6216978712490385e-05,
+      "loss": 0.5336,
+      "step": 21546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621633752244165e-05,
+      "loss": 0.5808,
+      "step": 21547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621569633239292e-05,
+      "loss": 0.431,
+      "step": 21548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6215055142344194e-05,
+      "loss": 0.4518,
+      "step": 21549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621441395229546e-05,
+      "loss": 0.5571,
+      "step": 21550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621377276224674e-05,
+      "loss": 0.635,
+      "step": 21551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6213131572198e-05,
+      "loss": 0.4979,
+      "step": 21552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6212490382149274e-05,
+      "loss": 0.4467,
+      "step": 21553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621184919210054e-05,
+      "loss": 0.314,
+      "step": 21554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621120800205181e-05,
+      "loss": 0.5231,
+      "step": 21555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6210566812003075e-05,
+      "loss": 0.4193,
+      "step": 21556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620992562195435e-05,
+      "loss": 0.5216,
+      "step": 21557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620928443190562e-05,
+      "loss": 0.4106,
+      "step": 21558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620864324185689e-05,
+      "loss": 0.5828,
+      "step": 21559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620800205180816e-05,
+      "loss": 0.4304,
+      "step": 21560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620736086175943e-05,
+      "loss": 0.4651,
+      "step": 21561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62067196717107e-05,
+      "loss": 0.5359,
+      "step": 21562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6206078481661964e-05,
+      "loss": 0.4889,
+      "step": 21563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6205437291613235e-05,
+      "loss": 0.4749,
+      "step": 21564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62047961015645e-05,
+      "loss": 0.4186,
+      "step": 21565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620415491151577e-05,
+      "loss": 0.5683,
+      "step": 21566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620351372146705e-05,
+      "loss": 0.5469,
+      "step": 21567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6202872531418316e-05,
+      "loss": 0.4376,
+      "step": 21568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620223134136959e-05,
+      "loss": 0.4976,
+      "step": 21569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620159015132085e-05,
+      "loss": 0.5801,
+      "step": 21570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6200948961272124e-05,
+      "loss": 0.5862,
+      "step": 21571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620030777122339e-05,
+      "loss": 0.4628,
+      "step": 21572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619966658117466e-05,
+      "loss": 0.6101,
+      "step": 21573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619902539112593e-05,
+      "loss": 0.4737,
+      "step": 21574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6198384201077204e-05,
+      "loss": 0.4841,
+      "step": 21575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6197743011028476e-05,
+      "loss": 0.5013,
+      "step": 21576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619710182097974e-05,
+      "loss": 0.4384,
+      "step": 21577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619646063093101e-05,
+      "loss": 0.6,
+      "step": 21578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619581944088228e-05,
+      "loss": 0.4563,
+      "step": 21579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619517825083355e-05,
+      "loss": 0.4332,
+      "step": 21580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6194537060784814e-05,
+      "loss": 0.5147,
+      "step": 21581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6193895870736086e-05,
+      "loss": 0.4322,
+      "step": 21582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619325468068736e-05,
+      "loss": 0.4051,
+      "step": 21583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619261349063863e-05,
+      "loss": 0.5585,
+      "step": 21584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.61919723005899e-05,
+      "loss": 0.4289,
+      "step": 21585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6191331110541166e-05,
+      "loss": 0.4897,
+      "step": 21586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619068992049244e-05,
+      "loss": 0.4522,
+      "step": 21587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.61900487304437e-05,
+      "loss": 0.57,
+      "step": 21588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6189407540394974e-05,
+      "loss": 0.4246,
+      "step": 21589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6188766350346246e-05,
+      "loss": 0.5123,
+      "step": 21590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618812516029752e-05,
+      "loss": 0.6214,
+      "step": 21591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618748397024878e-05,
+      "loss": 0.4924,
+      "step": 21592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6186842780200054e-05,
+      "loss": 0.4624,
+      "step": 21593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6186201590151326e-05,
+      "loss": 0.5555,
+      "step": 21594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618556040010259e-05,
+      "loss": 0.3996,
+      "step": 21595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618491921005386e-05,
+      "loss": 0.4087,
+      "step": 21596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618427802000513e-05,
+      "loss": 0.423,
+      "step": 21597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.61836368299564e-05,
+      "loss": 0.5383,
+      "step": 21598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618299563990767e-05,
+      "loss": 0.4198,
+      "step": 21599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618235444985894e-05,
+      "loss": 0.6484,
+      "step": 21600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618171325981021e-05,
+      "loss": 0.5023,
+      "step": 21601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618107206976148e-05,
+      "loss": 0.5092,
+      "step": 21602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618043087971275e-05,
+      "loss": 0.4581,
+      "step": 21603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6179789689664016e-05,
+      "loss": 0.5341,
+      "step": 21604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617914849961529e-05,
+      "loss": 0.5687,
+      "step": 21605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617850730956656e-05,
+      "loss": 0.5077,
+      "step": 21606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617786611951783e-05,
+      "loss": 0.5232,
+      "step": 21607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6177224929469096e-05,
+      "loss": 0.4447,
+      "step": 21608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617658373942037e-05,
+      "loss": 0.491,
+      "step": 21609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617594254937163e-05,
+      "loss": 0.4971,
+      "step": 21610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6175301359322904e-05,
+      "loss": 0.4311,
+      "step": 21611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6174660169274176e-05,
+      "loss": 0.4186,
+      "step": 21612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617401897922544e-05,
+      "loss": 0.649,
+      "step": 21613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617337778917671e-05,
+      "loss": 0.4692,
+      "step": 21614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6172736599127985e-05,
+      "loss": 0.4382,
+      "step": 21615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6172095409079256e-05,
+      "loss": 0.5316,
+      "step": 21616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617145421903052e-05,
+      "loss": 0.4865,
+      "step": 21617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617081302898179e-05,
+      "loss": 0.4413,
+      "step": 21618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617017183893306e-05,
+      "loss": 0.6015,
+      "step": 21619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616953064888433e-05,
+      "loss": 0.6094,
+      "step": 21620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.61688894588356e-05,
+      "loss": 0.4217,
+      "step": 21621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6168248268786866e-05,
+      "loss": 0.5773,
+      "step": 21622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6167607078738145e-05,
+      "loss": 0.5083,
+      "step": 21623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616696588868941e-05,
+      "loss": 0.4248,
+      "step": 21624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616632469864068e-05,
+      "loss": 0.5074,
+      "step": 21625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6165683508591946e-05,
+      "loss": 0.4529,
+      "step": 21626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616504231854322e-05,
+      "loss": 0.5379,
+      "step": 21627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616440112849448e-05,
+      "loss": 0.4136,
+      "step": 21628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6163759938445755e-05,
+      "loss": 0.456,
+      "step": 21629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6163118748397026e-05,
+      "loss": 0.4956,
+      "step": 21630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.61624775583483e-05,
+      "loss": 0.497,
+      "step": 21631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616183636829957e-05,
+      "loss": 0.5668,
+      "step": 21632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6161195178250835e-05,
+      "loss": 0.5272,
+      "step": 21633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6160553988202107e-05,
+      "loss": 0.4435,
+      "step": 21634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615991279815337e-05,
+      "loss": 0.4793,
+      "step": 21635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615927160810464e-05,
+      "loss": 0.5472,
+      "step": 21636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615863041805591e-05,
+      "loss": 0.4811,
+      "step": 21637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615798922800718e-05,
+      "loss": 0.5615,
+      "step": 21638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615734803795846e-05,
+      "loss": 0.4481,
+      "step": 21639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615670684790972e-05,
+      "loss": 0.4767,
+      "step": 21640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6156065657860995e-05,
+      "loss": 0.5364,
+      "step": 21641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615542446781226e-05,
+      "loss": 0.433,
+      "step": 21642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615478327776353e-05,
+      "loss": 0.5759,
+      "step": 21643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6154142087714797e-05,
+      "loss": 0.5024,
+      "step": 21644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615350089766607e-05,
+      "loss": 0.5265,
+      "step": 21645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615285970761734e-05,
+      "loss": 0.5423,
+      "step": 21646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615221851756861e-05,
+      "loss": 0.4615,
+      "step": 21647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6151577327519883e-05,
+      "loss": 0.4772,
+      "step": 21648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615093613747115e-05,
+      "loss": 0.371,
+      "step": 21649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615029494742242e-05,
+      "loss": 0.389,
+      "step": 21650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6149653757373685e-05,
+      "loss": 0.5916,
+      "step": 21651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614901256732496e-05,
+      "loss": 0.5237,
+      "step": 21652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614837137727622e-05,
+      "loss": 0.5467,
+      "step": 21653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6147730187227493e-05,
+      "loss": 0.3556,
+      "step": 21654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6147088997178765e-05,
+      "loss": 0.4915,
+      "step": 21655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614644780713004e-05,
+      "loss": 0.4236,
+      "step": 21656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614580661708131e-05,
+      "loss": 0.4516,
+      "step": 21657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6145165427032574e-05,
+      "loss": 0.3992,
+      "step": 21658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6144524236983845e-05,
+      "loss": 0.6475,
+      "step": 21659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614388304693511e-05,
+      "loss": 0.5197,
+      "step": 21660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614324185688638e-05,
+      "loss": 0.4378,
+      "step": 21661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6142600666837654e-05,
+      "loss": 0.5142,
+      "step": 21662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6141959476788925e-05,
+      "loss": 0.4941,
+      "step": 21663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614131828674019e-05,
+      "loss": 0.4828,
+      "step": 21664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614067709669146e-05,
+      "loss": 0.4537,
+      "step": 21665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6140035906642734e-05,
+      "loss": 0.4608,
+      "step": 21666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6139394716594e-05,
+      "loss": 0.4366,
+      "step": 21667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613875352654527e-05,
+      "loss": 0.553,
+      "step": 21668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6138112336496535e-05,
+      "loss": 0.5977,
+      "step": 21669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613747114644781e-05,
+      "loss": 0.5867,
+      "step": 21670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613682995639908e-05,
+      "loss": 0.5419,
+      "step": 21671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613618876635035e-05,
+      "loss": 0.4071,
+      "step": 21672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6135547576301615e-05,
+      "loss": 0.4826,
+      "step": 21673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613490638625289e-05,
+      "loss": 0.5626,
+      "step": 21674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613426519620416e-05,
+      "loss": 0.4385,
+      "step": 21675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6133624006155424e-05,
+      "loss": 0.5209,
+      "step": 21676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6132982816106695e-05,
+      "loss": 0.5045,
+      "step": 21677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613234162605797e-05,
+      "loss": 0.4856,
+      "step": 21678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613170043600924e-05,
+      "loss": 0.5065,
+      "step": 21679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6131059245960504e-05,
+      "loss": 0.4894,
+      "step": 21680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6130418055911776e-05,
+      "loss": 0.5528,
+      "step": 21681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612977686586304e-05,
+      "loss": 0.5086,
+      "step": 21682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612913567581431e-05,
+      "loss": 0.38,
+      "step": 21683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6128494485765584e-05,
+      "loss": 0.4533,
+      "step": 21684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612785329571685e-05,
+      "loss": 0.4015,
+      "step": 21685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612721210566812e-05,
+      "loss": 0.4272,
+      "step": 21686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612657091561939e-05,
+      "loss": 0.3811,
+      "step": 21687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6125929725570664e-05,
+      "loss": 0.5643,
+      "step": 21688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612528853552193e-05,
+      "loss": 0.5244,
+      "step": 21689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.61246473454732e-05,
+      "loss": 0.5685,
+      "step": 21690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6124006155424466e-05,
+      "loss": 0.5547,
+      "step": 21691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612336496537574e-05,
+      "loss": 0.533,
+      "step": 21692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612272377532701e-05,
+      "loss": 0.6401,
+      "step": 21693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6122082585278274e-05,
+      "loss": 0.4959,
+      "step": 21694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612144139522955e-05,
+      "loss": 0.4176,
+      "step": 21695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612080020518082e-05,
+      "loss": 0.4171,
+      "step": 21696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612015901513209e-05,
+      "loss": 0.6938,
+      "step": 21697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6119517825083354e-05,
+      "loss": 0.4926,
+      "step": 21698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6118876635034626e-05,
+      "loss": 0.5268,
+      "step": 21699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611823544498589e-05,
+      "loss": 0.5176,
+      "step": 21700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611759425493716e-05,
+      "loss": 0.5797,
+      "step": 21701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6116953064888434e-05,
+      "loss": 0.4625,
+      "step": 21702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6116311874839706e-05,
+      "loss": 0.5335,
+      "step": 21703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611567068479098e-05,
+      "loss": 0.4801,
+      "step": 21704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611502949474224e-05,
+      "loss": 0.4772,
+      "step": 21705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6114388304693514e-05,
+      "loss": 0.4491,
+      "step": 21706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611374711464478e-05,
+      "loss": 0.464,
+      "step": 21707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611310592459605e-05,
+      "loss": 0.5962,
+      "step": 21708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6112464734547316e-05,
+      "loss": 0.4462,
+      "step": 21709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611182354449859e-05,
+      "loss": 0.5786,
+      "step": 21710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6111182354449866e-05,
+      "loss": 0.4255,
+      "step": 21711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611054116440113e-05,
+      "loss": 0.5036,
+      "step": 21712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.61098999743524e-05,
+      "loss": 0.6152,
+      "step": 21713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.610925878430367e-05,
+      "loss": 0.4049,
+      "step": 21714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.610861759425494e-05,
+      "loss": 0.4987,
+      "step": 21715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6107976404206204e-05,
+      "loss": 0.5359,
+      "step": 21716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6107335214157476e-05,
+      "loss": 0.6399,
+      "step": 21717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.610669402410875e-05,
+      "loss": 0.4826,
+      "step": 21718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610605283406002e-05,
+      "loss": 0.5607,
+      "step": 21719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610541164401129e-05,
+      "loss": 0.4902,
+      "step": 21720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6104770453962556e-05,
+      "loss": 0.5687,
+      "step": 21721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610412926391383e-05,
+      "loss": 0.3524,
+      "step": 21722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610348807386509e-05,
+      "loss": 0.4642,
+      "step": 21723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6102846883816365e-05,
+      "loss": 0.5211,
+      "step": 21724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610220569376763e-05,
+      "loss": 0.4102,
+      "step": 21725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.61015645037189e-05,
+      "loss": 0.5514,
+      "step": 21726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610092331367018e-05,
+      "loss": 0.4703,
+      "step": 21727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6100282123621445e-05,
+      "loss": 0.5325,
+      "step": 21728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6099640933572716e-05,
+      "loss": 0.4673,
+      "step": 21729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609899974352398e-05,
+      "loss": 0.5674,
+      "step": 21730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609835855347525e-05,
+      "loss": 0.4695,
+      "step": 21731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609771736342652e-05,
+      "loss": 0.4655,
+      "step": 21732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609707617337779e-05,
+      "loss": 0.5549,
+      "step": 21733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609643498332906e-05,
+      "loss": 0.4674,
+      "step": 21734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609579379328033e-05,
+      "loss": 0.4112,
+      "step": 21735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6095152603231605e-05,
+      "loss": 0.5669,
+      "step": 21736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609451141318287e-05,
+      "loss": 0.4512,
+      "step": 21737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609387022313414e-05,
+      "loss": 0.523,
+      "step": 21738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6093229033085406e-05,
+      "loss": 0.6165,
+      "step": 21739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609258784303668e-05,
+      "loss": 0.4849,
+      "step": 21740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609194665298794e-05,
+      "loss": 0.4723,
+      "step": 21741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6091305462939215e-05,
+      "loss": 0.47,
+      "step": 21742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6090664272890486e-05,
+      "loss": 0.4683,
+      "step": 21743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609002308284176e-05,
+      "loss": 0.4351,
+      "step": 21744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608938189279303e-05,
+      "loss": 0.5439,
+      "step": 21745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6088740702744295e-05,
+      "loss": 0.6232,
+      "step": 21746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6088099512695567e-05,
+      "loss": 0.5284,
+      "step": 21747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608745832264683e-05,
+      "loss": 0.4917,
+      "step": 21748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60868171325981e-05,
+      "loss": 0.4844,
+      "step": 21749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6086175942549375e-05,
+      "loss": 0.6006,
+      "step": 21750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608553475250065e-05,
+      "loss": 0.4468,
+      "step": 21751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608489356245191e-05,
+      "loss": 0.3941,
+      "step": 21752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608425237240318e-05,
+      "loss": 0.3896,
+      "step": 21753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6083611182354455e-05,
+      "loss": 0.425,
+      "step": 21754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608296999230572e-05,
+      "loss": 0.4503,
+      "step": 21755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608232880225699e-05,
+      "loss": 0.5581,
+      "step": 21756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608168761220826e-05,
+      "loss": 0.4755,
+      "step": 21757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608104642215953e-05,
+      "loss": 0.581,
+      "step": 21758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60804052321108e-05,
+      "loss": 0.4868,
+      "step": 21759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607976404206207e-05,
+      "loss": 0.5247,
+      "step": 21760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607912285201334e-05,
+      "loss": 0.4435,
+      "step": 21761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607848166196461e-05,
+      "loss": 0.4749,
+      "step": 21762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607784047191588e-05,
+      "loss": 0.4768,
+      "step": 21763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6077199281867145e-05,
+      "loss": 0.4267,
+      "step": 21764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607655809181842e-05,
+      "loss": 0.5214,
+      "step": 21765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607591690176969e-05,
+      "loss": 0.3295,
+      "step": 21766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607527571172096e-05,
+      "loss": 0.7009,
+      "step": 21767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6074634521672225e-05,
+      "loss": 0.4745,
+      "step": 21768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60739933316235e-05,
+      "loss": 0.4064,
+      "step": 21769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607335214157476e-05,
+      "loss": 0.4708,
+      "step": 21770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6072710951526034e-05,
+      "loss": 0.6675,
+      "step": 21771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6072069761477305e-05,
+      "loss": 0.5559,
+      "step": 21772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607142857142857e-05,
+      "loss": 0.6697,
+      "step": 21773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607078738137984e-05,
+      "loss": 0.4855,
+      "step": 21774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6070146191331114e-05,
+      "loss": 0.4584,
+      "step": 21775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6069505001282385e-05,
+      "loss": 0.501,
+      "step": 21776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606886381123365e-05,
+      "loss": 0.5028,
+      "step": 21777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606822262118492e-05,
+      "loss": 0.4217,
+      "step": 21778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606758143113619e-05,
+      "loss": 0.5324,
+      "step": 21779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606694024108746e-05,
+      "loss": 0.4421,
+      "step": 21780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606629905103873e-05,
+      "loss": 0.4927,
+      "step": 21781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6065657860989995e-05,
+      "loss": 0.4691,
+      "step": 21782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6065016670941274e-05,
+      "loss": 0.5694,
+      "step": 21783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606437548089254e-05,
+      "loss": 0.5737,
+      "step": 21784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606373429084381e-05,
+      "loss": 0.5552,
+      "step": 21785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6063093100795075e-05,
+      "loss": 0.5154,
+      "step": 21786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606245191074635e-05,
+      "loss": 0.4671,
+      "step": 21787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606181072069761e-05,
+      "loss": 0.4334,
+      "step": 21788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6061169530648884e-05,
+      "loss": 0.5747,
+      "step": 21789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6060528340600156e-05,
+      "loss": 0.5255,
+      "step": 21790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605988715055143e-05,
+      "loss": 0.4887,
+      "step": 21791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60592459605027e-05,
+      "loss": 0.5137,
+      "step": 21792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6058604770453964e-05,
+      "loss": 0.453,
+      "step": 21793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6057963580405236e-05,
+      "loss": 0.4517,
+      "step": 21794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60573223903565e-05,
+      "loss": 0.5837,
+      "step": 21795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605668120030777e-05,
+      "loss": 0.5739,
+      "step": 21796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605604001025904e-05,
+      "loss": 0.4353,
+      "step": 21797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605539882021031e-05,
+      "loss": 0.4037,
+      "step": 21798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605475763016159e-05,
+      "loss": 0.5122,
+      "step": 21799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605411644011285e-05,
+      "loss": 0.5409,
+      "step": 21800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6053475250064124e-05,
+      "loss": 0.6248,
+      "step": 21801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605283406001539e-05,
+      "loss": 0.5005,
+      "step": 21802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605219286996666e-05,
+      "loss": 0.4678,
+      "step": 21803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6051551679917926e-05,
+      "loss": 0.4334,
+      "step": 21804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60509104898692e-05,
+      "loss": 0.5033,
+      "step": 21805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605026929982047e-05,
+      "loss": 0.4976,
+      "step": 21806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604962810977174e-05,
+      "loss": 0.4514,
+      "step": 21807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604898691972301e-05,
+      "loss": 0.4987,
+      "step": 21808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604834572967428e-05,
+      "loss": 0.4724,
+      "step": 21809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604770453962555e-05,
+      "loss": 0.4745,
+      "step": 21810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6047063349576814e-05,
+      "loss": 0.439,
+      "step": 21811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6046422159528086e-05,
+      "loss": 0.4514,
+      "step": 21812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604578096947935e-05,
+      "loss": 0.4963,
+      "step": 21813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604513977943062e-05,
+      "loss": 0.5076,
+      "step": 21814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6044498589381894e-05,
+      "loss": 0.7132,
+      "step": 21815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6043857399333166e-05,
+      "loss": 0.4418,
+      "step": 21816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604321620928444e-05,
+      "loss": 0.4756,
+      "step": 21817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60425750192357e-05,
+      "loss": 0.3741,
+      "step": 21818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6041933829186974e-05,
+      "loss": 0.4807,
+      "step": 21819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604129263913824e-05,
+      "loss": 0.6205,
+      "step": 21820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604065144908951e-05,
+      "loss": 0.3524,
+      "step": 21821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604001025904078e-05,
+      "loss": 0.5979,
+      "step": 21822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6039369068992054e-05,
+      "loss": 0.6092,
+      "step": 21823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603872787894332e-05,
+      "loss": 0.55,
+      "step": 21824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603808668889459e-05,
+      "loss": 0.5845,
+      "step": 21825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603744549884586e-05,
+      "loss": 0.6112,
+      "step": 21826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603680430879713e-05,
+      "loss": 0.5738,
+      "step": 21827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60361631187484e-05,
+      "loss": 0.5193,
+      "step": 21828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6035521928699664e-05,
+      "loss": 0.4559,
+      "step": 21829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6034880738650936e-05,
+      "loss": 0.5963,
+      "step": 21830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603423954860221e-05,
+      "loss": 0.5788,
+      "step": 21831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603359835855348e-05,
+      "loss": 0.5974,
+      "step": 21832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6032957168504744e-05,
+      "loss": 0.4508,
+      "step": 21833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6032315978456016e-05,
+      "loss": 0.5691,
+      "step": 21834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603167478840729e-05,
+      "loss": 0.5182,
+      "step": 21835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603103359835855e-05,
+      "loss": 0.5113,
+      "step": 21836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6030392408309825e-05,
+      "loss": 0.5066,
+      "step": 21837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6029751218261096e-05,
+      "loss": 0.5532,
+      "step": 21838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602911002821237e-05,
+      "loss": 0.5749,
+      "step": 21839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602846883816363e-05,
+      "loss": 0.4856,
+      "step": 21840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6027827648114905e-05,
+      "loss": 0.6437,
+      "step": 21841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602718645806617e-05,
+      "loss": 0.5107,
+      "step": 21842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602654526801744e-05,
+      "loss": 0.5159,
+      "step": 21843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602590407796871e-05,
+      "loss": 0.461,
+      "step": 21844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602526288791998e-05,
+      "loss": 0.45,
+      "step": 21845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602462169787125e-05,
+      "loss": 0.4572,
+      "step": 21846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602398050782252e-05,
+      "loss": 0.5289,
+      "step": 21847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602333931777379e-05,
+      "loss": 0.4804,
+      "step": 21848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602269812772506e-05,
+      "loss": 0.5122,
+      "step": 21849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602205693767633e-05,
+      "loss": 0.4251,
+      "step": 21850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6021415747627595e-05,
+      "loss": 0.5088,
+      "step": 21851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6020774557578866e-05,
+      "loss": 0.5442,
+      "step": 21852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602013336753014e-05,
+      "loss": 0.4734,
+      "step": 21853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601949217748141e-05,
+      "loss": 0.5376,
+      "step": 21854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601885098743268e-05,
+      "loss": 0.4578,
+      "step": 21855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6018209797383947e-05,
+      "loss": 0.5035,
+      "step": 21856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601756860733522e-05,
+      "loss": 0.6708,
+      "step": 21857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601692741728648e-05,
+      "loss": 0.4253,
+      "step": 21858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6016286227237755e-05,
+      "loss": 0.3781,
+      "step": 21859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601564503718902e-05,
+      "loss": 0.3275,
+      "step": 21860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601500384714029e-05,
+      "loss": 0.4619,
+      "step": 21861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601436265709156e-05,
+      "loss": 0.3188,
+      "step": 21862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6013721467042835e-05,
+      "loss": 0.485,
+      "step": 21863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601308027699411e-05,
+      "loss": 0.4356,
+      "step": 21864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601243908694537e-05,
+      "loss": 0.44,
+      "step": 21865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601179789689664e-05,
+      "loss": 0.5409,
+      "step": 21866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601115670684791e-05,
+      "loss": 0.4568,
+      "step": 21867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601051551679918e-05,
+      "loss": 0.5212,
+      "step": 21868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6009874326750445e-05,
+      "loss": 0.5109,
+      "step": 21869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600923313670172e-05,
+      "loss": 0.5984,
+      "step": 21870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6008591946652995e-05,
+      "loss": 0.4104,
+      "step": 21871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600795075660426e-05,
+      "loss": 0.4843,
+      "step": 21872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600730956655553e-05,
+      "loss": 0.4641,
+      "step": 21873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60066683765068e-05,
+      "loss": 0.5455,
+      "step": 21874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600602718645807e-05,
+      "loss": 0.4621,
+      "step": 21875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6005385996409333e-05,
+      "loss": 0.44,
+      "step": 21876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6004744806360605e-05,
+      "loss": 0.3812,
+      "step": 21877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600410361631188e-05,
+      "loss": 0.6563,
+      "step": 21878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600346242626315e-05,
+      "loss": 0.487,
+      "step": 21879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600282123621442e-05,
+      "loss": 0.4355,
+      "step": 21880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6002180046165685e-05,
+      "loss": 0.4428,
+      "step": 21881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600153885611696e-05,
+      "loss": 0.4626,
+      "step": 21882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600089766606822e-05,
+      "loss": 0.5317,
+      "step": 21883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6000256476019494e-05,
+      "loss": 0.4112,
+      "step": 21884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599961528597076e-05,
+      "loss": 0.4292,
+      "step": 21885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599897409592203e-05,
+      "loss": 0.47,
+      "step": 21886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.59983329058733e-05,
+      "loss": 0.4553,
+      "step": 21887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5997691715824574e-05,
+      "loss": 0.4912,
+      "step": 21888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5997050525775845e-05,
+      "loss": 0.4665,
+      "step": 21889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599640933572711e-05,
+      "loss": 0.3954,
+      "step": 21890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599576814567838e-05,
+      "loss": 0.4068,
+      "step": 21891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599512695562965e-05,
+      "loss": 0.537,
+      "step": 21892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599448576558092e-05,
+      "loss": 0.5836,
+      "step": 21893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599384457553219e-05,
+      "loss": 0.5865,
+      "step": 21894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599320338548346e-05,
+      "loss": 0.6558,
+      "step": 21895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599256219543473e-05,
+      "loss": 0.4939,
+      "step": 21896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5991921005386e-05,
+      "loss": 0.5093,
+      "step": 21897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599127981533727e-05,
+      "loss": 0.4709,
+      "step": 21898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5990638625288535e-05,
+      "loss": 0.4309,
+      "step": 21899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598999743523981e-05,
+      "loss": 0.5815,
+      "step": 21900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598935624519107e-05,
+      "loss": 0.5006,
+      "step": 21901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5988715055142344e-05,
+      "loss": 0.5402,
+      "step": 21902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5988073865093616e-05,
+      "loss": 0.5602,
+      "step": 21903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598743267504489e-05,
+      "loss": 0.5826,
+      "step": 21904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598679148499615e-05,
+      "loss": 0.5101,
+      "step": 21905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5986150294947424e-05,
+      "loss": 0.531,
+      "step": 21906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5985509104898696e-05,
+      "loss": 0.6236,
+      "step": 21907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598486791484996e-05,
+      "loss": 0.6068,
+      "step": 21908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598422672480123e-05,
+      "loss": 0.533,
+      "step": 21909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5983585534752504e-05,
+      "loss": 0.5386,
+      "step": 21910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5982944344703776e-05,
+      "loss": 0.5528,
+      "step": 21911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598230315465504e-05,
+      "loss": 0.538,
+      "step": 21912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598166196460631e-05,
+      "loss": 0.4862,
+      "step": 21913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5981020774557584e-05,
+      "loss": 0.5257,
+      "step": 21914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.598037958450885e-05,
+      "loss": 0.4266,
+      "step": 21915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597973839446012e-05,
+      "loss": 0.5262,
+      "step": 21916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5979097204411386e-05,
+      "loss": 0.5884,
+      "step": 21917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597845601436266e-05,
+      "loss": 0.4423,
+      "step": 21918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597781482431393e-05,
+      "loss": 0.4629,
+      "step": 21919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.59771736342652e-05,
+      "loss": 0.5004,
+      "step": 21920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5976532444216466e-05,
+      "loss": 0.4522,
+      "step": 21921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597589125416774e-05,
+      "loss": 0.5246,
+      "step": 21922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597525006411901e-05,
+      "loss": 0.5519,
+      "step": 21923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5974608874070274e-05,
+      "loss": 0.4454,
+      "step": 21924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5973967684021546e-05,
+      "loss": 0.5279,
+      "step": 21925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597332649397282e-05,
+      "loss": 0.6292,
+      "step": 21926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597268530392409e-05,
+      "loss": 0.5338,
+      "step": 21927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5972044113875354e-05,
+      "loss": 0.4216,
+      "step": 21928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5971402923826626e-05,
+      "loss": 0.4643,
+      "step": 21929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597076173377789e-05,
+      "loss": 0.4425,
+      "step": 21930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.597012054372916e-05,
+      "loss": 0.6044,
+      "step": 21931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5969479353680434e-05,
+      "loss": 0.4716,
+      "step": 21932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.59688381636317e-05,
+      "loss": 0.6377,
+      "step": 21933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596819697358297e-05,
+      "loss": 0.4844,
+      "step": 21934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596755578353424e-05,
+      "loss": 0.5832,
+      "step": 21935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5966914593485514e-05,
+      "loss": 0.5755,
+      "step": 21936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596627340343678e-05,
+      "loss": 0.4638,
+      "step": 21937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596563221338805e-05,
+      "loss": 0.4807,
+      "step": 21938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5964991023339316e-05,
+      "loss": 0.4192,
+      "step": 21939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596434983329059e-05,
+      "loss": 0.4651,
+      "step": 21940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596370864324186e-05,
+      "loss": 0.5549,
+      "step": 21941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5963067453193124e-05,
+      "loss": 0.5468,
+      "step": 21942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.59624262631444e-05,
+      "loss": 0.5592,
+      "step": 21943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596178507309567e-05,
+      "loss": 0.4791,
+      "step": 21944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596114388304694e-05,
+      "loss": 0.5205,
+      "step": 21945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5960502692998204e-05,
+      "loss": 0.4264,
+      "step": 21946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5959861502949476e-05,
+      "loss": 0.5283,
+      "step": 21947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595922031290074e-05,
+      "loss": 0.4323,
+      "step": 21948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595857912285201e-05,
+      "loss": 0.5278,
+      "step": 21949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5957937932803285e-05,
+      "loss": 0.492,
+      "step": 21950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5957296742754556e-05,
+      "loss": 0.484,
+      "step": 21951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595665555270583e-05,
+      "loss": 0.4532,
+      "step": 21952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595601436265709e-05,
+      "loss": 0.4053,
+      "step": 21953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5955373172608365e-05,
+      "loss": 0.5007,
+      "step": 21954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595473198255963e-05,
+      "loss": 0.4271,
+      "step": 21955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.59540907925109e-05,
+      "loss": 0.5578,
+      "step": 21956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5953449602462166e-05,
+      "loss": 0.6154,
+      "step": 21957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595280841241344e-05,
+      "loss": 0.618,
+      "step": 21958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5952167222364716e-05,
+      "loss": 0.4635,
+      "step": 21959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595152603231598e-05,
+      "loss": 0.5624,
+      "step": 21960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595088484226725e-05,
+      "loss": 0.4566,
+      "step": 21961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595024365221852e-05,
+      "loss": 0.5093,
+      "step": 21962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594960246216979e-05,
+      "loss": 0.4981,
+      "step": 21963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5948961272121055e-05,
+      "loss": 0.5591,
+      "step": 21964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5948320082072326e-05,
+      "loss": 0.4932,
+      "step": 21965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.59476788920236e-05,
+      "loss": 0.4837,
+      "step": 21966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594703770197487e-05,
+      "loss": 0.3956,
+      "step": 21967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594639651192614e-05,
+      "loss": 0.4612,
+      "step": 21968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5945755321877407e-05,
+      "loss": 0.4889,
+      "step": 21969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594511413182868e-05,
+      "loss": 0.4579,
+      "step": 21970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594447294177994e-05,
+      "loss": 0.5552,
+      "step": 21971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5943831751731215e-05,
+      "loss": 0.4813,
+      "step": 21972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594319056168248e-05,
+      "loss": 0.5594,
+      "step": 21973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594254937163375e-05,
+      "loss": 0.5393,
+      "step": 21974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594190818158502e-05,
+      "loss": 0.531,
+      "step": 21975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5941266991536295e-05,
+      "loss": 0.6136,
+      "step": 21976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594062580148757e-05,
+      "loss": 0.424,
+      "step": 21977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.593998461143883e-05,
+      "loss": 0.4618,
+      "step": 21978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59393434213901e-05,
+      "loss": 0.5338,
+      "step": 21979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593870223134137e-05,
+      "loss": 0.542,
+      "step": 21980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593806104129264e-05,
+      "loss": 0.4889,
+      "step": 21981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593741985124391e-05,
+      "loss": 0.444,
+      "step": 21982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5936778661195183e-05,
+      "loss": 0.5267,
+      "step": 21983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593613747114645e-05,
+      "loss": 0.4571,
+      "step": 21984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593549628109772e-05,
+      "loss": 0.5361,
+      "step": 21985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593485509104899e-05,
+      "loss": 0.4618,
+      "step": 21986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593421390100026e-05,
+      "loss": 0.4507,
+      "step": 21987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593357271095153e-05,
+      "loss": 0.4599,
+      "step": 21988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5932931520902793e-05,
+      "loss": 0.5798,
+      "step": 21989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5932290330854065e-05,
+      "loss": 0.5008,
+      "step": 21990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593164914080534e-05,
+      "loss": 0.4204,
+      "step": 21991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593100795075661e-05,
+      "loss": 0.4866,
+      "step": 21992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5930366760707874e-05,
+      "loss": 0.5339,
+      "step": 21993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5929725570659145e-05,
+      "loss": 0.4635,
+      "step": 21994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592908438061042e-05,
+      "loss": 0.5634,
+      "step": 21995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592844319056168e-05,
+      "loss": 0.4608,
+      "step": 21996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5927802000512954e-05,
+      "loss": 0.4431,
+      "step": 21997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5927160810464225e-05,
+      "loss": 0.5457,
+      "step": 21998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59265196204155e-05,
+      "loss": 0.4626,
+      "step": 21999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592587843036676e-05,
+      "loss": 0.6074,
+      "step": 22000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5925237240318034e-05,
+      "loss": 0.4393,
+      "step": 22001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59245960502693e-05,
+      "loss": 0.4819,
+      "step": 22002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592395486022057e-05,
+      "loss": 0.4222,
+      "step": 22003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592331367017184e-05,
+      "loss": 0.4017,
+      "step": 22004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592267248012311e-05,
+      "loss": 0.4493,
+      "step": 22005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592203129007438e-05,
+      "loss": 0.5453,
+      "step": 22006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592139010002565e-05,
+      "loss": 0.5765,
+      "step": 22007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592074890997692e-05,
+      "loss": 0.5165,
+      "step": 22008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592010771992819e-05,
+      "loss": 0.4813,
+      "step": 22009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591946652987946e-05,
+      "loss": 0.5679,
+      "step": 22010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5918825339830724e-05,
+      "loss": 0.4941,
+      "step": 22011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5918184149781995e-05,
+      "loss": 0.6454,
+      "step": 22012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591754295973327e-05,
+      "loss": 0.4918,
+      "step": 22013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591690176968454e-05,
+      "loss": 0.4418,
+      "step": 22014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591626057963581e-05,
+      "loss": 0.4293,
+      "step": 22015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5915619389587076e-05,
+      "loss": 0.473,
+      "step": 22016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591497819953835e-05,
+      "loss": 0.3982,
+      "step": 22017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591433700948961e-05,
+      "loss": 0.5299,
+      "step": 22018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5913695819440884e-05,
+      "loss": 0.4631,
+      "step": 22019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591305462939215e-05,
+      "loss": 0.5171,
+      "step": 22020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591241343934342e-05,
+      "loss": 0.4732,
+      "step": 22021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591177224929469e-05,
+      "loss": 0.4643,
+      "step": 22022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5911131059245964e-05,
+      "loss": 0.6147,
+      "step": 22023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5910489869197236e-05,
+      "loss": 0.4331,
+      "step": 22024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59098486791485e-05,
+      "loss": 0.5691,
+      "step": 22025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590920748909977e-05,
+      "loss": 0.4766,
+      "step": 22026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590856629905104e-05,
+      "loss": 0.5558,
+      "step": 22027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590792510900231e-05,
+      "loss": 0.4436,
+      "step": 22028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5907283918953574e-05,
+      "loss": 0.5091,
+      "step": 22029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5906642728904846e-05,
+      "loss": 0.4407,
+      "step": 22030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5906001538856124e-05,
+      "loss": 0.4828,
+      "step": 22031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590536034880739e-05,
+      "loss": 0.589,
+      "step": 22032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590471915875866e-05,
+      "loss": 0.481,
+      "step": 22033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5904077968709926e-05,
+      "loss": 0.4884,
+      "step": 22034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59034367786612e-05,
+      "loss": 0.5641,
+      "step": 22035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590279558861246e-05,
+      "loss": 0.3718,
+      "step": 22036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5902154398563734e-05,
+      "loss": 0.4534,
+      "step": 22037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5901513208515006e-05,
+      "loss": 0.4096,
+      "step": 22038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590087201846628e-05,
+      "loss": 0.4034,
+      "step": 22039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590023082841755e-05,
+      "loss": 0.3912,
+      "step": 22040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5899589638368814e-05,
+      "loss": 0.4794,
+      "step": 22041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5898948448320086e-05,
+      "loss": 0.4855,
+      "step": 22042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589830725827135e-05,
+      "loss": 0.5439,
+      "step": 22043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589766606822262e-05,
+      "loss": 0.4482,
+      "step": 22044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589702487817389e-05,
+      "loss": 0.4696,
+      "step": 22045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589638368812516e-05,
+      "loss": 0.463,
+      "step": 22046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589574249807643e-05,
+      "loss": 0.5784,
+      "step": 22047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58951013080277e-05,
+      "loss": 0.4077,
+      "step": 22048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5894460117978974e-05,
+      "loss": 0.482,
+      "step": 22049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589381892793024e-05,
+      "loss": 0.4053,
+      "step": 22050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589317773788151e-05,
+      "loss": 0.6512,
+      "step": 22051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5892536547832776e-05,
+      "loss": 0.5518,
+      "step": 22052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589189535778405e-05,
+      "loss": 0.4566,
+      "step": 22053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589125416773532e-05,
+      "loss": 0.4719,
+      "step": 22054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589061297768659e-05,
+      "loss": 0.512,
+      "step": 22055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5889971787637856e-05,
+      "loss": 0.5341,
+      "step": 22056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588933059758913e-05,
+      "loss": 0.5054,
+      "step": 22057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58886894075404e-05,
+      "loss": 0.5422,
+      "step": 22058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5888048217491665e-05,
+      "loss": 0.4657,
+      "step": 22059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5887407027442936e-05,
+      "loss": 0.4109,
+      "step": 22060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58867658373942e-05,
+      "loss": 0.5155,
+      "step": 22061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588612464734547e-05,
+      "loss": 0.4426,
+      "step": 22062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5885483457296745e-05,
+      "loss": 0.4982,
+      "step": 22063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5884842267248016e-05,
+      "loss": 0.5423,
+      "step": 22064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588420107719928e-05,
+      "loss": 0.5851,
+      "step": 22065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588355988715055e-05,
+      "loss": 0.495,
+      "step": 22066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5882918697101825e-05,
+      "loss": 0.462,
+      "step": 22067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588227750705309e-05,
+      "loss": 0.5283,
+      "step": 22068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588163631700436e-05,
+      "loss": 0.5034,
+      "step": 22069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588099512695563e-05,
+      "loss": 0.4886,
+      "step": 22070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5880353936906905e-05,
+      "loss": 0.4565,
+      "step": 22071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587971274685817e-05,
+      "loss": 0.5412,
+      "step": 22072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587907155680944e-05,
+      "loss": 0.4744,
+      "step": 22073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5878430366760706e-05,
+      "loss": 0.5535,
+      "step": 22074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587778917671198e-05,
+      "loss": 0.4227,
+      "step": 22075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587714798666325e-05,
+      "loss": 0.5821,
+      "step": 22076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5876506796614515e-05,
+      "loss": 0.4939,
+      "step": 22077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5875865606565786e-05,
+      "loss": 0.5301,
+      "step": 22078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587522441651706e-05,
+      "loss": 0.5847,
+      "step": 22079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587458322646833e-05,
+      "loss": 0.4845,
+      "step": 22080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5873942036419595e-05,
+      "loss": 0.5688,
+      "step": 22081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5873300846370867e-05,
+      "loss": 0.4523,
+      "step": 22082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587265965632213e-05,
+      "loss": 0.4976,
+      "step": 22083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58720184662734e-05,
+      "loss": 0.5255,
+      "step": 22084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5871377276224675e-05,
+      "loss": 0.5202,
+      "step": 22085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587073608617595e-05,
+      "loss": 0.5,
+      "step": 22086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587009489612722e-05,
+      "loss": 0.4952,
+      "step": 22087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586945370607848e-05,
+      "loss": 0.5348,
+      "step": 22088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5868812516029755e-05,
+      "loss": 0.4022,
+      "step": 22089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586817132598102e-05,
+      "loss": 0.5221,
+      "step": 22090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586753013593229e-05,
+      "loss": 0.4224,
+      "step": 22091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586688894588356e-05,
+      "loss": 0.4711,
+      "step": 22092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586624775583483e-05,
+      "loss": 0.4157,
+      "step": 22093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58656065657861e-05,
+      "loss": 0.4437,
+      "step": 22094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586496537573737e-05,
+      "loss": 0.4325,
+      "step": 22095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5864324185688643e-05,
+      "loss": 0.6081,
+      "step": 22096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586368299563991e-05,
+      "loss": 0.5659,
+      "step": 22097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586304180559118e-05,
+      "loss": 0.4105,
+      "step": 22098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5862400615542445e-05,
+      "loss": 0.5258,
+      "step": 22099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586175942549372e-05,
+      "loss": 0.5205,
+      "step": 22100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586111823544499e-05,
+      "loss": 0.5069,
+      "step": 22101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586047704539626e-05,
+      "loss": 0.5783,
+      "step": 22102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585983585534753e-05,
+      "loss": 0.6467,
+      "step": 22103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58591946652988e-05,
+      "loss": 0.4976,
+      "step": 22104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585855347525007e-05,
+      "loss": 0.4976,
+      "step": 22105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5857912285201334e-05,
+      "loss": 0.4483,
+      "step": 22106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5857271095152605e-05,
+      "loss": 0.5737,
+      "step": 22107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585662990510387e-05,
+      "loss": 0.3979,
+      "step": 22108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585598871505514e-05,
+      "loss": 0.5161,
+      "step": 22109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5855347525006414e-05,
+      "loss": 0.5005,
+      "step": 22110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5854706334957685e-05,
+      "loss": 0.4214,
+      "step": 22111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585406514490896e-05,
+      "loss": 0.5318,
+      "step": 22112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585342395486022e-05,
+      "loss": 0.4805,
+      "step": 22113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5852782764811494e-05,
+      "loss": 0.4753,
+      "step": 22114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585214157476276e-05,
+      "loss": 0.5043,
+      "step": 22115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585150038471403e-05,
+      "loss": 0.5459,
+      "step": 22116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5850859194665295e-05,
+      "loss": 0.4423,
+      "step": 22117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585021800461657e-05,
+      "loss": 0.5893,
+      "step": 22118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5849576814567846e-05,
+      "loss": 0.459,
+      "step": 22119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584893562451911e-05,
+      "loss": 0.456,
+      "step": 22120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584829443447038e-05,
+      "loss": 0.5251,
+      "step": 22121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584765324442165e-05,
+      "loss": 0.5206,
+      "step": 22122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584701205437292e-05,
+      "loss": 0.5164,
+      "step": 22123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5846370864324184e-05,
+      "loss": 0.4575,
+      "step": 22124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5845729674275456e-05,
+      "loss": 0.4542,
+      "step": 22125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584508848422673e-05,
+      "loss": 0.5744,
+      "step": 22126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5844447294178e-05,
+      "loss": 0.5583,
+      "step": 22127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584380610412927e-05,
+      "loss": 0.4627,
+      "step": 22128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5843164914080536e-05,
+      "loss": 0.447,
+      "step": 22129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584252372403181e-05,
+      "loss": 0.4369,
+      "step": 22130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584188253398307e-05,
+      "loss": 0.6338,
+      "step": 22131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5841241343934344e-05,
+      "loss": 0.5413,
+      "step": 22132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584060015388561e-05,
+      "loss": 0.5127,
+      "step": 22133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583995896383688e-05,
+      "loss": 0.4515,
+      "step": 22134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583931777378815e-05,
+      "loss": 0.438,
+      "step": 22135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5838676583739424e-05,
+      "loss": 0.4882,
+      "step": 22136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5838035393690696e-05,
+      "loss": 0.4801,
+      "step": 22137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583739420364196e-05,
+      "loss": 0.4945,
+      "step": 22138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583675301359323e-05,
+      "loss": 0.5761,
+      "step": 22139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58361118235445e-05,
+      "loss": 0.515,
+      "step": 22140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583547063349577e-05,
+      "loss": 0.5082,
+      "step": 22141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583482944344704e-05,
+      "loss": 0.4191,
+      "step": 22142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583418825339831e-05,
+      "loss": 0.4898,
+      "step": 22143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583354706334958e-05,
+      "loss": 0.5491,
+      "step": 22144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583290587330085e-05,
+      "loss": 0.5116,
+      "step": 22145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583226468325212e-05,
+      "loss": 0.4239,
+      "step": 22146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5831623493203386e-05,
+      "loss": 0.3829,
+      "step": 22147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583098230315466e-05,
+      "loss": 0.4527,
+      "step": 22148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583034111310592e-05,
+      "loss": 0.4127,
+      "step": 22149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5829699923057194e-05,
+      "loss": 0.4505,
+      "step": 22150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5829058733008466e-05,
+      "loss": 0.5654,
+      "step": 22151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582841754295974e-05,
+      "loss": 0.4836,
+      "step": 22152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5827776352911e-05,
+      "loss": 0.4409,
+      "step": 22153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5827135162862274e-05,
+      "loss": 0.5401,
+      "step": 22154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5826493972813546e-05,
+      "loss": 0.4943,
+      "step": 22155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582585278276481e-05,
+      "loss": 0.5623,
+      "step": 22156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582521159271608e-05,
+      "loss": 0.4752,
+      "step": 22157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5824570402667354e-05,
+      "loss": 0.5076,
+      "step": 22158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5823929212618626e-05,
+      "loss": 0.6048,
+      "step": 22159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582328802256989e-05,
+      "loss": 0.4036,
+      "step": 22160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582264683252116e-05,
+      "loss": 0.4562,
+      "step": 22161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582200564247243e-05,
+      "loss": 0.4888,
+      "step": 22162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58213644524237e-05,
+      "loss": 0.4488,
+      "step": 22163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582072326237497e-05,
+      "loss": 0.5588,
+      "step": 22164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5820082072326236e-05,
+      "loss": 0.6175,
+      "step": 22165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581944088227751e-05,
+      "loss": 0.4789,
+      "step": 22166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581879969222878e-05,
+      "loss": 0.534,
+      "step": 22167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581815850218005e-05,
+      "loss": 0.4316,
+      "step": 22168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5817517312131316e-05,
+      "loss": 0.4916,
+      "step": 22169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581687612208259e-05,
+      "loss": 0.5873,
+      "step": 22170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581623493203385e-05,
+      "loss": 0.4549,
+      "step": 22171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5815593741985125e-05,
+      "loss": 0.4306,
+      "step": 22172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5814952551936396e-05,
+      "loss": 0.4138,
+      "step": 22173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581431136188767e-05,
+      "loss": 0.3925,
+      "step": 22174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581367017183894e-05,
+      "loss": 0.4589,
+      "step": 22175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5813028981790205e-05,
+      "loss": 0.632,
+      "step": 22176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5812387791741476e-05,
+      "loss": 0.5077,
+      "step": 22177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581174660169274e-05,
+      "loss": 0.4002,
+      "step": 22178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581110541164401e-05,
+      "loss": 0.4367,
+      "step": 22179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581046422159528e-05,
+      "loss": 0.3763,
+      "step": 22180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580982303154655e-05,
+      "loss": 0.5664,
+      "step": 22181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580918184149782e-05,
+      "loss": 0.5368,
+      "step": 22182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580854065144909e-05,
+      "loss": 0.4719,
+      "step": 22183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5807899461400365e-05,
+      "loss": 0.4939,
+      "step": 22184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580725827135163e-05,
+      "loss": 0.395,
+      "step": 22185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58066170813029e-05,
+      "loss": 0.4624,
+      "step": 22186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5805975891254166e-05,
+      "loss": 0.4796,
+      "step": 22187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580533470120544e-05,
+      "loss": 0.511,
+      "step": 22188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58046935111567e-05,
+      "loss": 0.5278,
+      "step": 22189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5804052321107975e-05,
+      "loss": 0.577,
+      "step": 22190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580341113105925e-05,
+      "loss": 0.6693,
+      "step": 22191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580276994101052e-05,
+      "loss": 0.4766,
+      "step": 22192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580212875096179e-05,
+      "loss": 0.6106,
+      "step": 22193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5801487560913055e-05,
+      "loss": 0.4759,
+      "step": 22194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5800846370864327e-05,
+      "loss": 0.58,
+      "step": 22195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580020518081559e-05,
+      "loss": 0.5931,
+      "step": 22196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579956399076686e-05,
+      "loss": 0.5259,
+      "step": 22197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5798922800718135e-05,
+      "loss": 0.4608,
+      "step": 22198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579828161066941e-05,
+      "loss": 0.4645,
+      "step": 22199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579764042062068e-05,
+      "loss": 0.4876,
+      "step": 22200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579699923057194e-05,
+      "loss": 0.5182,
+      "step": 22201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5796358040523215e-05,
+      "loss": 0.4058,
+      "step": 22202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579571685047448e-05,
+      "loss": 0.4733,
+      "step": 22203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579507566042575e-05,
+      "loss": 0.444,
+      "step": 22204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579443447037702e-05,
+      "loss": 0.5109,
+      "step": 22205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579379328032829e-05,
+      "loss": 0.4952,
+      "step": 22206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579315209027956e-05,
+      "loss": 0.4855,
+      "step": 22207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579251090023083e-05,
+      "loss": 0.6015,
+      "step": 22208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5791869710182104e-05,
+      "loss": 0.516,
+      "step": 22209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579122852013337e-05,
+      "loss": 0.4548,
+      "step": 22210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579058733008464e-05,
+      "loss": 0.3875,
+      "step": 22211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5789946140035905e-05,
+      "loss": 0.5358,
+      "step": 22212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578930494998718e-05,
+      "loss": 0.5342,
+      "step": 22213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578866375993845e-05,
+      "loss": 0.5129,
+      "step": 22214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578802256988972e-05,
+      "loss": 0.4777,
+      "step": 22215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5787381379840985e-05,
+      "loss": 0.4659,
+      "step": 22216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578674018979226e-05,
+      "loss": 0.5592,
+      "step": 22217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578609899974353e-05,
+      "loss": 0.5415,
+      "step": 22218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5785457809694794e-05,
+      "loss": 0.4659,
+      "step": 22219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5784816619646065e-05,
+      "loss": 0.4905,
+      "step": 22220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578417542959733e-05,
+      "loss": 0.3488,
+      "step": 22221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.57835342395486e-05,
+      "loss": 0.5311,
+      "step": 22222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5782893049499874e-05,
+      "loss": 0.5775,
+      "step": 22223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5782251859451145e-05,
+      "loss": 0.5527,
+      "step": 22224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578161066940241e-05,
+      "loss": 0.4114,
+      "step": 22225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578096947935368e-05,
+      "loss": 0.497,
+      "step": 22226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5780328289304954e-05,
+      "loss": 0.5021,
+      "step": 22227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.577968709925622e-05,
+      "loss": 0.4332,
+      "step": 22228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.577904590920749e-05,
+      "loss": 0.4202,
+      "step": 22229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.577840471915876e-05,
+      "loss": 0.4936,
+      "step": 22230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5777763529110034e-05,
+      "loss": 0.5055,
+      "step": 22231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.57771223390613e-05,
+      "loss": 0.495,
+      "step": 22232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.577648114901257e-05,
+      "loss": 0.4959,
+      "step": 22233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5775839958963835e-05,
+      "loss": 0.5046,
+      "step": 22234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.577519876891511e-05,
+      "loss": 0.6257,
+      "step": 22235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.577455757886638e-05,
+      "loss": 0.6127,
+      "step": 22236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5773916388817644e-05,
+      "loss": 0.5143,
+      "step": 22237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5773275198768916e-05,
+      "loss": 0.5334,
+      "step": 22238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577263400872019e-05,
+      "loss": 0.5048,
+      "step": 22239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577199281867146e-05,
+      "loss": 0.5806,
+      "step": 22240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5771351628622724e-05,
+      "loss": 0.4998,
+      "step": 22241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5770710438573996e-05,
+      "loss": 0.5321,
+      "step": 22242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577006924852526e-05,
+      "loss": 0.4777,
+      "step": 22243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576942805847653e-05,
+      "loss": 0.4651,
+      "step": 22244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5768786868427804e-05,
+      "loss": 0.5164,
+      "step": 22245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5768145678379076e-05,
+      "loss": 0.5667,
+      "step": 22246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576750448833035e-05,
+      "loss": 0.5725,
+      "step": 22247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576686329828161e-05,
+      "loss": 0.4436,
+      "step": 22248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5766222108232884e-05,
+      "loss": 0.4381,
+      "step": 22249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576558091818415e-05,
+      "loss": 0.5192,
+      "step": 22250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576493972813542e-05,
+      "loss": 0.5171,
+      "step": 22251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5764298538086686e-05,
+      "loss": 0.4726,
+      "step": 22252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576365734803796e-05,
+      "loss": 0.4378,
+      "step": 22253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576301615798923e-05,
+      "loss": 0.4185,
+      "step": 22254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57623749679405e-05,
+      "loss": 0.5847,
+      "step": 22255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576173377789177e-05,
+      "loss": 0.5449,
+      "step": 22256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576109258784304e-05,
+      "loss": 0.4265,
+      "step": 22257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576045139779431e-05,
+      "loss": 0.5557,
+      "step": 22258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5759810207745574e-05,
+      "loss": 0.5611,
+      "step": 22259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5759169017696846e-05,
+      "loss": 0.4825,
+      "step": 22260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575852782764811e-05,
+      "loss": 0.3844,
+      "step": 22261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575788663759939e-05,
+      "loss": 0.5129,
+      "step": 22262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575724544755066e-05,
+      "loss": 0.5553,
+      "step": 22263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5756604257501926e-05,
+      "loss": 0.4965,
+      "step": 22264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57559630674532e-05,
+      "loss": 0.4942,
+      "step": 22265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575532187740446e-05,
+      "loss": 0.4138,
+      "step": 22266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5754680687355734e-05,
+      "loss": 0.4877,
+      "step": 22267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5754039497307e-05,
+      "loss": 0.4311,
+      "step": 22268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575339830725827e-05,
+      "loss": 0.6152,
+      "step": 22269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575275711720954e-05,
+      "loss": 0.4675,
+      "step": 22270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5752115927160814e-05,
+      "loss": 0.4772,
+      "step": 22271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5751474737112086e-05,
+      "loss": 0.5833,
+      "step": 22272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575083354706335e-05,
+      "loss": 0.5734,
+      "step": 22273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575019235701462e-05,
+      "loss": 0.4913,
+      "step": 22274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574955116696589e-05,
+      "loss": 0.4689,
+      "step": 22275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574890997691716e-05,
+      "loss": 0.5355,
+      "step": 22276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5748268786868424e-05,
+      "loss": 0.451,
+      "step": 22277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5747627596819696e-05,
+      "loss": 0.5208,
+      "step": 22278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574698640677097e-05,
+      "loss": 0.3925,
+      "step": 22279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574634521672224e-05,
+      "loss": 0.5106,
+      "step": 22280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574570402667351e-05,
+      "loss": 0.534,
+      "step": 22281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5745062836624776e-05,
+      "loss": 0.4476,
+      "step": 22282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574442164657605e-05,
+      "loss": 0.437,
+      "step": 22283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574378045652731e-05,
+      "loss": 0.5225,
+      "step": 22284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5743139266478585e-05,
+      "loss": 0.5874,
+      "step": 22285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5742498076429856e-05,
+      "loss": 0.4934,
+      "step": 22286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574185688638113e-05,
+      "loss": 0.4375,
+      "step": 22287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574121569633239e-05,
+      "loss": 0.5452,
+      "step": 22288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5740574506283665e-05,
+      "loss": 0.4212,
+      "step": 22289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5739933316234936e-05,
+      "loss": 0.4602,
+      "step": 22290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57392921261862e-05,
+      "loss": 0.6454,
+      "step": 22291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573865093613747e-05,
+      "loss": 0.585,
+      "step": 22292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573800974608874e-05,
+      "loss": 0.4966,
+      "step": 22293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573736855604001e-05,
+      "loss": 0.4938,
+      "step": 22294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573672736599128e-05,
+      "loss": 0.509,
+      "step": 22295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573608617594255e-05,
+      "loss": 0.582,
+      "step": 22296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5735444985893825e-05,
+      "loss": 0.4551,
+      "step": 22297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573480379584509e-05,
+      "loss": 0.5112,
+      "step": 22298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573416260579636e-05,
+      "loss": 0.4823,
+      "step": 22299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5733521415747626e-05,
+      "loss": 0.5488,
+      "step": 22300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57328802256989e-05,
+      "loss": 0.4954,
+      "step": 22301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573223903565017e-05,
+      "loss": 0.4902,
+      "step": 22302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573159784560144e-05,
+      "loss": 0.4986,
+      "step": 22303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5730956655552707e-05,
+      "loss": 0.4707,
+      "step": 22304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573031546550398e-05,
+      "loss": 0.5053,
+      "step": 22305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572967427545525e-05,
+      "loss": 0.525,
+      "step": 22306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5729033085406515e-05,
+      "loss": 0.5485,
+      "step": 22307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572839189535779e-05,
+      "loss": 0.5267,
+      "step": 22308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572775070530905e-05,
+      "loss": 0.4862,
+      "step": 22309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572710951526032e-05,
+      "loss": 0.4401,
+      "step": 22310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5726468325211595e-05,
+      "loss": 0.4588,
+      "step": 22311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572582713516287e-05,
+      "loss": 0.484,
+      "step": 22312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572518594511413e-05,
+      "loss": 0.5029,
+      "step": 22313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57245447550654e-05,
+      "loss": 0.5646,
+      "step": 22314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5723903565016675e-05,
+      "loss": 0.4751,
+      "step": 22315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572326237496794e-05,
+      "loss": 0.5515,
+      "step": 22316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572262118491921e-05,
+      "loss": 0.4187,
+      "step": 22317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5721979994870483e-05,
+      "loss": 0.5494,
+      "step": 22318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5721338804821755e-05,
+      "loss": 0.4325,
+      "step": 22319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572069761477302e-05,
+      "loss": 0.4679,
+      "step": 22320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572005642472429e-05,
+      "loss": 0.5055,
+      "step": 22321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571941523467556e-05,
+      "loss": 0.4369,
+      "step": 22322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571877404462683e-05,
+      "loss": 0.4727,
+      "step": 22323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57181328545781e-05,
+      "loss": 0.5619,
+      "step": 22324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5717491664529365e-05,
+      "loss": 0.5207,
+      "step": 22325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571685047448064e-05,
+      "loss": 0.4896,
+      "step": 22326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571620928443191e-05,
+      "loss": 0.4165,
+      "step": 22327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571556809438318e-05,
+      "loss": 0.5314,
+      "step": 22328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5714926904334445e-05,
+      "loss": 0.4876,
+      "step": 22329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.4557,
+      "step": 22330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571364452423698e-05,
+      "loss": 0.4904,
+      "step": 22331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5713003334188254e-05,
+      "loss": 0.4054,
+      "step": 22332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5712362144139525e-05,
+      "loss": 0.4884,
+      "step": 22333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57117209540908e-05,
+      "loss": 0.3936,
+      "step": 22334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571107976404207e-05,
+      "loss": 0.5072,
+      "step": 22335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5710438573993334e-05,
+      "loss": 0.4628,
+      "step": 22336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5709797383944605e-05,
+      "loss": 0.53,
+      "step": 22337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570915619389587e-05,
+      "loss": 0.382,
+      "step": 22338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570851500384714e-05,
+      "loss": 0.5434,
+      "step": 22339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570787381379841e-05,
+      "loss": 0.5449,
+      "step": 22340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570723262374968e-05,
+      "loss": 0.4305,
+      "step": 22341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570659143370095e-05,
+      "loss": 0.4574,
+      "step": 22342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570595024365222e-05,
+      "loss": 0.5262,
+      "step": 22343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5705309053603494e-05,
+      "loss": 0.5342,
+      "step": 22344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570466786355476e-05,
+      "loss": 0.4658,
+      "step": 22345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570402667350603e-05,
+      "loss": 0.515,
+      "step": 22346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5703385483457295e-05,
+      "loss": 0.5732,
+      "step": 22347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570274429340857e-05,
+      "loss": 0.4997,
+      "step": 22348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570210310335983e-05,
+      "loss": 0.5437,
+      "step": 22349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5701461913311104e-05,
+      "loss": 0.4055,
+      "step": 22350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570082072326238e-05,
+      "loss": 0.5288,
+      "step": 22351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570017953321365e-05,
+      "loss": 0.4501,
+      "step": 22352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569953834316492e-05,
+      "loss": 0.4815,
+      "step": 22353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5698897153116184e-05,
+      "loss": 0.4843,
+      "step": 22354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5698255963067456e-05,
+      "loss": 0.4471,
+      "step": 22355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569761477301872e-05,
+      "loss": 0.4644,
+      "step": 22356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569697358296999e-05,
+      "loss": 0.4065,
+      "step": 22357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5696332392921264e-05,
+      "loss": 0.5376,
+      "step": 22358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5695691202872536e-05,
+      "loss": 0.553,
+      "step": 22359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569505001282381e-05,
+      "loss": 0.6132,
+      "step": 22360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569440882277507e-05,
+      "loss": 0.5977,
+      "step": 22361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5693767632726344e-05,
+      "loss": 0.5571,
+      "step": 22362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569312644267761e-05,
+      "loss": 0.4512,
+      "step": 22363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569248525262888e-05,
+      "loss": 0.3981,
+      "step": 22364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5691844062580146e-05,
+      "loss": 0.4588,
+      "step": 22365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569120287253142e-05,
+      "loss": 0.5096,
+      "step": 22366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569056168248269e-05,
+      "loss": 0.5335,
+      "step": 22367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568992049243396e-05,
+      "loss": 0.4142,
+      "step": 22368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568927930238523e-05,
+      "loss": 0.4592,
+      "step": 22369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56886381123365e-05,
+      "loss": 0.4283,
+      "step": 22370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568799692228777e-05,
+      "loss": 0.4091,
+      "step": 22371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5687355732239034e-05,
+      "loss": 0.4696,
+      "step": 22372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5686714542190306e-05,
+      "loss": 0.4348,
+      "step": 22373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568607335214158e-05,
+      "loss": 0.5147,
+      "step": 22374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568543216209285e-05,
+      "loss": 0.4955,
+      "step": 22375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5684790972044114e-05,
+      "loss": 0.4618,
+      "step": 22376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5684149781995386e-05,
+      "loss": 0.4004,
+      "step": 22377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568350859194666e-05,
+      "loss": 0.6169,
+      "step": 22378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568286740189792e-05,
+      "loss": 0.5243,
+      "step": 22379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5682226211849194e-05,
+      "loss": 0.5987,
+      "step": 22380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568158502180046e-05,
+      "loss": 0.5334,
+      "step": 22381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568094383175173e-05,
+      "loss": 0.5641,
+      "step": 22382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5680302641703e-05,
+      "loss": 0.5347,
+      "step": 22383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5679661451654274e-05,
+      "loss": 0.4866,
+      "step": 22384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567902026160554e-05,
+      "loss": 0.4083,
+      "step": 22385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567837907155681e-05,
+      "loss": 0.5061,
+      "step": 22386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567773788150808e-05,
+      "loss": 0.5544,
+      "step": 22387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567709669145935e-05,
+      "loss": 0.5021,
+      "step": 22388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567645550141062e-05,
+      "loss": 0.5547,
+      "step": 22389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567581431136189e-05,
+      "loss": 0.3869,
+      "step": 22390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567517312131316e-05,
+      "loss": 0.4354,
+      "step": 22391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567453193126443e-05,
+      "loss": 0.5611,
+      "step": 22392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56738907412157e-05,
+      "loss": 0.4485,
+      "step": 22393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5673249551166965e-05,
+      "loss": 0.5403,
+      "step": 22394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5672608361118236e-05,
+      "loss": 0.4981,
+      "step": 22395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567196717106951e-05,
+      "loss": 0.4969,
+      "step": 22396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567132598102077e-05,
+      "loss": 0.5193,
+      "step": 22397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5670684790972045e-05,
+      "loss": 0.4255,
+      "step": 22398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5670043600923316e-05,
+      "loss": 0.4292,
+      "step": 22399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566940241087459e-05,
+      "loss": 0.5747,
+      "step": 22400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566876122082585e-05,
+      "loss": 0.5559,
+      "step": 22401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5668120030777125e-05,
+      "loss": 0.5756,
+      "step": 22402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566747884072839e-05,
+      "loss": 0.4499,
+      "step": 22403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566683765067966e-05,
+      "loss": 0.4843,
+      "step": 22404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566619646063093e-05,
+      "loss": 0.5353,
+      "step": 22405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5665555270582205e-05,
+      "loss": 0.4956,
+      "step": 22406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5664914080533477e-05,
+      "loss": 0.6127,
+      "step": 22407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566427289048474e-05,
+      "loss": 0.4415,
+      "step": 22408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566363170043601e-05,
+      "loss": 0.5437,
+      "step": 22409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566299051038728e-05,
+      "loss": 0.5803,
+      "step": 22410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566234932033855e-05,
+      "loss": 0.4959,
+      "step": 22411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5661708130289815e-05,
+      "loss": 0.5238,
+      "step": 22412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5661066940241086e-05,
+      "loss": 0.5727,
+      "step": 22413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566042575019236e-05,
+      "loss": 0.5943,
+      "step": 22414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565978456014363e-05,
+      "loss": 0.5223,
+      "step": 22415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56591433700949e-05,
+      "loss": 0.573,
+      "step": 22416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5658502180046167e-05,
+      "loss": 0.5688,
+      "step": 22417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565786098999744e-05,
+      "loss": 0.3956,
+      "step": 22418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56572197999487e-05,
+      "loss": 0.5439,
+      "step": 22419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5656578609899975e-05,
+      "loss": 0.5163,
+      "step": 22420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565593741985124e-05,
+      "loss": 0.5248,
+      "step": 22421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565529622980252e-05,
+      "loss": 0.5007,
+      "step": 22422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565465503975379e-05,
+      "loss": 0.5429,
+      "step": 22423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5654013849705055e-05,
+      "loss": 0.419,
+      "step": 22424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565337265965633e-05,
+      "loss": 0.5556,
+      "step": 22425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565273146960759e-05,
+      "loss": 0.5027,
+      "step": 22426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5652090279558863e-05,
+      "loss": 0.4605,
+      "step": 22427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565144908951013e-05,
+      "loss": 0.5569,
+      "step": 22428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56508078994614e-05,
+      "loss": 0.5469,
+      "step": 22429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565016670941267e-05,
+      "loss": 0.4446,
+      "step": 22430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5649525519363944e-05,
+      "loss": 0.454,
+      "step": 22431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5648884329315215e-05,
+      "loss": 0.5011,
+      "step": 22432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564824313926648e-05,
+      "loss": 0.3732,
+      "step": 22433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564760194921775e-05,
+      "loss": 0.4953,
+      "step": 22434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564696075916902e-05,
+      "loss": 0.569,
+      "step": 22435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564631956912029e-05,
+      "loss": 0.4707,
+      "step": 22436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5645678379071553e-05,
+      "loss": 0.4172,
+      "step": 22437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5645037189022825e-05,
+      "loss": 0.4447,
+      "step": 22438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56443959989741e-05,
+      "loss": 0.5358,
+      "step": 22439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564375480892537e-05,
+      "loss": 0.5476,
+      "step": 22440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564311361887664e-05,
+      "loss": 0.4838,
+      "step": 22441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5642472428827905e-05,
+      "loss": 0.5355,
+      "step": 22442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564183123877918e-05,
+      "loss": 0.4773,
+      "step": 22443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564119004873044e-05,
+      "loss": 0.4756,
+      "step": 22444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5640548858681714e-05,
+      "loss": 0.4167,
+      "step": 22445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5639907668632985e-05,
+      "loss": 0.4044,
+      "step": 22446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563926647858426e-05,
+      "loss": 0.5399,
+      "step": 22447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563862528853552e-05,
+      "loss": 0.4558,
+      "step": 22448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5637984098486794e-05,
+      "loss": 0.3816,
+      "step": 22449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5637342908438065e-05,
+      "loss": 0.4762,
+      "step": 22450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563670171838933e-05,
+      "loss": 0.4138,
+      "step": 22451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56360605283406e-05,
+      "loss": 0.4288,
+      "step": 22452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563541933829187e-05,
+      "loss": 0.5711,
+      "step": 22453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563477814824314e-05,
+      "loss": 0.4356,
+      "step": 22454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563413695819441e-05,
+      "loss": 0.5089,
+      "step": 22455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563349576814568e-05,
+      "loss": 0.6022,
+      "step": 22456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563285457809695e-05,
+      "loss": 0.4765,
+      "step": 22457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563221338804822e-05,
+      "loss": 0.52,
+      "step": 22458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563157219799949e-05,
+      "loss": 0.3799,
+      "step": 22459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5630931007950756e-05,
+      "loss": 0.5088,
+      "step": 22460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563028981790203e-05,
+      "loss": 0.4963,
+      "step": 22461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56296486278533e-05,
+      "loss": 0.4821,
+      "step": 22462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562900743780457e-05,
+      "loss": 0.4803,
+      "step": 22463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5628366247755836e-05,
+      "loss": 0.4795,
+      "step": 22464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562772505770711e-05,
+      "loss": 0.3586,
+      "step": 22465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562708386765837e-05,
+      "loss": 0.4551,
+      "step": 22466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5626442677609644e-05,
+      "loss": 0.4789,
+      "step": 22467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5625801487560916e-05,
+      "loss": 0.4981,
+      "step": 22468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562516029751218e-05,
+      "loss": 0.6034,
+      "step": 22469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562451910746345e-05,
+      "loss": 0.4452,
+      "step": 22470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5623877917414724e-05,
+      "loss": 0.3788,
+      "step": 22471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5623236727365996e-05,
+      "loss": 0.4142,
+      "step": 22472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562259553731726e-05,
+      "loss": 0.4484,
+      "step": 22473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562195434726853e-05,
+      "loss": 0.5498,
+      "step": 22474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56213131572198e-05,
+      "loss": 0.561,
+      "step": 22475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562067196717107e-05,
+      "loss": 0.4846,
+      "step": 22476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562003077712234e-05,
+      "loss": 0.4841,
+      "step": 22477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561938958707361e-05,
+      "loss": 0.6672,
+      "step": 22478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5618748397024884e-05,
+      "loss": 0.618,
+      "step": 22479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561810720697615e-05,
+      "loss": 0.5755,
+      "step": 22480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561746601692742e-05,
+      "loss": 0.5264,
+      "step": 22481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5616824826878686e-05,
+      "loss": 0.514,
+      "step": 22482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561618363682996e-05,
+      "loss": 0.4614,
+      "step": 22483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561554244678123e-05,
+      "loss": 0.5166,
+      "step": 22484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5614901256732494e-05,
+      "loss": 0.6423,
+      "step": 22485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5614260066683766e-05,
+      "loss": 0.5874,
+      "step": 22486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561361887663504e-05,
+      "loss": 0.5657,
+      "step": 22487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561297768658631e-05,
+      "loss": 0.5525,
+      "step": 22488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5612336496537574e-05,
+      "loss": 0.5889,
+      "step": 22489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5611695306488846e-05,
+      "loss": 0.398,
+      "step": 22490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561105411644011e-05,
+      "loss": 0.4401,
+      "step": 22491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561041292639138e-05,
+      "loss": 0.5044,
+      "step": 22492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5609771736342654e-05,
+      "loss": 0.456,
+      "step": 22493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5609130546293926e-05,
+      "loss": 0.5886,
+      "step": 22494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56084893562452e-05,
+      "loss": 0.529,
+      "step": 22495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.560784816619646e-05,
+      "loss": 0.54,
+      "step": 22496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5607206976147734e-05,
+      "loss": 0.4625,
+      "step": 22497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5606565786099e-05,
+      "loss": 0.4707,
+      "step": 22498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560592459605027e-05,
+      "loss": 0.52,
+      "step": 22499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5605283406001536e-05,
+      "loss": 0.4361,
+      "step": 22500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560464221595281e-05,
+      "loss": 0.4485,
+      "step": 22501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560400102590408e-05,
+      "loss": 0.5027,
+      "step": 22502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560335983585535e-05,
+      "loss": 0.4875,
+      "step": 22503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560271864580662e-05,
+      "loss": 0.4695,
+      "step": 22504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560207745575789e-05,
+      "loss": 0.525,
+      "step": 22505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560143626570916e-05,
+      "loss": 0.4542,
+      "step": 22506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5600795075660425e-05,
+      "loss": 0.5636,
+      "step": 22507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5600153885611696e-05,
+      "loss": 0.5358,
+      "step": 22508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559951269556296e-05,
+      "loss": 0.4813,
+      "step": 22509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559887150551424e-05,
+      "loss": 0.5991,
+      "step": 22510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559823031546551e-05,
+      "loss": 0.5107,
+      "step": 22511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5597589125416776e-05,
+      "loss": 0.6058,
+      "step": 22512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559694793536805e-05,
+      "loss": 0.4818,
+      "step": 22513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559630674531931e-05,
+      "loss": 0.5897,
+      "step": 22514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5595665555270585e-05,
+      "loss": 0.4353,
+      "step": 22515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559502436522185e-05,
+      "loss": 0.5195,
+      "step": 22516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559438317517312e-05,
+      "loss": 0.4428,
+      "step": 22517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559374198512439e-05,
+      "loss": 0.4834,
+      "step": 22518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5593100795075665e-05,
+      "loss": 0.4672,
+      "step": 22519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5592459605026937e-05,
+      "loss": 0.3944,
+      "step": 22520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55918184149782e-05,
+      "loss": 0.5041,
+      "step": 22521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559117722492947e-05,
+      "loss": 0.4832,
+      "step": 22522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559053603488074e-05,
+      "loss": 0.4463,
+      "step": 22523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558989484483201e-05,
+      "loss": 0.5737,
+      "step": 22524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5589253654783275e-05,
+      "loss": 0.5176,
+      "step": 22525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5588612464734547e-05,
+      "loss": 0.4906,
+      "step": 22526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558797127468582e-05,
+      "loss": 0.4432,
+      "step": 22527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558733008463709e-05,
+      "loss": 0.5137,
+      "step": 22528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558668889458836e-05,
+      "loss": 0.4318,
+      "step": 22529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5586047704539627e-05,
+      "loss": 0.5598,
+      "step": 22530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55854065144909e-05,
+      "loss": 0.5172,
+      "step": 22531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558476532444216e-05,
+      "loss": 0.5705,
+      "step": 22532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5584124134393435e-05,
+      "loss": 0.4672,
+      "step": 22533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558348294434471e-05,
+      "loss": 0.6287,
+      "step": 22534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558284175429598e-05,
+      "loss": 0.5016,
+      "step": 22535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558220056424724e-05,
+      "loss": 0.5992,
+      "step": 22536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5581559374198515e-05,
+      "loss": 0.4487,
+      "step": 22537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558091818414979e-05,
+      "loss": 0.5302,
+      "step": 22538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558027699410105e-05,
+      "loss": 0.5539,
+      "step": 22539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5579635804052323e-05,
+      "loss": 0.466,
+      "step": 22540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557899461400359e-05,
+      "loss": 0.5075,
+      "step": 22541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557835342395486e-05,
+      "loss": 0.5456,
+      "step": 22542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557771223390613e-05,
+      "loss": 0.5984,
+      "step": 22543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5577071043857404e-05,
+      "loss": 0.4563,
+      "step": 22544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557642985380867e-05,
+      "loss": 0.583,
+      "step": 22545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557578866375994e-05,
+      "loss": 0.4245,
+      "step": 22546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557514747371121e-05,
+      "loss": 0.445,
+      "step": 22547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557450628366248e-05,
+      "loss": 0.5707,
+      "step": 22548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557386509361375e-05,
+      "loss": 0.4423,
+      "step": 22549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557322390356502e-05,
+      "loss": 0.4211,
+      "step": 22550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557258271351629e-05,
+      "loss": 0.5703,
+      "step": 22551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557194152346756e-05,
+      "loss": 0.477,
+      "step": 22552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557130033341883e-05,
+      "loss": 0.3952,
+      "step": 22553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5570659143370094e-05,
+      "loss": 0.5696,
+      "step": 22554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5570017953321365e-05,
+      "loss": 0.4597,
+      "step": 22555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556937676327264e-05,
+      "loss": 0.4657,
+      "step": 22556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55687355732239e-05,
+      "loss": 0.5616,
+      "step": 22557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5568094383175174e-05,
+      "loss": 0.5336,
+      "step": 22558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5567453193126445e-05,
+      "loss": 0.5264,
+      "step": 22559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556681200307772e-05,
+      "loss": 0.3551,
+      "step": 22560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556617081302898e-05,
+      "loss": 0.4623,
+      "step": 22561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5565529622980254e-05,
+      "loss": 0.5505,
+      "step": 22562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556488843293152e-05,
+      "loss": 0.6317,
+      "step": 22563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556424724288279e-05,
+      "loss": 0.4395,
+      "step": 22564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556360605283406e-05,
+      "loss": 0.4292,
+      "step": 22565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5562964862785334e-05,
+      "loss": 0.4955,
+      "step": 22566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5562323672736606e-05,
+      "loss": 0.4474,
+      "step": 22567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556168248268787e-05,
+      "loss": 0.5456,
+      "step": 22568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556104129263914e-05,
+      "loss": 0.551,
+      "step": 22569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556040010259041e-05,
+      "loss": 0.49,
+      "step": 22570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555975891254168e-05,
+      "loss": 0.4487,
+      "step": 22571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5559117722492944e-05,
+      "loss": 0.403,
+      "step": 22572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5558476532444216e-05,
+      "loss": 0.4603,
+      "step": 22573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555783534239549e-05,
+      "loss": 0.5975,
+      "step": 22574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555719415234676e-05,
+      "loss": 0.54,
+      "step": 22575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555655296229803e-05,
+      "loss": 0.4709,
+      "step": 22576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5555911772249296e-05,
+      "loss": 0.4219,
+      "step": 22577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555527058220057e-05,
+      "loss": 0.3645,
+      "step": 22578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555462939215183e-05,
+      "loss": 0.4,
+      "step": 22579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5553988202103104e-05,
+      "loss": 0.4622,
+      "step": 22580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555334701205437e-05,
+      "loss": 0.4914,
+      "step": 22581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555270582200565e-05,
+      "loss": 0.5229,
+      "step": 22582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555206463195692e-05,
+      "loss": 0.5796,
+      "step": 22583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5551423441908184e-05,
+      "loss": 0.4932,
+      "step": 22584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5550782251859456e-05,
+      "loss": 0.4446,
+      "step": 22585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555014106181072e-05,
+      "loss": 0.605,
+      "step": 22586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554949987176199e-05,
+      "loss": 0.6238,
+      "step": 22587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554885868171326e-05,
+      "loss": 0.4694,
+      "step": 22588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554821749166453e-05,
+      "loss": 0.4687,
+      "step": 22589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55475763016158e-05,
+      "loss": 0.4381,
+      "step": 22590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554693511156707e-05,
+      "loss": 0.5064,
+      "step": 22591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5546293921518344e-05,
+      "loss": 0.5106,
+      "step": 22592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554565273146961e-05,
+      "loss": 0.5056,
+      "step": 22593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554501154142088e-05,
+      "loss": 0.5325,
+      "step": 22594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5544370351372146e-05,
+      "loss": 0.5329,
+      "step": 22595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554372916132342e-05,
+      "loss": 0.4522,
+      "step": 22596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554308797127468e-05,
+      "loss": 0.5248,
+      "step": 22597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5542446781225954e-05,
+      "loss": 0.6309,
+      "step": 22598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5541805591177226e-05,
+      "loss": 0.4522,
+      "step": 22599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55411644011285e-05,
+      "loss": 0.473,
+      "step": 22600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554052321107977e-05,
+      "loss": 0.5685,
+      "step": 22601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5539882021031034e-05,
+      "loss": 0.4921,
+      "step": 22602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5539240830982306e-05,
+      "loss": 0.5068,
+      "step": 22603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553859964093357e-05,
+      "loss": 0.4855,
+      "step": 22604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553795845088484e-05,
+      "loss": 0.5123,
+      "step": 22605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5537317260836114e-05,
+      "loss": 0.5426,
+      "step": 22606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5536676070787386e-05,
+      "loss": 0.5284,
+      "step": 22607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553603488073865e-05,
+      "loss": 0.5253,
+      "step": 22608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553539369068992e-05,
+      "loss": 0.4336,
+      "step": 22609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5534752500641195e-05,
+      "loss": 0.5766,
+      "step": 22610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553411131059246e-05,
+      "loss": 0.5108,
+      "step": 22611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553347012054373e-05,
+      "loss": 0.4282,
+      "step": 22612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5532828930494996e-05,
+      "loss": 0.4997,
+      "step": 22613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553218774044627e-05,
+      "loss": 0.4009,
+      "step": 22614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553154655039754e-05,
+      "loss": 0.4644,
+      "step": 22615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553090536034881e-05,
+      "loss": 0.4817,
+      "step": 22616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5530264170300076e-05,
+      "loss": 0.369,
+      "step": 22617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552962298025135e-05,
+      "loss": 0.4415,
+      "step": 22618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552898179020262e-05,
+      "loss": 0.4192,
+      "step": 22619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5528340600153885e-05,
+      "loss": 0.5629,
+      "step": 22620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5527699410105156e-05,
+      "loss": 0.5415,
+      "step": 22621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552705822005643e-05,
+      "loss": 0.6457,
+      "step": 22622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55264170300077e-05,
+      "loss": 0.5805,
+      "step": 22623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5525775839958965e-05,
+      "loss": 0.4731,
+      "step": 22624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5525134649910236e-05,
+      "loss": 0.4628,
+      "step": 22625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55244934598615e-05,
+      "loss": 0.4183,
+      "step": 22626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552385226981277e-05,
+      "loss": 0.5163,
+      "step": 22627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5523211079764045e-05,
+      "loss": 0.5078,
+      "step": 22628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552256988971531e-05,
+      "loss": 0.5285,
+      "step": 22629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552192869966658e-05,
+      "loss": 0.4922,
+      "step": 22630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552128750961785e-05,
+      "loss": 0.4911,
+      "step": 22631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5520646319569125e-05,
+      "loss": 0.475,
+      "step": 22632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552000512952039e-05,
+      "loss": 0.4973,
+      "step": 22633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551936393947166e-05,
+      "loss": 0.5708,
+      "step": 22634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5518722749422926e-05,
+      "loss": 0.466,
+      "step": 22635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55180815593742e-05,
+      "loss": 0.5177,
+      "step": 22636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551744036932547e-05,
+      "loss": 0.608,
+      "step": 22637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551679917927674e-05,
+      "loss": 0.5218,
+      "step": 22638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551615798922801e-05,
+      "loss": 0.4421,
+      "step": 22639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551551679917928e-05,
+      "loss": 0.5688,
+      "step": 22640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551487560913055e-05,
+      "loss": 0.3702,
+      "step": 22641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5514234419081815e-05,
+      "loss": 0.5053,
+      "step": 22642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551359322903309e-05,
+      "loss": 0.4492,
+      "step": 22643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551295203898435e-05,
+      "loss": 0.4659,
+      "step": 22644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551231084893562e-05,
+      "loss": 0.4423,
+      "step": 22645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5511669658886895e-05,
+      "loss": 0.513,
+      "step": 22646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551102846883817e-05,
+      "loss": 0.482,
+      "step": 22647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551038727878944e-05,
+      "loss": 0.4558,
+      "step": 22648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5509746088740703e-05,
+      "loss": 0.5047,
+      "step": 22649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5509104898691975e-05,
+      "loss": 0.5601,
+      "step": 22650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550846370864324e-05,
+      "loss": 0.4962,
+      "step": 22651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550782251859451e-05,
+      "loss": 0.4621,
+      "step": 22652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550718132854578e-05,
+      "loss": 0.5403,
+      "step": 22653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5506540138497055e-05,
+      "loss": 0.4761,
+      "step": 22654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550589894844833e-05,
+      "loss": 0.6824,
+      "step": 22655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550525775839959e-05,
+      "loss": 0.4126,
+      "step": 22656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5504616568350864e-05,
+      "loss": 0.5043,
+      "step": 22657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550397537830213e-05,
+      "loss": 0.493,
+      "step": 22658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55033341882534e-05,
+      "loss": 0.3937,
+      "step": 22659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5502692998204665e-05,
+      "loss": 0.496,
+      "step": 22660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550205180815594e-05,
+      "loss": 0.5164,
+      "step": 22661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550141061810721e-05,
+      "loss": 0.4862,
+      "step": 22662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550076942805848e-05,
+      "loss": 0.6011,
+      "step": 22663
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550012823800975e-05,
+      "loss": 0.4505,
+      "step": 22664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549948704796102e-05,
+      "loss": 0.3958,
+      "step": 22665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549884585791229e-05,
+      "loss": 0.4728,
+      "step": 22666
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5498204667863554e-05,
+      "loss": 0.4986,
+      "step": 22667
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5497563477814825e-05,
+      "loss": 0.3943,
+      "step": 22668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549692228776609e-05,
+      "loss": 0.4981,
+      "step": 22669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549628109771737e-05,
+      "loss": 0.5602,
+      "step": 22670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549563990766864e-05,
+      "loss": 0.4077,
+      "step": 22671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5494998717619905e-05,
+      "loss": 0.4105,
+      "step": 22672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549435752757118e-05,
+      "loss": 0.4735,
+      "step": 22673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549371633752244e-05,
+      "loss": 0.4755,
+      "step": 22674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5493075147473714e-05,
+      "loss": 0.4839,
+      "step": 22675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549243395742498e-05,
+      "loss": 0.5638,
+      "step": 22676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549179276737625e-05,
+      "loss": 0.4836,
+      "step": 22677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549115157732752e-05,
+      "loss": 0.4073,
+      "step": 22678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5490510387278794e-05,
+      "loss": 0.5021,
+      "step": 22679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5489869197230066e-05,
+      "loss": 0.4409,
+      "step": 22680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548922800718133e-05,
+      "loss": 0.553,
+      "step": 22681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.54885868171326e-05,
+      "loss": 0.6161,
+      "step": 22682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548794562708387e-05,
+      "loss": 0.4734,
+      "step": 22683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548730443703514e-05,
+      "loss": 0.5117,
+      "step": 22684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5486663246986404e-05,
+      "loss": 0.4445,
+      "step": 22685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5486022056937676e-05,
+      "loss": 0.5298,
+      "step": 22686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548538086688895e-05,
+      "loss": 0.4353,
+      "step": 22687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548473967684022e-05,
+      "loss": 0.4782,
+      "step": 22688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548409848679149e-05,
+      "loss": 0.4856,
+      "step": 22689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5483457296742756e-05,
+      "loss": 0.4723,
+      "step": 22690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548281610669403e-05,
+      "loss": 0.4755,
+      "step": 22691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548217491664529e-05,
+      "loss": 0.5098,
+      "step": 22692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5481533726596564e-05,
+      "loss": 0.3836,
+      "step": 22693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5480892536547836e-05,
+      "loss": 0.416,
+      "step": 22694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548025134649911e-05,
+      "loss": 0.4408,
+      "step": 22695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547961015645037e-05,
+      "loss": 0.5725,
+      "step": 22696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5478968966401644e-05,
+      "loss": 0.457,
+      "step": 22697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5478327776352916e-05,
+      "loss": 0.3719,
+      "step": 22698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547768658630418e-05,
+      "loss": 0.4704,
+      "step": 22699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547704539625545e-05,
+      "loss": 0.4224,
+      "step": 22700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547640420620672e-05,
+      "loss": 0.5585,
+      "step": 22701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547576301615799e-05,
+      "loss": 0.5157,
+      "step": 22702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547512182610926e-05,
+      "loss": 0.6623,
+      "step": 22703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547448063606053e-05,
+      "loss": 0.4143,
+      "step": 22704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.54738394460118e-05,
+      "loss": 0.456,
+      "step": 22705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547319825596307e-05,
+      "loss": 0.4988,
+      "step": 22706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547255706591434e-05,
+      "loss": 0.4872,
+      "step": 22707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5471915875865606e-05,
+      "loss": 0.5407,
+      "step": 22708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547127468581688e-05,
+      "loss": 0.5273,
+      "step": 22709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547063349576815e-05,
+      "loss": 0.6564,
+      "step": 22710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546999230571942e-05,
+      "loss": 0.4659,
+      "step": 22711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5469351115670686e-05,
+      "loss": 0.5412,
+      "step": 22712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546870992562196e-05,
+      "loss": 0.5171,
+      "step": 22713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546806873557322e-05,
+      "loss": 0.3686,
+      "step": 22714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5467427545524494e-05,
+      "loss": 0.5025,
+      "step": 22715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5466786355475766e-05,
+      "loss": 0.5051,
+      "step": 22716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546614516542703e-05,
+      "loss": 0.4393,
+      "step": 22717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.54655039753783e-05,
+      "loss": 0.4587,
+      "step": 22718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5464862785329574e-05,
+      "loss": 0.5149,
+      "step": 22719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5464221595280846e-05,
+      "loss": 0.5345,
+      "step": 22720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546358040523211e-05,
+      "loss": 0.5529,
+      "step": 22721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546293921518338e-05,
+      "loss": 0.5161,
+      "step": 22722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546229802513465e-05,
+      "loss": 0.472,
+      "step": 22723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546165683508592e-05,
+      "loss": 0.5276,
+      "step": 22724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546101564503719e-05,
+      "loss": 0.4284,
+      "step": 22725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546037445498846e-05,
+      "loss": 0.487,
+      "step": 22726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5459733264939735e-05,
+      "loss": 0.3341,
+      "step": 22727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5459092074891e-05,
+      "loss": 0.4557,
+      "step": 22728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.545845088484227e-05,
+      "loss": 0.5688,
+      "step": 22729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5457809694793536e-05,
+      "loss": 0.4725,
+      "step": 22730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.545716850474481e-05,
+      "loss": 0.6405,
+      "step": 22731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.545652731469607e-05,
+      "loss": 0.561,
+      "step": 22732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5455886124647345e-05,
+      "loss": 0.4196,
+      "step": 22733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5455244934598616e-05,
+      "loss": 0.5102,
+      "step": 22734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.545460374454989e-05,
+      "loss": 0.5318,
+      "step": 22735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.545396255450116e-05,
+      "loss": 0.5367,
+      "step": 22736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5453321364452425e-05,
+      "loss": 0.4566,
+      "step": 22737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5452680174403696e-05,
+      "loss": 0.5222,
+      "step": 22738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.545203898435496e-05,
+      "loss": 0.4879,
+      "step": 22739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.545139779430623e-05,
+      "loss": 0.491,
+      "step": 22740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.54507566042575e-05,
+      "loss": 0.5136,
+      "step": 22741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5450115414208777e-05,
+      "loss": 0.3932,
+      "step": 22742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544947422416005e-05,
+      "loss": 0.4747,
+      "step": 22743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544883303411131e-05,
+      "loss": 0.4995,
+      "step": 22744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5448191844062585e-05,
+      "loss": 0.4501,
+      "step": 22745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544755065401385e-05,
+      "loss": 0.5284,
+      "step": 22746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544690946396512e-05,
+      "loss": 0.5194,
+      "step": 22747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5446268273916386e-05,
+      "loss": 0.5293,
+      "step": 22748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544562708386766e-05,
+      "loss": 0.5731,
+      "step": 22749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544498589381893e-05,
+      "loss": 0.514,
+      "step": 22750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.54443447037702e-05,
+      "loss": 0.5782,
+      "step": 22751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544370351372147e-05,
+      "loss": 0.5016,
+      "step": 22752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544306232367274e-05,
+      "loss": 0.6087,
+      "step": 22753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544242113362401e-05,
+      "loss": 0.502,
+      "step": 22754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5441779943575275e-05,
+      "loss": 0.459,
+      "step": 22755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544113875352655e-05,
+      "loss": 0.5164,
+      "step": 22756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.544049756347781e-05,
+      "loss": 0.543,
+      "step": 22757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.543985637342908e-05,
+      "loss": 0.5228,
+      "step": 22758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5439215183380355e-05,
+      "loss": 0.5278,
+      "step": 22759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543857399333163e-05,
+      "loss": 0.3913,
+      "step": 22760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54379328032829e-05,
+      "loss": 0.5269,
+      "step": 22761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5437291613234163e-05,
+      "loss": 0.5,
+      "step": 22762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5436650423185435e-05,
+      "loss": 0.5918,
+      "step": 22763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54360092331367e-05,
+      "loss": 0.4356,
+      "step": 22764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543536804308797e-05,
+      "loss": 0.5261,
+      "step": 22765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5434726853039244e-05,
+      "loss": 0.5128,
+      "step": 22766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5434085662990515e-05,
+      "loss": 0.5251,
+      "step": 22767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543344447294178e-05,
+      "loss": 0.5555,
+      "step": 22768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543280328289305e-05,
+      "loss": 0.5741,
+      "step": 22769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5432162092844324e-05,
+      "loss": 0.4889,
+      "step": 22770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543152090279559e-05,
+      "loss": 0.4353,
+      "step": 22771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543087971274686e-05,
+      "loss": 0.456,
+      "step": 22772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5430238522698125e-05,
+      "loss": 0.6156,
+      "step": 22773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54295973326494e-05,
+      "loss": 0.5801,
+      "step": 22774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542895614260067e-05,
+      "loss": 0.5254,
+      "step": 22775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542831495255194e-05,
+      "loss": 0.4448,
+      "step": 22776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5427673762503205e-05,
+      "loss": 0.5489,
+      "step": 22777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542703257245448e-05,
+      "loss": 0.4296,
+      "step": 22778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542639138240575e-05,
+      "loss": 0.5396,
+      "step": 22779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5425750192357014e-05,
+      "loss": 0.5286,
+      "step": 22780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5425109002308285e-05,
+      "loss": 0.4931,
+      "step": 22781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542446781225956e-05,
+      "loss": 0.4621,
+      "step": 22782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542382662221083e-05,
+      "loss": 0.5373,
+      "step": 22783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5423185432162094e-05,
+      "loss": 0.5356,
+      "step": 22784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5422544242113365e-05,
+      "loss": 0.4114,
+      "step": 22785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542190305206463e-05,
+      "loss": 0.4028,
+      "step": 22786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54212618620159e-05,
+      "loss": 0.507,
+      "step": 22787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5420620671967174e-05,
+      "loss": 0.4953,
+      "step": 22788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541997948191844e-05,
+      "loss": 0.452,
+      "step": 22789
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541933829186971e-05,
+      "loss": 0.5572,
+      "step": 22790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541869710182098e-05,
+      "loss": 0.5763,
+      "step": 22791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5418055911772254e-05,
+      "loss": 0.6569,
+      "step": 22792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541741472172352e-05,
+      "loss": 0.4583,
+      "step": 22793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541677353167479e-05,
+      "loss": 0.5408,
+      "step": 22794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5416132341626056e-05,
+      "loss": 0.4206,
+      "step": 22795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541549115157733e-05,
+      "loss": 0.5046,
+      "step": 22796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54148499615286e-05,
+      "loss": 0.5383,
+      "step": 22797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541420877147987e-05,
+      "loss": 0.635,
+      "step": 22798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541356758143114e-05,
+      "loss": 0.571,
+      "step": 22799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541292639138241e-05,
+      "loss": 0.4563,
+      "step": 22800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541228520133368e-05,
+      "loss": 0.4639,
+      "step": 22801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5411644011284944e-05,
+      "loss": 0.4479,
+      "step": 22802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5411002821236216e-05,
+      "loss": 0.5511,
+      "step": 22803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541036163118748e-05,
+      "loss": 0.4726,
+      "step": 22804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540972044113875e-05,
+      "loss": 0.5346,
+      "step": 22805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5409079251090024e-05,
+      "loss": 0.4217,
+      "step": 22806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5408438061041296e-05,
+      "loss": 0.5044,
+      "step": 22807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540779687099257e-05,
+      "loss": 0.5159,
+      "step": 22808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540715568094383e-05,
+      "loss": 0.4842,
+      "step": 22809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5406514490895104e-05,
+      "loss": 0.5806,
+      "step": 22810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540587330084637e-05,
+      "loss": 0.3982,
+      "step": 22811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540523211079764e-05,
+      "loss": 0.5415,
+      "step": 22812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5404590920748906e-05,
+      "loss": 0.5888,
+      "step": 22813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5403949730700184e-05,
+      "loss": 0.5635,
+      "step": 22814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5403308540651456e-05,
+      "loss": 0.4083,
+      "step": 22815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540266735060272e-05,
+      "loss": 0.4782,
+      "step": 22816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540202616055399e-05,
+      "loss": 0.6501,
+      "step": 22817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540138497050526e-05,
+      "loss": 0.5384,
+      "step": 22818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540074378045653e-05,
+      "loss": 0.3736,
+      "step": 22819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5400102590407794e-05,
+      "loss": 0.5861,
+      "step": 22820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5399461400359066e-05,
+      "loss": 0.5106,
+      "step": 22821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539882021031034e-05,
+      "loss": 0.4428,
+      "step": 22822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539817902026161e-05,
+      "loss": 0.5407,
+      "step": 22823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539753783021288e-05,
+      "loss": 0.4454,
+      "step": 22824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5396896640164146e-05,
+      "loss": 0.529,
+      "step": 22825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539625545011542e-05,
+      "loss": 0.403,
+      "step": 22826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539561426006668e-05,
+      "loss": 0.4104,
+      "step": 22827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5394973070017954e-05,
+      "loss": 0.4866,
+      "step": 22828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539433187996922e-05,
+      "loss": 0.5112,
+      "step": 22829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53936906899205e-05,
+      "loss": 0.5728,
+      "step": 22830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539304949987176e-05,
+      "loss": 0.4985,
+      "step": 22831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5392408309823035e-05,
+      "loss": 0.5082,
+      "step": 22832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5391767119774306e-05,
+      "loss": 0.5316,
+      "step": 22833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539112592972557e-05,
+      "loss": 0.6441,
+      "step": 22834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539048473967684e-05,
+      "loss": 0.4836,
+      "step": 22835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538984354962811e-05,
+      "loss": 0.6336,
+      "step": 22836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538920235957938e-05,
+      "loss": 0.477,
+      "step": 22837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538856116953065e-05,
+      "loss": 0.4756,
+      "step": 22838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538791997948192e-05,
+      "loss": 0.5025,
+      "step": 22839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538727878943319e-05,
+      "loss": 0.4456,
+      "step": 22840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538663759938446e-05,
+      "loss": 0.5406,
+      "step": 22841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538599640933573e-05,
+      "loss": 0.4223,
+      "step": 22842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5385355219286996e-05,
+      "loss": 0.493,
+      "step": 22843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538471402923827e-05,
+      "loss": 0.4576,
+      "step": 22844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538407283918953e-05,
+      "loss": 0.5024,
+      "step": 22845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5383431649140805e-05,
+      "loss": 0.5531,
+      "step": 22846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5382790459092076e-05,
+      "loss": 0.4147,
+      "step": 22847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538214926904335e-05,
+      "loss": 0.4516,
+      "step": 22848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538150807899461e-05,
+      "loss": 0.4856,
+      "step": 22849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5380866888945885e-05,
+      "loss": 0.4785,
+      "step": 22850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5380225698897156e-05,
+      "loss": 0.4662,
+      "step": 22851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537958450884842e-05,
+      "loss": 0.4758,
+      "step": 22852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537894331879969e-05,
+      "loss": 0.6002,
+      "step": 22853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5378302128750965e-05,
+      "loss": 0.4617,
+      "step": 22854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5377660938702237e-05,
+      "loss": 0.5007,
+      "step": 22855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53770197486535e-05,
+      "loss": 0.4631,
+      "step": 22856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537637855860477e-05,
+      "loss": 0.5615,
+      "step": 22857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5375737368556045e-05,
+      "loss": 0.4528,
+      "step": 22858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537509617850731e-05,
+      "loss": 0.4539,
+      "step": 22859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537445498845858e-05,
+      "loss": 0.5283,
+      "step": 22860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5373813798409847e-05,
+      "loss": 0.5486,
+      "step": 22861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537317260836112e-05,
+      "loss": 0.5569,
+      "step": 22862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537253141831239e-05,
+      "loss": 0.4856,
+      "step": 22863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537189022826366e-05,
+      "loss": 0.5755,
+      "step": 22864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537124903821493e-05,
+      "loss": 0.5316,
+      "step": 22865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53706078481662e-05,
+      "loss": 0.45,
+      "step": 22866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536996665811747e-05,
+      "loss": 0.5119,
+      "step": 22867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5369325468068735e-05,
+      "loss": 0.5276,
+      "step": 22868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536868427802001e-05,
+      "loss": 0.5271,
+      "step": 22869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536804308797128e-05,
+      "loss": 0.5431,
+      "step": 22870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536740189792255e-05,
+      "loss": 0.3734,
+      "step": 22871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5366760707873815e-05,
+      "loss": 0.4586,
+      "step": 22872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536611951782509e-05,
+      "loss": 0.5985,
+      "step": 22873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536547832777635e-05,
+      "loss": 0.5297,
+      "step": 22874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5364837137727623e-05,
+      "loss": 0.5899,
+      "step": 22875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5364195947678895e-05,
+      "loss": 0.3989,
+      "step": 22876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536355475763016e-05,
+      "loss": 0.4161,
+      "step": 22877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536291356758143e-05,
+      "loss": 0.4594,
+      "step": 22878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5362272377532704e-05,
+      "loss": 0.465,
+      "step": 22879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5361631187483975e-05,
+      "loss": 0.4322,
+      "step": 22880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536098999743524e-05,
+      "loss": 0.5258,
+      "step": 22881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536034880738651e-05,
+      "loss": 0.4841,
+      "step": 22882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535970761733778e-05,
+      "loss": 0.5676,
+      "step": 22883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535906642728905e-05,
+      "loss": 0.5214,
+      "step": 22884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535842523724032e-05,
+      "loss": 0.4876,
+      "step": 22885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535778404719159e-05,
+      "loss": 0.5698,
+      "step": 22886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5357142857142864e-05,
+      "loss": 0.3442,
+      "step": 22887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535650166709413e-05,
+      "loss": 0.5265,
+      "step": 22888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53558604770454e-05,
+      "loss": 0.5312,
+      "step": 22889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5355219286996665e-05,
+      "loss": 0.4673,
+      "step": 22890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535457809694794e-05,
+      "loss": 0.3167,
+      "step": 22891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53539369068992e-05,
+      "loss": 0.547,
+      "step": 22892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5353295716850474e-05,
+      "loss": 0.3986,
+      "step": 22893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5352654526801745e-05,
+      "loss": 0.4096,
+      "step": 22894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535201333675302e-05,
+      "loss": 0.4497,
+      "step": 22895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535137214670429e-05,
+      "loss": 0.363,
+      "step": 22896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5350730956655554e-05,
+      "loss": 0.4719,
+      "step": 22897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5350089766606826e-05,
+      "loss": 0.5234,
+      "step": 22898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534944857655809e-05,
+      "loss": 0.4278,
+      "step": 22899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534880738650936e-05,
+      "loss": 0.451,
+      "step": 22900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534816619646063e-05,
+      "loss": 0.5349,
+      "step": 22901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5347525006411906e-05,
+      "loss": 0.5583,
+      "step": 22902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534688381636318e-05,
+      "loss": 0.4055,
+      "step": 22903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534624262631444e-05,
+      "loss": 0.5605,
+      "step": 22904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5345601436265714e-05,
+      "loss": 0.5736,
+      "step": 22905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534496024621698e-05,
+      "loss": 0.4471,
+      "step": 22906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534431905616825e-05,
+      "loss": 0.451,
+      "step": 22907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5343677866119516e-05,
+      "loss": 0.4814,
+      "step": 22908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534303667607079e-05,
+      "loss": 0.3637,
+      "step": 22909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534239548602206e-05,
+      "loss": 0.5077,
+      "step": 22910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534175429597333e-05,
+      "loss": 0.4618,
+      "step": 22911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53411131059246e-05,
+      "loss": 0.4185,
+      "step": 22912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534047191587587e-05,
+      "loss": 0.5396,
+      "step": 22913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533983072582714e-05,
+      "loss": 0.5212,
+      "step": 22914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5339189535778404e-05,
+      "loss": 0.3981,
+      "step": 22915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5338548345729676e-05,
+      "loss": 0.4856,
+      "step": 22916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533790715568094e-05,
+      "loss": 0.5026,
+      "step": 22917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533726596563222e-05,
+      "loss": 0.5305,
+      "step": 22918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5336624775583484e-05,
+      "loss": 0.4875,
+      "step": 22919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5335983585534756e-05,
+      "loss": 0.484,
+      "step": 22920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533534239548603e-05,
+      "loss": 0.595,
+      "step": 22921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533470120543729e-05,
+      "loss": 0.4586,
+      "step": 22922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5334060015388564e-05,
+      "loss": 0.3729,
+      "step": 22923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533341882533983e-05,
+      "loss": 0.4784,
+      "step": 22924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53327776352911e-05,
+      "loss": 0.6003,
+      "step": 22925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533213644524237e-05,
+      "loss": 0.4813,
+      "step": 22926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5331495255193644e-05,
+      "loss": 0.551,
+      "step": 22927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533085406514491e-05,
+      "loss": 0.4144,
+      "step": 22928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533021287509618e-05,
+      "loss": 0.5712,
+      "step": 22929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532957168504745e-05,
+      "loss": 0.4444,
+      "step": 22930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532893049499872e-05,
+      "loss": 0.6027,
+      "step": 22931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532828930494999e-05,
+      "loss": 0.448,
+      "step": 22932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5327648114901254e-05,
+      "loss": 0.5655,
+      "step": 22933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5327006924852526e-05,
+      "loss": 0.5674,
+      "step": 22934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53263657348038e-05,
+      "loss": 0.517,
+      "step": 22935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532572454475507e-05,
+      "loss": 0.4715,
+      "step": 22936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5325083354706334e-05,
+      "loss": 0.5293,
+      "step": 22937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5324442164657606e-05,
+      "loss": 0.5372,
+      "step": 22938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532380097460888e-05,
+      "loss": 0.4115,
+      "step": 22939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532315978456014e-05,
+      "loss": 0.4484,
+      "step": 22940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5322518594511414e-05,
+      "loss": 0.4835,
+      "step": 22941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5321877404462686e-05,
+      "loss": 0.5211,
+      "step": 22942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532123621441396e-05,
+      "loss": 0.5443,
+      "step": 22943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532059502436522e-05,
+      "loss": 0.4044,
+      "step": 22944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5319953834316495e-05,
+      "loss": 0.5351,
+      "step": 22945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531931264426776e-05,
+      "loss": 0.5702,
+      "step": 22946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531867145421903e-05,
+      "loss": 0.5812,
+      "step": 22947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53180302641703e-05,
+      "loss": 0.5668,
+      "step": 22948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531738907412157e-05,
+      "loss": 0.5444,
+      "step": 22949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531674788407284e-05,
+      "loss": 0.5428,
+      "step": 22950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531610669402411e-05,
+      "loss": 0.4639,
+      "step": 22951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531546550397538e-05,
+      "loss": 0.6148,
+      "step": 22952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531482431392665e-05,
+      "loss": 0.5887,
+      "step": 22953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531418312387792e-05,
+      "loss": 0.4947,
+      "step": 22954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5313541933829185e-05,
+      "loss": 0.5771,
+      "step": 22955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5312900743780456e-05,
+      "loss": 0.4312,
+      "step": 22956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531225955373173e-05,
+      "loss": 0.5727,
+      "step": 22957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5311618363683e-05,
+      "loss": 0.5134,
+      "step": 22958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531097717363427e-05,
+      "loss": 0.4365,
+      "step": 22959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5310335983585536e-05,
+      "loss": 0.533,
+      "step": 22960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530969479353681e-05,
+      "loss": 0.5178,
+      "step": 22961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530905360348807e-05,
+      "loss": 0.4232,
+      "step": 22962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5308412413439345e-05,
+      "loss": 0.573,
+      "step": 22963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530777122339061e-05,
+      "loss": 0.5668,
+      "step": 22964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530713003334188e-05,
+      "loss": 0.5069,
+      "step": 22965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530648884329315e-05,
+      "loss": 0.4435,
+      "step": 22966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5305847653244425e-05,
+      "loss": 0.5632,
+      "step": 22967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5305206463195697e-05,
+      "loss": 0.5133,
+      "step": 22968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530456527314696e-05,
+      "loss": 0.4968,
+      "step": 22969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530392408309823e-05,
+      "loss": 0.432,
+      "step": 22970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53032828930495e-05,
+      "loss": 0.4485,
+      "step": 22971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530264170300077e-05,
+      "loss": 0.4902,
+      "step": 22972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5302000512952035e-05,
+      "loss": 0.5086,
+      "step": 22973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530135932290331e-05,
+      "loss": 0.4487,
+      "step": 22974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5300718132854585e-05,
+      "loss": 0.3956,
+      "step": 22975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530007694280585e-05,
+      "loss": 0.503,
+      "step": 22976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529943575275712e-05,
+      "loss": 0.5427,
+      "step": 22977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529879456270839e-05,
+      "loss": 0.4747,
+      "step": 22978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529815337265966e-05,
+      "loss": 0.4573,
+      "step": 22979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529751218261092e-05,
+      "loss": 0.5293,
+      "step": 22980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5296870992562195e-05,
+      "loss": 0.4678,
+      "step": 22981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529622980251347e-05,
+      "loss": 0.5549,
+      "step": 22982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529558861246474e-05,
+      "loss": 0.5179,
+      "step": 22983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529494742241601e-05,
+      "loss": 0.478,
+      "step": 22984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5294306232367275e-05,
+      "loss": 0.5108,
+      "step": 22985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529366504231855e-05,
+      "loss": 0.4951,
+      "step": 22986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529302385226981e-05,
+      "loss": 0.5321,
+      "step": 22987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5292382662221083e-05,
+      "loss": 0.4439,
+      "step": 22988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529174147217235e-05,
+      "loss": 0.5501,
+      "step": 22989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529110028212363e-05,
+      "loss": 0.5036,
+      "step": 22990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529045909207489e-05,
+      "loss": 0.6323,
+      "step": 22991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5289817902026164e-05,
+      "loss": 0.4234,
+      "step": 22992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5289176711977435e-05,
+      "loss": 0.2996,
+      "step": 22993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.52885355219287e-05,
+      "loss": 0.5841,
+      "step": 22994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528789433187997e-05,
+      "loss": 0.5238,
+      "step": 22995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528725314183124e-05,
+      "loss": 0.4829,
+      "step": 22996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528661195178251e-05,
+      "loss": 0.4607,
+      "step": 22997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528597076173378e-05,
+      "loss": 0.541,
+      "step": 22998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528532957168505e-05,
+      "loss": 0.5127,
+      "step": 22999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528468838163632e-05,
+      "loss": 0.4184,
+      "step": 23000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528404719158759e-05,
+      "loss": 0.4776,
+      "step": 23001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528340600153886e-05,
+      "loss": 0.5096,
+      "step": 23002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5282764811490125e-05,
+      "loss": 0.4868,
+      "step": 23003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.52821236214414e-05,
+      "loss": 0.5473,
+      "step": 23004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528148243139266e-05,
+      "loss": 0.5429,
+      "step": 23005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5280841241343934e-05,
+      "loss": 0.5249,
+      "step": 23006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5280200051295205e-05,
+      "loss": 0.4271,
+      "step": 23007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527955886124648e-05,
+      "loss": 0.3823,
+      "step": 23008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527891767119774e-05,
+      "loss": 0.3392,
+      "step": 23009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5278276481149014e-05,
+      "loss": 0.4589,
+      "step": 23010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5277635291100286e-05,
+      "loss": 0.4436,
+      "step": 23011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527699410105155e-05,
+      "loss": 0.4291,
+      "step": 23012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527635291100282e-05,
+      "loss": 0.5397,
+      "step": 23013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5275711720954094e-05,
+      "loss": 0.344,
+      "step": 23014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5275070530905366e-05,
+      "loss": 0.43,
+      "step": 23015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527442934085663e-05,
+      "loss": 0.3873,
+      "step": 23016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.52737881508079e-05,
+      "loss": 0.4571,
+      "step": 23017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527314696075917e-05,
+      "loss": 0.504,
+      "step": 23018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527250577071044e-05,
+      "loss": 0.415,
+      "step": 23019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527186458066171e-05,
+      "loss": 0.5376,
+      "step": 23020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5271223390612976e-05,
+      "loss": 0.5068,
+      "step": 23021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527058220056425e-05,
+      "loss": 0.5748,
+      "step": 23022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526994101051552e-05,
+      "loss": 0.4927,
+      "step": 23023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526929982046679e-05,
+      "loss": 0.5717,
+      "step": 23024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5268658630418056e-05,
+      "loss": 0.5574,
+      "step": 23025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526801744036933e-05,
+      "loss": 0.5327,
+      "step": 23026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526737625032059e-05,
+      "loss": 0.4045,
+      "step": 23027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5266735060271864e-05,
+      "loss": 0.5521,
+      "step": 23028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5266093870223136e-05,
+      "loss": 0.4514,
+      "step": 23029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526545268017441e-05,
+      "loss": 0.5316,
+      "step": 23030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526481149012568e-05,
+      "loss": 0.4744,
+      "step": 23031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5264170300076944e-05,
+      "loss": 0.4954,
+      "step": 23032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5263529110028216e-05,
+      "loss": 0.4939,
+      "step": 23033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526288791997948e-05,
+      "loss": 0.5718,
+      "step": 23034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526224672993075e-05,
+      "loss": 0.4188,
+      "step": 23035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526160553988202e-05,
+      "loss": 0.5307,
+      "step": 23036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526096434983329e-05,
+      "loss": 0.4166,
+      "step": 23037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526032315978456e-05,
+      "loss": 0.4691,
+      "step": 23038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525968196973583e-05,
+      "loss": 0.558,
+      "step": 23039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5259040779687104e-05,
+      "loss": 0.5307,
+      "step": 23040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525839958963837e-05,
+      "loss": 0.4123,
+      "step": 23041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525775839958964e-05,
+      "loss": 0.5004,
+      "step": 23042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5257117209540906e-05,
+      "loss": 0.5418,
+      "step": 23043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525647601949218e-05,
+      "loss": 0.4718,
+      "step": 23044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525583482944344e-05,
+      "loss": 0.4165,
+      "step": 23045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525519363939472e-05,
+      "loss": 0.5605,
+      "step": 23046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525455244934599e-05,
+      "loss": 0.4499,
+      "step": 23047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525391125929726e-05,
+      "loss": 0.4217,
+      "step": 23048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525327006924853e-05,
+      "loss": 0.5432,
+      "step": 23049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5252628879199794e-05,
+      "loss": 0.4099,
+      "step": 23050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5251987689151066e-05,
+      "loss": 0.4858,
+      "step": 23051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525134649910233e-05,
+      "loss": 0.4591,
+      "step": 23052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52507053090536e-05,
+      "loss": 0.4783,
+      "step": 23053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5250064119004874e-05,
+      "loss": 0.4999,
+      "step": 23054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5249422928956146e-05,
+      "loss": 0.5947,
+      "step": 23055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524878173890742e-05,
+      "loss": 0.4868,
+      "step": 23056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524814054885868e-05,
+      "loss": 0.5302,
+      "step": 23057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5247499358809955e-05,
+      "loss": 0.4813,
+      "step": 23058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524685816876122e-05,
+      "loss": 0.6136,
+      "step": 23059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524621697871249e-05,
+      "loss": 0.48,
+      "step": 23060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5245575788663756e-05,
+      "loss": 0.5812,
+      "step": 23061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5244934598615035e-05,
+      "loss": 0.5482,
+      "step": 23062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5244293408566306e-05,
+      "loss": 0.5483,
+      "step": 23063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524365221851757e-05,
+      "loss": 0.4263,
+      "step": 23064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524301102846884e-05,
+      "loss": 0.4771,
+      "step": 23065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524236983842011e-05,
+      "loss": 0.4848,
+      "step": 23066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524172864837138e-05,
+      "loss": 0.4749,
+      "step": 23067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5241087458322645e-05,
+      "loss": 0.4331,
+      "step": 23068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5240446268273916e-05,
+      "loss": 0.5109,
+      "step": 23069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523980507822519e-05,
+      "loss": 0.5474,
+      "step": 23070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523916388817646e-05,
+      "loss": 0.5584,
+      "step": 23071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523852269812773e-05,
+      "loss": 0.5468,
+      "step": 23072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5237881508078996e-05,
+      "loss": 0.4071,
+      "step": 23073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523724031803027e-05,
+      "loss": 0.4908,
+      "step": 23074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523659912798153e-05,
+      "loss": 0.5191,
+      "step": 23075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5235957937932805e-05,
+      "loss": 0.4699,
+      "step": 23076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523531674788407e-05,
+      "loss": 0.5917,
+      "step": 23077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523467555783535e-05,
+      "loss": 0.4186,
+      "step": 23078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523403436778661e-05,
+      "loss": 0.4506,
+      "step": 23079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5233393177737885e-05,
+      "loss": 0.5047,
+      "step": 23080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523275198768916e-05,
+      "loss": 0.5178,
+      "step": 23081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523211079764042e-05,
+      "loss": 0.5224,
+      "step": 23082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523146960759169e-05,
+      "loss": 0.5161,
+      "step": 23083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523082841754296e-05,
+      "loss": 0.5665,
+      "step": 23084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523018722749423e-05,
+      "loss": 0.5386,
+      "step": 23085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52295460374455e-05,
+      "loss": 0.5782,
+      "step": 23086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522890484739677e-05,
+      "loss": 0.413,
+      "step": 23087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522826365734804e-05,
+      "loss": 0.435,
+      "step": 23088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522762246729931e-05,
+      "loss": 0.5381,
+      "step": 23089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522698127725058e-05,
+      "loss": 0.5883,
+      "step": 23090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522634008720185e-05,
+      "loss": 0.5251,
+      "step": 23091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522569889715312e-05,
+      "loss": 0.3864,
+      "step": 23092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522505770710438e-05,
+      "loss": 0.5912,
+      "step": 23093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5224416517055655e-05,
+      "loss": 0.4757,
+      "step": 23094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522377532700693e-05,
+      "loss": 0.5745,
+      "step": 23095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52231341369582e-05,
+      "loss": 0.5078,
+      "step": 23096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5222492946909463e-05,
+      "loss": 0.5496,
+      "step": 23097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5221851756860735e-05,
+      "loss": 0.325,
+      "step": 23098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522121056681201e-05,
+      "loss": 0.4322,
+      "step": 23099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522056937676327e-05,
+      "loss": 0.5168,
+      "step": 23100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5219928186714544e-05,
+      "loss": 0.4449,
+      "step": 23101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5219286996665815e-05,
+      "loss": 0.4748,
+      "step": 23102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521864580661709e-05,
+      "loss": 0.4824,
+      "step": 23103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521800461656835e-05,
+      "loss": 0.4731,
+      "step": 23104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5217363426519624e-05,
+      "loss": 0.4599,
+      "step": 23105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521672223647089e-05,
+      "loss": 0.5195,
+      "step": 23106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521608104642216e-05,
+      "loss": 0.5245,
+      "step": 23107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521543985637343e-05,
+      "loss": 0.5268,
+      "step": 23108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52147986663247e-05,
+      "loss": 0.5762,
+      "step": 23109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521415747627597e-05,
+      "loss": 0.4971,
+      "step": 23110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521351628622724e-05,
+      "loss": 0.5675,
+      "step": 23111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521287509617851e-05,
+      "loss": 0.5204,
+      "step": 23112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521223390612978e-05,
+      "loss": 0.4787,
+      "step": 23113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521159271608105e-05,
+      "loss": 0.4588,
+      "step": 23114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5210951526032314e-05,
+      "loss": 0.4118,
+      "step": 23115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5210310335983585e-05,
+      "loss": 0.4636,
+      "step": 23116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520966914593486e-05,
+      "loss": 0.4524,
+      "step": 23117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520902795588613e-05,
+      "loss": 0.4548,
+      "step": 23118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52083867658374e-05,
+      "loss": 0.4923,
+      "step": 23119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5207745575788665e-05,
+      "loss": 0.4298,
+      "step": 23120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520710438573994e-05,
+      "loss": 0.4741,
+      "step": 23121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52064631956912e-05,
+      "loss": 0.4741,
+      "step": 23122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5205822005642474e-05,
+      "loss": 0.4915,
+      "step": 23123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520518081559374e-05,
+      "loss": 0.5901,
+      "step": 23124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520453962554501e-05,
+      "loss": 0.5569,
+      "step": 23125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520389843549628e-05,
+      "loss": 0.5108,
+      "step": 23126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5203257245447554e-05,
+      "loss": 0.6174,
+      "step": 23127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5202616055398826e-05,
+      "loss": 0.3811,
+      "step": 23128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520197486535009e-05,
+      "loss": 0.6184,
+      "step": 23129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520133367530136e-05,
+      "loss": 0.3436,
+      "step": 23130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520069248525263e-05,
+      "loss": 0.539,
+      "step": 23131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52000512952039e-05,
+      "loss": 0.6045,
+      "step": 23132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5199410105155164e-05,
+      "loss": 0.526,
+      "step": 23133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519876891510644e-05,
+      "loss": 0.48,
+      "step": 23134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5198127725057714e-05,
+      "loss": 0.4864,
+      "step": 23135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519748653500898e-05,
+      "loss": 0.5664,
+      "step": 23136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519684534496025e-05,
+      "loss": 0.4419,
+      "step": 23137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5196204154911516e-05,
+      "loss": 0.5593,
+      "step": 23138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519556296486279e-05,
+      "loss": 0.5422,
+      "step": 23139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519492177481405e-05,
+      "loss": 0.5642,
+      "step": 23140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5194280584765324e-05,
+      "loss": 0.5439,
+      "step": 23141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5193639394716596e-05,
+      "loss": 0.5044,
+      "step": 23142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519299820466787e-05,
+      "loss": 0.5304,
+      "step": 23143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519235701461914e-05,
+      "loss": 0.5009,
+      "step": 23144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5191715824570404e-05,
+      "loss": 0.4354,
+      "step": 23145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5191074634521676e-05,
+      "loss": 0.5251,
+      "step": 23146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519043344447294e-05,
+      "loss": 0.4542,
+      "step": 23147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518979225442421e-05,
+      "loss": 0.5514,
+      "step": 23148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518915106437548e-05,
+      "loss": 0.5093,
+      "step": 23149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5188509874326756e-05,
+      "loss": 0.6223,
+      "step": 23150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518786868427802e-05,
+      "loss": 0.4717,
+      "step": 23151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518722749422929e-05,
+      "loss": 0.5215,
+      "step": 23152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5186586304180564e-05,
+      "loss": 0.4851,
+      "step": 23153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518594511413183e-05,
+      "loss": 0.3916,
+      "step": 23154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51853039240831e-05,
+      "loss": 0.5252,
+      "step": 23155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5184662734034366e-05,
+      "loss": 0.3829,
+      "step": 23156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518402154398564e-05,
+      "loss": 0.5286,
+      "step": 23157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518338035393691e-05,
+      "loss": 0.4144,
+      "step": 23158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518273916388818e-05,
+      "loss": 0.5162,
+      "step": 23159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5182097973839446e-05,
+      "loss": 0.4888,
+      "step": 23160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518145678379072e-05,
+      "loss": 0.5326,
+      "step": 23161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518081559374199e-05,
+      "loss": 0.501,
+      "step": 23162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5180174403693254e-05,
+      "loss": 0.4656,
+      "step": 23163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5179533213644526e-05,
+      "loss": 0.4687,
+      "step": 23164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517889202359579e-05,
+      "loss": 0.4698,
+      "step": 23165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517825083354706e-05,
+      "loss": 0.5151,
+      "step": 23166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5177609643498335e-05,
+      "loss": 0.5718,
+      "step": 23167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5176968453449606e-05,
+      "loss": 0.5273,
+      "step": 23168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517632726340087e-05,
+      "loss": 0.5299,
+      "step": 23169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517568607335214e-05,
+      "loss": 0.4184,
+      "step": 23170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5175044883303415e-05,
+      "loss": 0.5136,
+      "step": 23171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517440369325468e-05,
+      "loss": 0.4398,
+      "step": 23172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517376250320595e-05,
+      "loss": 0.4204,
+      "step": 23173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517312131315722e-05,
+      "loss": 0.4792,
+      "step": 23174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5172480123108495e-05,
+      "loss": 0.3965,
+      "step": 23175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517183893305976e-05,
+      "loss": 0.4091,
+      "step": 23176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517119774301103e-05,
+      "loss": 0.463,
+      "step": 23177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5170556552962296e-05,
+      "loss": 0.5629,
+      "step": 23178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516991536291357e-05,
+      "loss": 0.4531,
+      "step": 23179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516927417286484e-05,
+      "loss": 0.3626,
+      "step": 23180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5168632982816105e-05,
+      "loss": 0.4919,
+      "step": 23181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5167991792767376e-05,
+      "loss": 0.4476,
+      "step": 23182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516735060271865e-05,
+      "loss": 0.5278,
+      "step": 23183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516670941266992e-05,
+      "loss": 0.4982,
+      "step": 23184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5166068222621185e-05,
+      "loss": 0.4313,
+      "step": 23185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5165427032572456e-05,
+      "loss": 0.4321,
+      "step": 23186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516478584252372e-05,
+      "loss": 0.466,
+      "step": 23187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516414465247499e-05,
+      "loss": 0.479,
+      "step": 23188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5163503462426265e-05,
+      "loss": 0.5728,
+      "step": 23189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5162862272377537e-05,
+      "loss": 0.6758,
+      "step": 23190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516222108232881e-05,
+      "loss": 0.5328,
+      "step": 23191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516157989228007e-05,
+      "loss": 0.4289,
+      "step": 23192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5160938702231345e-05,
+      "loss": 0.5732,
+      "step": 23193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516029751218261e-05,
+      "loss": 0.5387,
+      "step": 23194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515965632213388e-05,
+      "loss": 0.4929,
+      "step": 23195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5159015132085147e-05,
+      "loss": 0.469,
+      "step": 23196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515837394203642e-05,
+      "loss": 0.47,
+      "step": 23197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515773275198769e-05,
+      "loss": 0.425,
+      "step": 23198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515709156193896e-05,
+      "loss": 0.468,
+      "step": 23199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5156450371890233e-05,
+      "loss": 0.4946,
+      "step": 23200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51558091818415e-05,
+      "loss": 0.4842,
+      "step": 23201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515516799179277e-05,
+      "loss": 0.5246,
+      "step": 23202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5154526801744035e-05,
+      "loss": 0.3943,
+      "step": 23203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515388561169531e-05,
+      "loss": 0.5002,
+      "step": 23204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515324442164658e-05,
+      "loss": 0.3873,
+      "step": 23205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515260323159785e-05,
+      "loss": 0.4608,
+      "step": 23206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515196204154912e-05,
+      "loss": 0.4494,
+      "step": 23207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515132085150039e-05,
+      "loss": 0.4257,
+      "step": 23208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515067966145166e-05,
+      "loss": 0.558,
+      "step": 23209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5150038471402923e-05,
+      "loss": 0.3853,
+      "step": 23210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5149397281354195e-05,
+      "loss": 0.6376,
+      "step": 23211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514875609130546e-05,
+      "loss": 0.4276,
+      "step": 23212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514811490125673e-05,
+      "loss": 0.4066,
+      "step": 23213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5147473711208004e-05,
+      "loss": 0.5052,
+      "step": 23214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5146832521159275e-05,
+      "loss": 0.5998,
+      "step": 23215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514619133111055e-05,
+      "loss": 0.4125,
+      "step": 23216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514555014106181e-05,
+      "loss": 0.5148,
+      "step": 23217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5144908951013084e-05,
+      "loss": 0.5265,
+      "step": 23218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514426776096435e-05,
+      "loss": 0.467,
+      "step": 23219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514362657091562e-05,
+      "loss": 0.4738,
+      "step": 23220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5142985380866885e-05,
+      "loss": 0.4442,
+      "step": 23221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5142344190818164e-05,
+      "loss": 0.3991,
+      "step": 23222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514170300076943e-05,
+      "loss": 0.573,
+      "step": 23223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51410618107207e-05,
+      "loss": 0.4646,
+      "step": 23224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.514042062067197e-05,
+      "loss": 0.4205,
+      "step": 23225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513977943062324e-05,
+      "loss": 0.5885,
+      "step": 23226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513913824057451e-05,
+      "loss": 0.4824,
+      "step": 23227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5138497050525774e-05,
+      "loss": 0.5632,
+      "step": 23228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5137855860477045e-05,
+      "loss": 0.51,
+      "step": 23229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513721467042832e-05,
+      "loss": 0.5281,
+      "step": 23230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513657348037959e-05,
+      "loss": 0.596,
+      "step": 23231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5135932290330854e-05,
+      "loss": 0.3845,
+      "step": 23232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5135291100282126e-05,
+      "loss": 0.4138,
+      "step": 23233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51346499102334e-05,
+      "loss": 0.5394,
+      "step": 23234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513400872018466e-05,
+      "loss": 0.6017,
+      "step": 23235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5133367530135934e-05,
+      "loss": 0.4483,
+      "step": 23236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51327263400872e-05,
+      "loss": 0.5395,
+      "step": 23237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513208515003848e-05,
+      "loss": 0.3532,
+      "step": 23238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513144395998974e-05,
+      "loss": 0.4019,
+      "step": 23239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5130802769941014e-05,
+      "loss": 0.5968,
+      "step": 23240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5130161579892286e-05,
+      "loss": 0.6697,
+      "step": 23241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512952038984355e-05,
+      "loss": 0.3966,
+      "step": 23242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512887919979482e-05,
+      "loss": 0.4252,
+      "step": 23243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512823800974609e-05,
+      "loss": 0.4678,
+      "step": 23244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512759681969736e-05,
+      "loss": 0.5369,
+      "step": 23245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512695562964863e-05,
+      "loss": 0.4178,
+      "step": 23246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51263144395999e-05,
+      "loss": 0.5251,
+      "step": 23247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512567324955117e-05,
+      "loss": 0.4982,
+      "step": 23248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512503205950244e-05,
+      "loss": 0.5465,
+      "step": 23249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512439086945371e-05,
+      "loss": 0.4539,
+      "step": 23250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5123749679404976e-05,
+      "loss": 0.3857,
+      "step": 23251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512310848935625e-05,
+      "loss": 0.5323,
+      "step": 23252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512246729930751e-05,
+      "loss": 0.4605,
+      "step": 23253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5121826109258784e-05,
+      "loss": 0.4663,
+      "step": 23254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5121184919210056e-05,
+      "loss": 0.4543,
+      "step": 23255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512054372916133e-05,
+      "loss": 0.4762,
+      "step": 23256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511990253911259e-05,
+      "loss": 0.5037,
+      "step": 23257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5119261349063864e-05,
+      "loss": 0.4269,
+      "step": 23258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5118620159015136e-05,
+      "loss": 0.4301,
+      "step": 23259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51179789689664e-05,
+      "loss": 0.4523,
+      "step": 23260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511733777891767e-05,
+      "loss": 0.493,
+      "step": 23261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5116696588868944e-05,
+      "loss": 0.5217,
+      "step": 23262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5116055398820216e-05,
+      "loss": 0.425,
+      "step": 23263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511541420877148e-05,
+      "loss": 0.4355,
+      "step": 23264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511477301872275e-05,
+      "loss": 0.7961,
+      "step": 23265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511413182867402e-05,
+      "loss": 0.4994,
+      "step": 23266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511349063862529e-05,
+      "loss": 0.5416,
+      "step": 23267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511284944857656e-05,
+      "loss": 0.3718,
+      "step": 23268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5112208258527826e-05,
+      "loss": 0.5269,
+      "step": 23269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51115670684791e-05,
+      "loss": 0.4242,
+      "step": 23270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511092587843037e-05,
+      "loss": 0.4764,
+      "step": 23271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511028468838164e-05,
+      "loss": 0.5845,
+      "step": 23272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5109643498332906e-05,
+      "loss": 0.5361,
+      "step": 23273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.510900230828418e-05,
+      "loss": 0.4754,
+      "step": 23274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.510836111823544e-05,
+      "loss": 0.5312,
+      "step": 23275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5107719928186714e-05,
+      "loss": 0.476,
+      "step": 23276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5107078738137986e-05,
+      "loss": 0.5362,
+      "step": 23277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.510643754808926e-05,
+      "loss": 0.455,
+      "step": 23278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510579635804053e-05,
+      "loss": 0.4785,
+      "step": 23279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5105155167991795e-05,
+      "loss": 0.5668,
+      "step": 23280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5104513977943066e-05,
+      "loss": 0.4041,
+      "step": 23281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510387278789433e-05,
+      "loss": 0.474,
+      "step": 23282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.51032315978456e-05,
+      "loss": 0.4937,
+      "step": 23283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510259040779687e-05,
+      "loss": 0.4487,
+      "step": 23284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510194921774814e-05,
+      "loss": 0.4824,
+      "step": 23285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510130802769941e-05,
+      "loss": 0.5437,
+      "step": 23286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510066683765068e-05,
+      "loss": 0.5665,
+      "step": 23287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5100025647601955e-05,
+      "loss": 0.3989,
+      "step": 23288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509938445755322e-05,
+      "loss": 0.3758,
+      "step": 23289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509874326750449e-05,
+      "loss": 0.56,
+      "step": 23290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5098102077455756e-05,
+      "loss": 0.3754,
+      "step": 23291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509746088740703e-05,
+      "loss": 0.4721,
+      "step": 23292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509681969735829e-05,
+      "loss": 0.4642,
+      "step": 23293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509617850730957e-05,
+      "loss": 0.468,
+      "step": 23294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509553731726084e-05,
+      "loss": 0.5783,
+      "step": 23295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509489612721211e-05,
+      "loss": 0.57,
+      "step": 23296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509425493716338e-05,
+      "loss": 0.5107,
+      "step": 23297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5093613747114645e-05,
+      "loss": 0.4848,
+      "step": 23298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5092972557065917e-05,
+      "loss": 0.5167,
+      "step": 23299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509233136701718e-05,
+      "loss": 0.5889,
+      "step": 23300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509169017696845e-05,
+      "loss": 0.4483,
+      "step": 23301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5091048986919725e-05,
+      "loss": 0.4961,
+      "step": 23302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5090407796870997e-05,
+      "loss": 0.5486,
+      "step": 23303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508976660682227e-05,
+      "loss": 0.5308,
+      "step": 23304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508912541677353e-05,
+      "loss": 0.4329,
+      "step": 23305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5088484226724805e-05,
+      "loss": 0.5274,
+      "step": 23306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508784303667607e-05,
+      "loss": 0.5597,
+      "step": 23307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508720184662734e-05,
+      "loss": 0.4735,
+      "step": 23308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5086560656578607e-05,
+      "loss": 0.5342,
+      "step": 23309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5085919466529885e-05,
+      "loss": 0.4919,
+      "step": 23310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508527827648115e-05,
+      "loss": 0.4862,
+      "step": 23311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508463708643242e-05,
+      "loss": 0.5251,
+      "step": 23312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5083995896383693e-05,
+      "loss": 0.3805,
+      "step": 23313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508335470633496e-05,
+      "loss": 0.479,
+      "step": 23314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508271351628623e-05,
+      "loss": 0.5134,
+      "step": 23315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5082072326237495e-05,
+      "loss": 0.4947,
+      "step": 23316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508143113618877e-05,
+      "loss": 0.5461,
+      "step": 23317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508078994614004e-05,
+      "loss": 0.4311,
+      "step": 23318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508014875609131e-05,
+      "loss": 0.4936,
+      "step": 23319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5079507566042575e-05,
+      "loss": 0.4783,
+      "step": 23320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507886637599385e-05,
+      "loss": 0.4523,
+      "step": 23321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507822518594512e-05,
+      "loss": 0.5281,
+      "step": 23322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5077583995896383e-05,
+      "loss": 0.4412,
+      "step": 23323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5076942805847655e-05,
+      "loss": 0.4021,
+      "step": 23324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507630161579892e-05,
+      "loss": 0.593,
+      "step": 23325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50756604257502e-05,
+      "loss": 0.5563,
+      "step": 23326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5075019235701464e-05,
+      "loss": 0.5501,
+      "step": 23327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5074378045652735e-05,
+      "loss": 0.4617,
+      "step": 23328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5073736855604e-05,
+      "loss": 0.4551,
+      "step": 23329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507309566555527e-05,
+      "loss": 0.5727,
+      "step": 23330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5072454475506544e-05,
+      "loss": 0.4643,
+      "step": 23331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507181328545781e-05,
+      "loss": 0.4385,
+      "step": 23332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507117209540908e-05,
+      "loss": 0.5703,
+      "step": 23333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507053090536035e-05,
+      "loss": 0.5351,
+      "step": 23334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5069889715311624e-05,
+      "loss": 0.4407,
+      "step": 23335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506924852526289e-05,
+      "loss": 0.6224,
+      "step": 23336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506860733521416e-05,
+      "loss": 0.5991,
+      "step": 23337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5067966145165425e-05,
+      "loss": 0.5023,
+      "step": 23338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50673249551167e-05,
+      "loss": 0.3718,
+      "step": 23339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506668376506797e-05,
+      "loss": 0.5376,
+      "step": 23340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5066042575019234e-05,
+      "loss": 0.4966,
+      "step": 23341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5065401384970505e-05,
+      "loss": 0.4249,
+      "step": 23342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506476019492178e-05,
+      "loss": 0.5692,
+      "step": 23343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506411900487305e-05,
+      "loss": 0.5365,
+      "step": 23344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5063477814824314e-05,
+      "loss": 0.5818,
+      "step": 23345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5062836624775586e-05,
+      "loss": 0.5543,
+      "step": 23346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506219543472685e-05,
+      "loss": 0.5178,
+      "step": 23347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506155424467812e-05,
+      "loss": 0.4669,
+      "step": 23348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5060913054629394e-05,
+      "loss": 0.4327,
+      "step": 23349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5060271864580666e-05,
+      "loss": 0.4683,
+      "step": 23350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505963067453194e-05,
+      "loss": 0.5813,
+      "step": 23351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50589894844832e-05,
+      "loss": 0.4702,
+      "step": 23352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5058348294434474e-05,
+      "loss": 0.5559,
+      "step": 23353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505770710438574e-05,
+      "loss": 0.4771,
+      "step": 23354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505706591433701e-05,
+      "loss": 0.4098,
+      "step": 23355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5056424724288276e-05,
+      "loss": 0.5709,
+      "step": 23356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505578353423955e-05,
+      "loss": 0.6598,
+      "step": 23357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505514234419082e-05,
+      "loss": 0.5744,
+      "step": 23358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505450115414209e-05,
+      "loss": 0.4889,
+      "step": 23359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505385996409336e-05,
+      "loss": 0.4759,
+      "step": 23360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505321877404463e-05,
+      "loss": 0.5832,
+      "step": 23361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50525775839959e-05,
+      "loss": 0.4362,
+      "step": 23362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5051936393947164e-05,
+      "loss": 0.4451,
+      "step": 23363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5051295203898436e-05,
+      "loss": 0.4914,
+      "step": 23364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505065401384971e-05,
+      "loss": 0.4824,
+      "step": 23365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505001282380098e-05,
+      "loss": 0.493,
+      "step": 23366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504937163375225e-05,
+      "loss": 0.4735,
+      "step": 23367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5048730443703516e-05,
+      "loss": 0.643,
+      "step": 23368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504808925365479e-05,
+      "loss": 0.5601,
+      "step": 23369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504744806360605e-05,
+      "loss": 0.5271,
+      "step": 23370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5046806873557324e-05,
+      "loss": 0.5435,
+      "step": 23371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504616568350859e-05,
+      "loss": 0.6014,
+      "step": 23372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504552449345986e-05,
+      "loss": 0.4348,
+      "step": 23373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504488330341113e-05,
+      "loss": 0.5006,
+      "step": 23374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5044242113362404e-05,
+      "loss": 0.5259,
+      "step": 23375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5043600923313676e-05,
+      "loss": 0.4946,
+      "step": 23376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504295973326494e-05,
+      "loss": 0.5873,
+      "step": 23377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504231854321621e-05,
+      "loss": 0.3948,
+      "step": 23378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504167735316748e-05,
+      "loss": 0.4895,
+      "step": 23379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504103616311875e-05,
+      "loss": 0.4493,
+      "step": 23380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5040394973070014e-05,
+      "loss": 0.4257,
+      "step": 23381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503975378302129e-05,
+      "loss": 0.4001,
+      "step": 23382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503911259297256e-05,
+      "loss": 0.4428,
+      "step": 23383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503847140292383e-05,
+      "loss": 0.5374,
+      "step": 23384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50378302128751e-05,
+      "loss": 0.4114,
+      "step": 23385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5037189022826366e-05,
+      "loss": 0.5881,
+      "step": 23386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503654783277764e-05,
+      "loss": 0.5245,
+      "step": 23387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50359066427289e-05,
+      "loss": 0.4524,
+      "step": 23388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5035265452680174e-05,
+      "loss": 0.4426,
+      "step": 23389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5034624262631446e-05,
+      "loss": 0.574,
+      "step": 23390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503398307258272e-05,
+      "loss": 0.5932,
+      "step": 23391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503334188253398e-05,
+      "loss": 0.4299,
+      "step": 23392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5032700692485255e-05,
+      "loss": 0.5485,
+      "step": 23393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5032059502436526e-05,
+      "loss": 0.4519,
+      "step": 23394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503141831238779e-05,
+      "loss": 0.575,
+      "step": 23395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503077712233906e-05,
+      "loss": 0.5003,
+      "step": 23396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503013593229033e-05,
+      "loss": 0.504,
+      "step": 23397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5029494742241606e-05,
+      "loss": 0.4747,
+      "step": 23398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502885355219287e-05,
+      "loss": 0.6083,
+      "step": 23399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502821236214414e-05,
+      "loss": 0.5496,
+      "step": 23400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502757117209541e-05,
+      "loss": 0.5507,
+      "step": 23401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502692998204668e-05,
+      "loss": 0.5284,
+      "step": 23402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502628879199795e-05,
+      "loss": 0.4885,
+      "step": 23403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5025647601949216e-05,
+      "loss": 0.4547,
+      "step": 23404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502500641190049e-05,
+      "loss": 0.4282,
+      "step": 23405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502436522185176e-05,
+      "loss": 0.4203,
+      "step": 23406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502372403180303e-05,
+      "loss": 0.4862,
+      "step": 23407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5023082841754296e-05,
+      "loss": 0.4763,
+      "step": 23408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502244165170557e-05,
+      "loss": 0.5737,
+      "step": 23409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502180046165683e-05,
+      "loss": 0.3697,
+      "step": 23410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5021159271608105e-05,
+      "loss": 0.5724,
+      "step": 23411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5020518081559377e-05,
+      "loss": 0.4099,
+      "step": 23412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501987689151064e-05,
+      "loss": 0.4113,
+      "step": 23413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501923570146191e-05,
+      "loss": 0.4998,
+      "step": 23414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5018594511413185e-05,
+      "loss": 0.402,
+      "step": 23415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501795332136446e-05,
+      "loss": 0.4903,
+      "step": 23416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501731213131572e-05,
+      "loss": 0.5168,
+      "step": 23417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501667094126699e-05,
+      "loss": 0.551,
+      "step": 23418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501602975121826e-05,
+      "loss": 0.3584,
+      "step": 23419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501538856116953e-05,
+      "loss": 0.4564,
+      "step": 23420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50147473711208e-05,
+      "loss": 0.5599,
+      "step": 23421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501410618107207e-05,
+      "loss": 0.3959,
+      "step": 23422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5013464991023345e-05,
+      "loss": 0.4931,
+      "step": 23423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501282380097461e-05,
+      "loss": 0.5861,
+      "step": 23424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501218261092588e-05,
+      "loss": 0.4264,
+      "step": 23425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501154142087715e-05,
+      "loss": 0.4506,
+      "step": 23426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501090023082842e-05,
+      "loss": 0.4382,
+      "step": 23427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501025904077969e-05,
+      "loss": 0.5112,
+      "step": 23428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5009617850730955e-05,
+      "loss": 0.5221,
+      "step": 23429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500897666068223e-05,
+      "loss": 0.4233,
+      "step": 23430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50083354706335e-05,
+      "loss": 0.5901,
+      "step": 23431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500769428058477e-05,
+      "loss": 0.4703,
+      "step": 23432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5007053090536035e-05,
+      "loss": 0.4138,
+      "step": 23433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500641190048731e-05,
+      "loss": 0.4596,
+      "step": 23434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500577071043857e-05,
+      "loss": 0.442,
+      "step": 23435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5005129520389844e-05,
+      "loss": 0.5431,
+      "step": 23436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5004488330341115e-05,
+      "loss": 0.4902,
+      "step": 23437
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500384714029239e-05,
+      "loss": 0.4638,
+      "step": 23438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500320595024366e-05,
+      "loss": 0.5338,
+      "step": 23439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5002564760194924e-05,
+      "loss": 0.5446,
+      "step": 23440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5001923570146195e-05,
+      "loss": 0.4697,
+      "step": 23441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500128238009746e-05,
+      "loss": 0.4974,
+      "step": 23442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500064119004873e-05,
+      "loss": 0.5344,
+      "step": 23443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5e-05,
+      "loss": 0.4287,
+      "step": 23444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499935880995127e-05,
+      "loss": 0.4938,
+      "step": 23445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499871761990254e-05,
+      "loss": 0.4096,
+      "step": 23446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499807642985381e-05,
+      "loss": 0.5736,
+      "step": 23447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4997435239805084e-05,
+      "loss": 0.4979,
+      "step": 23448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499679404975635e-05,
+      "loss": 0.5458,
+      "step": 23449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499615285970762e-05,
+      "loss": 0.4644,
+      "step": 23450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4995511669658885e-05,
+      "loss": 0.4972,
+      "step": 23451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499487047961016e-05,
+      "loss": 0.4655,
+      "step": 23452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499422928956142e-05,
+      "loss": 0.4513,
+      "step": 23453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.49935880995127e-05,
+      "loss": 0.4712,
+      "step": 23454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499294690946397e-05,
+      "loss": 0.3723,
+      "step": 23455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499230571941524e-05,
+      "loss": 0.4952,
+      "step": 23456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499166452936651e-05,
+      "loss": 0.4082,
+      "step": 23457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4991023339317774e-05,
+      "loss": 0.6532,
+      "step": 23458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4990382149269046e-05,
+      "loss": 0.7212,
+      "step": 23459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498974095922031e-05,
+      "loss": 0.4969,
+      "step": 23460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498909976917158e-05,
+      "loss": 0.5267,
+      "step": 23461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4988458579122854e-05,
+      "loss": 0.5424,
+      "step": 23462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4987817389074126e-05,
+      "loss": 0.5624,
+      "step": 23463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.49871761990254e-05,
+      "loss": 0.4603,
+      "step": 23464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498653500897666e-05,
+      "loss": 0.4138,
+      "step": 23465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4985893818927934e-05,
+      "loss": 0.3961,
+      "step": 23466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.49852526288792e-05,
+      "loss": 0.4611,
+      "step": 23467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498461143883047e-05,
+      "loss": 0.5471,
+      "step": 23468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4983970248781736e-05,
+      "loss": 0.5543,
+      "step": 23469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4983329058733014e-05,
+      "loss": 0.4996,
+      "step": 23470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498268786868428e-05,
+      "loss": 0.4282,
+      "step": 23471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498204667863555e-05,
+      "loss": 0.5117,
+      "step": 23472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498140548858682e-05,
+      "loss": 0.4201,
+      "step": 23473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498076429853809e-05,
+      "loss": 0.5903,
+      "step": 23474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498012310848936e-05,
+      "loss": 0.5389,
+      "step": 23475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4979481918440624e-05,
+      "loss": 0.4705,
+      "step": 23476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4978840728391896e-05,
+      "loss": 0.5279,
+      "step": 23477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.497819953834317e-05,
+      "loss": 0.4081,
+      "step": 23478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.497755834829444e-05,
+      "loss": 0.4746,
+      "step": 23479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4976917158245704e-05,
+      "loss": 0.4675,
+      "step": 23480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4976275968196976e-05,
+      "loss": 0.4541,
+      "step": 23481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.497563477814825e-05,
+      "loss": 0.6093,
+      "step": 23482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.497499358809951e-05,
+      "loss": 0.4917,
+      "step": 23483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4974352398050784e-05,
+      "loss": 0.4744,
+      "step": 23484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.497371120800205e-05,
+      "loss": 0.537,
+      "step": 23485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.497307001795333e-05,
+      "loss": 0.4491,
+      "step": 23486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.497242882790459e-05,
+      "loss": 0.4624,
+      "step": 23487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4971787637855864e-05,
+      "loss": 0.4936,
+      "step": 23488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.497114644780713e-05,
+      "loss": 0.5432,
+      "step": 23489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.49705052577584e-05,
+      "loss": 0.4924,
+      "step": 23490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496986406770967e-05,
+      "loss": 0.4713,
+      "step": 23491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496922287766094e-05,
+      "loss": 0.4269,
+      "step": 23492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496858168761221e-05,
+      "loss": 0.4692,
+      "step": 23493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496794049756348e-05,
+      "loss": 0.6302,
+      "step": 23494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496729930751475e-05,
+      "loss": 0.4352,
+      "step": 23495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496665811746602e-05,
+      "loss": 0.5113,
+      "step": 23496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496601692741729e-05,
+      "loss": 0.4704,
+      "step": 23497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4965375737368554e-05,
+      "loss": 0.4796,
+      "step": 23498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4964734547319826e-05,
+      "loss": 0.7008,
+      "step": 23499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.49640933572711e-05,
+      "loss": 0.3277,
+      "step": 23500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496345216722236e-05,
+      "loss": 0.4781,
+      "step": 23501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4962810977173635e-05,
+      "loss": 0.3996,
+      "step": 23502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4962169787124906e-05,
+      "loss": 0.4302,
+      "step": 23503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496152859707618e-05,
+      "loss": 0.4717,
+      "step": 23504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496088740702744e-05,
+      "loss": 0.4739,
+      "step": 23505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4960246216978715e-05,
+      "loss": 0.5028,
+      "step": 23506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495960502692998e-05,
+      "loss": 0.4363,
+      "step": 23507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495896383688125e-05,
+      "loss": 0.4701,
+      "step": 23508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495832264683252e-05,
+      "loss": 0.3468,
+      "step": 23509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4957681456783795e-05,
+      "loss": 0.6928,
+      "step": 23510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4957040266735066e-05,
+      "loss": 0.4758,
+      "step": 23511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495639907668633e-05,
+      "loss": 0.4649,
+      "step": 23512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.49557578866376e-05,
+      "loss": 0.3651,
+      "step": 23513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495511669658887e-05,
+      "loss": 0.411,
+      "step": 23514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495447550654014e-05,
+      "loss": 0.5584,
+      "step": 23515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4953834316491405e-05,
+      "loss": 0.4304,
+      "step": 23516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4953193126442676e-05,
+      "loss": 0.508,
+      "step": 23517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495255193639395e-05,
+      "loss": 0.4611,
+      "step": 23518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495191074634522e-05,
+      "loss": 0.4415,
+      "step": 23519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495126955629649e-05,
+      "loss": 0.5666,
+      "step": 23520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4950628366247756e-05,
+      "loss": 0.3722,
+      "step": 23521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494998717619903e-05,
+      "loss": 0.3986,
+      "step": 23522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494934598615029e-05,
+      "loss": 0.4512,
+      "step": 23523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4948704796101565e-05,
+      "loss": 0.4595,
+      "step": 23524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4948063606052837e-05,
+      "loss": 0.5215,
+      "step": 23525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494742241600411e-05,
+      "loss": 0.4919,
+      "step": 23526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494678122595538e-05,
+      "loss": 0.4162,
+      "step": 23527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4946140035906645e-05,
+      "loss": 0.3911,
+      "step": 23528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494549884585792e-05,
+      "loss": 0.397,
+      "step": 23529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494485765580918e-05,
+      "loss": 0.5098,
+      "step": 23530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494421646576045e-05,
+      "loss": 0.4866,
+      "step": 23531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494357527571172e-05,
+      "loss": 0.4519,
+      "step": 23532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494293408566299e-05,
+      "loss": 0.4282,
+      "step": 23533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494229289561426e-05,
+      "loss": 0.5456,
+      "step": 23534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4941651705565533e-05,
+      "loss": 0.5665,
+      "step": 23535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4941010515516805e-05,
+      "loss": 0.401,
+      "step": 23536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.494036932546807e-05,
+      "loss": 0.4332,
+      "step": 23537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.493972813541934e-05,
+      "loss": 0.4893,
+      "step": 23538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.493908694537061e-05,
+      "loss": 0.4073,
+      "step": 23539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493844575532188e-05,
+      "loss": 0.3557,
+      "step": 23540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493780456527314e-05,
+      "loss": 0.4351,
+      "step": 23541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493716337522442e-05,
+      "loss": 0.5316,
+      "step": 23542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493652218517569e-05,
+      "loss": 0.3706,
+      "step": 23543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493588099512696e-05,
+      "loss": 0.4643,
+      "step": 23544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493523980507823e-05,
+      "loss": 0.4237,
+      "step": 23545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4934598615029495e-05,
+      "loss": 0.4561,
+      "step": 23546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493395742498077e-05,
+      "loss": 0.4762,
+      "step": 23547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493331623493203e-05,
+      "loss": 0.5029,
+      "step": 23548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4932675044883304e-05,
+      "loss": 0.5673,
+      "step": 23549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4932033854834575e-05,
+      "loss": 0.6615,
+      "step": 23550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493139266478585e-05,
+      "loss": 0.4409,
+      "step": 23551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493075147473711e-05,
+      "loss": 0.4492,
+      "step": 23552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4930110284688384e-05,
+      "loss": 0.5499,
+      "step": 23553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4929469094639655e-05,
+      "loss": 0.4339,
+      "step": 23554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492882790459092e-05,
+      "loss": 0.3939,
+      "step": 23555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492818671454219e-05,
+      "loss": 0.4554,
+      "step": 23556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492754552449346e-05,
+      "loss": 0.4935,
+      "step": 23557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4926904334444735e-05,
+      "loss": 0.5541,
+      "step": 23558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4926263144396e-05,
+      "loss": 0.5769,
+      "step": 23559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492562195434727e-05,
+      "loss": 0.5834,
+      "step": 23560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492498076429854e-05,
+      "loss": 0.3745,
+      "step": 23561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492433957424981e-05,
+      "loss": 0.4691,
+      "step": 23562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492369838420108e-05,
+      "loss": 0.4003,
+      "step": 23563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4923057194152345e-05,
+      "loss": 0.4665,
+      "step": 23564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492241600410362e-05,
+      "loss": 0.4285,
+      "step": 23565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492177481405489e-05,
+      "loss": 0.5558,
+      "step": 23566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492113362400616e-05,
+      "loss": 0.5579,
+      "step": 23567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4920492433957426e-05,
+      "loss": 0.6228,
+      "step": 23568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49198512439087e-05,
+      "loss": 0.4868,
+      "step": 23569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491921005385996e-05,
+      "loss": 0.5527,
+      "step": 23570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4918568863811234e-05,
+      "loss": 0.4136,
+      "step": 23571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4917927673762506e-05,
+      "loss": 0.4545,
+      "step": 23572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491728648371377e-05,
+      "loss": 0.3736,
+      "step": 23573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491664529366505e-05,
+      "loss": 0.4038,
+      "step": 23574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4916004103616314e-05,
+      "loss": 0.4979,
+      "step": 23575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4915362913567586e-05,
+      "loss": 0.4721,
+      "step": 23576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491472172351885e-05,
+      "loss": 0.586,
+      "step": 23577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491408053347012e-05,
+      "loss": 0.4241,
+      "step": 23578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491343934342139e-05,
+      "loss": 0.588,
+      "step": 23579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491279815337266e-05,
+      "loss": 0.4677,
+      "step": 23580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491215696332393e-05,
+      "loss": 0.5569,
+      "step": 23581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49115157732752e-05,
+      "loss": 0.5936,
+      "step": 23582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4910874583226474e-05,
+      "loss": 0.4826,
+      "step": 23583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491023339317774e-05,
+      "loss": 0.3722,
+      "step": 23584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490959220312901e-05,
+      "loss": 0.4396,
+      "step": 23585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4908951013080276e-05,
+      "loss": 0.3664,
+      "step": 23586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490830982303155e-05,
+      "loss": 0.4011,
+      "step": 23587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490766863298281e-05,
+      "loss": 0.5567,
+      "step": 23588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4907027442934084e-05,
+      "loss": 0.402,
+      "step": 23589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4906386252885356e-05,
+      "loss": 0.572,
+      "step": 23590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490574506283663e-05,
+      "loss": 0.4583,
+      "step": 23591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49051038727879e-05,
+      "loss": 0.406,
+      "step": 23592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4904462682739164e-05,
+      "loss": 0.3934,
+      "step": 23593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4903821492690436e-05,
+      "loss": 0.4331,
+      "step": 23594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49031803026417e-05,
+      "loss": 0.4907,
+      "step": 23595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490253911259297e-05,
+      "loss": 0.5481,
+      "step": 23596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4901897922544244e-05,
+      "loss": 0.5327,
+      "step": 23597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4901256732495516e-05,
+      "loss": 0.5507,
+      "step": 23598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490061554244679e-05,
+      "loss": 0.4859,
+      "step": 23599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489997435239805e-05,
+      "loss": 0.4938,
+      "step": 23600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4899333162349324e-05,
+      "loss": 0.3962,
+      "step": 23601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489869197230059e-05,
+      "loss": 0.5831,
+      "step": 23602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489805078225186e-05,
+      "loss": 0.4118,
+      "step": 23603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4897409592203126e-05,
+      "loss": 0.5242,
+      "step": 23604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48967684021544e-05,
+      "loss": 0.4999,
+      "step": 23605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489612721210567e-05,
+      "loss": 0.4575,
+      "step": 23606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489548602205694e-05,
+      "loss": 0.3831,
+      "step": 23607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489484483200821e-05,
+      "loss": 0.4978,
+      "step": 23608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489420364195948e-05,
+      "loss": 0.4762,
+      "step": 23609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489356245191075e-05,
+      "loss": 0.464,
+      "step": 23610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4892921261862014e-05,
+      "loss": 0.4018,
+      "step": 23611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4892280071813286e-05,
+      "loss": 0.6131,
+      "step": 23612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489163888176456e-05,
+      "loss": 0.386,
+      "step": 23613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489099769171583e-05,
+      "loss": 0.5549,
+      "step": 23614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48903565016671e-05,
+      "loss": 0.3631,
+      "step": 23615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4889715311618366e-05,
+      "loss": 0.4688,
+      "step": 23616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488907412156964e-05,
+      "loss": 0.5206,
+      "step": 23617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48884329315209e-05,
+      "loss": 0.5737,
+      "step": 23618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4887791741472175e-05,
+      "loss": 0.5895,
+      "step": 23619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488715055142344e-05,
+      "loss": 0.513,
+      "step": 23620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488650936137471e-05,
+      "loss": 0.4537,
+      "step": 23621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488586817132598e-05,
+      "loss": 0.561,
+      "step": 23622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4885226981277255e-05,
+      "loss": 0.5301,
+      "step": 23623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4884585791228526e-05,
+      "loss": 0.4215,
+      "step": 23624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488394460117979e-05,
+      "loss": 0.6145,
+      "step": 23625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488330341113106e-05,
+      "loss": 0.5444,
+      "step": 23626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488266222108233e-05,
+      "loss": 0.454,
+      "step": 23627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48820210310336e-05,
+      "loss": 0.4123,
+      "step": 23628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4881379840984865e-05,
+      "loss": 0.5322,
+      "step": 23629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488073865093614e-05,
+      "loss": 0.4403,
+      "step": 23630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488009746088741e-05,
+      "loss": 0.5329,
+      "step": 23631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487945627083868e-05,
+      "loss": 0.4664,
+      "step": 23632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487881508078995e-05,
+      "loss": 0.4618,
+      "step": 23633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4878173890741217e-05,
+      "loss": 0.5086,
+      "step": 23634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487753270069249e-05,
+      "loss": 0.5919,
+      "step": 23635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487689151064375e-05,
+      "loss": 0.4742,
+      "step": 23636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4876250320595025e-05,
+      "loss": 0.4668,
+      "step": 23637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4875609130546297e-05,
+      "loss": 0.496,
+      "step": 23638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487496794049757e-05,
+      "loss": 0.5594,
+      "step": 23639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487432675044883e-05,
+      "loss": 0.5451,
+      "step": 23640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4873685560400105e-05,
+      "loss": 0.3655,
+      "step": 23641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487304437035138e-05,
+      "loss": 0.5391,
+      "step": 23642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487240318030264e-05,
+      "loss": 0.5424,
+      "step": 23643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487176199025391e-05,
+      "loss": 0.4768,
+      "step": 23644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487112080020518e-05,
+      "loss": 0.4704,
+      "step": 23645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487047961015646e-05,
+      "loss": 0.5468,
+      "step": 23646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486983842010772e-05,
+      "loss": 0.5597,
+      "step": 23647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4869197230058993e-05,
+      "loss": 0.3418,
+      "step": 23648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486855604001026e-05,
+      "loss": 0.484,
+      "step": 23649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486791484996153e-05,
+      "loss": 0.4823,
+      "step": 23650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48672736599128e-05,
+      "loss": 0.4571,
+      "step": 23651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486663246986407e-05,
+      "loss": 0.5113,
+      "step": 23652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486599127981534e-05,
+      "loss": 0.4368,
+      "step": 23653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486535008976661e-05,
+      "loss": 0.5854,
+      "step": 23654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486470889971788e-05,
+      "loss": 0.4678,
+      "step": 23655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486406770966915e-05,
+      "loss": 0.6399,
+      "step": 23656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486342651962042e-05,
+      "loss": 0.4676,
+      "step": 23657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4862785329571684e-05,
+      "loss": 0.3705,
+      "step": 23658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4862144139522955e-05,
+      "loss": 0.4284,
+      "step": 23659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486150294947423e-05,
+      "loss": 0.5436,
+      "step": 23660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486086175942549e-05,
+      "loss": 0.554,
+      "step": 23661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4860220569376764e-05,
+      "loss": 0.5546,
+      "step": 23662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4859579379328035e-05,
+      "loss": 0.6115,
+      "step": 23663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485893818927931e-05,
+      "loss": 0.4629,
+      "step": 23664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485829699923057e-05,
+      "loss": 0.4754,
+      "step": 23665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4857655809181844e-05,
+      "loss": 0.4563,
+      "step": 23666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485701461913311e-05,
+      "loss": 0.4012,
+      "step": 23667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485637342908438e-05,
+      "loss": 0.483,
+      "step": 23668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485573223903565e-05,
+      "loss": 0.4717,
+      "step": 23669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4855091048986924e-05,
+      "loss": 0.4106,
+      "step": 23670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4854449858938195e-05,
+      "loss": 0.5371,
+      "step": 23671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485380866888946e-05,
+      "loss": 0.4435,
+      "step": 23672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485316747884073e-05,
+      "loss": 0.4215,
+      "step": 23673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4852526288792e-05,
+      "loss": 0.3606,
+      "step": 23674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485188509874327e-05,
+      "loss": 0.5139,
+      "step": 23675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4851243908694534e-05,
+      "loss": 0.5208,
+      "step": 23676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4850602718645805e-05,
+      "loss": 0.4653,
+      "step": 23677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484996152859708e-05,
+      "loss": 0.4983,
+      "step": 23678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484932033854835e-05,
+      "loss": 0.3784,
+      "step": 23679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484867914849962e-05,
+      "loss": 0.5142,
+      "step": 23680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4848037958450886e-05,
+      "loss": 0.5726,
+      "step": 23681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484739676840216e-05,
+      "loss": 0.4747,
+      "step": 23682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484675557835342e-05,
+      "loss": 0.5357,
+      "step": 23683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4846114388304694e-05,
+      "loss": 0.4589,
+      "step": 23684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4845473198255966e-05,
+      "loss": 0.4397,
+      "step": 23685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484483200820724e-05,
+      "loss": 0.4937,
+      "step": 23686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484419081815851e-05,
+      "loss": 0.4542,
+      "step": 23687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4843549628109774e-05,
+      "loss": 0.5581,
+      "step": 23688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4842908438061046e-05,
+      "loss": 0.5192,
+      "step": 23689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484226724801231e-05,
+      "loss": 0.544,
+      "step": 23690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484162605796358e-05,
+      "loss": 0.6155,
+      "step": 23691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484098486791485e-05,
+      "loss": 0.5746,
+      "step": 23692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484034367786612e-05,
+      "loss": 0.4899,
+      "step": 23693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483970248781739e-05,
+      "loss": 0.4734,
+      "step": 23694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483906129776866e-05,
+      "loss": 0.4971,
+      "step": 23695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4838420107719934e-05,
+      "loss": 0.5322,
+      "step": 23696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48377789176712e-05,
+      "loss": 0.594,
+      "step": 23697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483713772762247e-05,
+      "loss": 0.4636,
+      "step": 23698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4836496537573736e-05,
+      "loss": 0.4676,
+      "step": 23699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483585534752501e-05,
+      "loss": 0.4885,
+      "step": 23700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483521415747627e-05,
+      "loss": 0.4338,
+      "step": 23701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483457296742755e-05,
+      "loss": 0.5592,
+      "step": 23702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4833931777378816e-05,
+      "loss": 0.5186,
+      "step": 23703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483329058733009e-05,
+      "loss": 0.4506,
+      "step": 23704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483264939728136e-05,
+      "loss": 0.4974,
+      "step": 23705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4832008207232624e-05,
+      "loss": 0.3994,
+      "step": 23706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4831367017183896e-05,
+      "loss": 0.5113,
+      "step": 23707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483072582713516e-05,
+      "loss": 0.5106,
+      "step": 23708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483008463708643e-05,
+      "loss": 0.4829,
+      "step": 23709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4829443447037704e-05,
+      "loss": 0.5215,
+      "step": 23710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4828802256988976e-05,
+      "loss": 0.5503,
+      "step": 23711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482816106694024e-05,
+      "loss": 0.4475,
+      "step": 23712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482751987689151e-05,
+      "loss": 0.3729,
+      "step": 23713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4826878686842784e-05,
+      "loss": 0.4339,
+      "step": 23714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482623749679405e-05,
+      "loss": 0.51,
+      "step": 23715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482559630674532e-05,
+      "loss": 0.4993,
+      "step": 23716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4824955116696586e-05,
+      "loss": 0.5487,
+      "step": 23717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4824313926647865e-05,
+      "loss": 0.4264,
+      "step": 23718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482367273659913e-05,
+      "loss": 0.5187,
+      "step": 23719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48230315465504e-05,
+      "loss": 0.4689,
+      "step": 23720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4822390356501666e-05,
+      "loss": 0.5635,
+      "step": 23721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482174916645294e-05,
+      "loss": 0.3609,
+      "step": 23722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482110797640421e-05,
+      "loss": 0.4947,
+      "step": 23723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4820466786355475e-05,
+      "loss": 0.5958,
+      "step": 23724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4819825596306746e-05,
+      "loss": 0.3832,
+      "step": 23725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481918440625802e-05,
+      "loss": 0.3911,
+      "step": 23726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481854321620929e-05,
+      "loss": 0.6073,
+      "step": 23727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4817902026160555e-05,
+      "loss": 0.4678,
+      "step": 23728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4817260836111826e-05,
+      "loss": 0.4215,
+      "step": 23729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481661964606309e-05,
+      "loss": 0.5558,
+      "step": 23730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481597845601436e-05,
+      "loss": 0.4164,
+      "step": 23731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4815337265965635e-05,
+      "loss": 0.4803,
+      "step": 23732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48146960759169e-05,
+      "loss": 0.412,
+      "step": 23733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481405488586818e-05,
+      "loss": 0.4555,
+      "step": 23734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481341369581944e-05,
+      "loss": 0.5392,
+      "step": 23735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4812772505770715e-05,
+      "loss": 0.5739,
+      "step": 23736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481213131572198e-05,
+      "loss": 0.5013,
+      "step": 23737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481149012567325e-05,
+      "loss": 0.45,
+      "step": 23738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4810848935624516e-05,
+      "loss": 0.6459,
+      "step": 23739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481020774557579e-05,
+      "loss": 0.572,
+      "step": 23740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480956655552706e-05,
+      "loss": 0.3603,
+      "step": 23741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480892536547833e-05,
+      "loss": 0.4764,
+      "step": 23742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48082841754296e-05,
+      "loss": 0.5423,
+      "step": 23743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480764298538087e-05,
+      "loss": 0.5573,
+      "step": 23744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480700179533214e-05,
+      "loss": 0.4918,
+      "step": 23745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4806360605283405e-05,
+      "loss": 0.5151,
+      "step": 23746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4805719415234677e-05,
+      "loss": 0.4724,
+      "step": 23747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480507822518594e-05,
+      "loss": 0.4651,
+      "step": 23748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480443703513721e-05,
+      "loss": 0.5205,
+      "step": 23749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4803795845088485e-05,
+      "loss": 0.4198,
+      "step": 23750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480315465503976e-05,
+      "loss": 0.5389,
+      "step": 23751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480251346499103e-05,
+      "loss": 0.5991,
+      "step": 23752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480187227494229e-05,
+      "loss": 0.4844,
+      "step": 23753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4801231084893565e-05,
+      "loss": 0.5147,
+      "step": 23754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480058989484483e-05,
+      "loss": 0.4956,
+      "step": 23755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.47999487047961e-05,
+      "loss": 0.4617,
+      "step": 23756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479930751474737e-05,
+      "loss": 0.5429,
+      "step": 23757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4798666324698645e-05,
+      "loss": 0.4627,
+      "step": 23758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479802513464992e-05,
+      "loss": 0.4896,
+      "step": 23759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479738394460118e-05,
+      "loss": 0.5073,
+      "step": 23760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4796742754552453e-05,
+      "loss": 0.5331,
+      "step": 23761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479610156450372e-05,
+      "loss": 0.5457,
+      "step": 23762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479546037445499e-05,
+      "loss": 0.4615,
+      "step": 23763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4794819184406255e-05,
+      "loss": 0.4077,
+      "step": 23764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479417799435753e-05,
+      "loss": 0.5166,
+      "step": 23765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.47935368043088e-05,
+      "loss": 0.4724,
+      "step": 23766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479289561426007e-05,
+      "loss": 0.4164,
+      "step": 23767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479225442421134e-05,
+      "loss": 0.524,
+      "step": 23768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479161323416261e-05,
+      "loss": 0.5134,
+      "step": 23769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.479097204411388e-05,
+      "loss": 0.3637,
+      "step": 23770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4790330854065144e-05,
+      "loss": 0.4569,
+      "step": 23771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4789689664016415e-05,
+      "loss": 0.4935,
+      "step": 23772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478904847396769e-05,
+      "loss": 0.4064,
+      "step": 23773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478840728391896e-05,
+      "loss": 0.4996,
+      "step": 23774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4787766093870224e-05,
+      "loss": 0.6057,
+      "step": 23775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4787124903821495e-05,
+      "loss": 0.4459,
+      "step": 23776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478648371377277e-05,
+      "loss": 0.6724,
+      "step": 23777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478584252372403e-05,
+      "loss": 0.5202,
+      "step": 23778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4785201333675304e-05,
+      "loss": 0.481,
+      "step": 23779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478456014362657e-05,
+      "loss": 0.3822,
+      "step": 23780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478391895357784e-05,
+      "loss": 0.4371,
+      "step": 23781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478327776352911e-05,
+      "loss": 0.5054,
+      "step": 23782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4782636573480384e-05,
+      "loss": 0.5027,
+      "step": 23783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478199538343165e-05,
+      "loss": 0.5398,
+      "step": 23784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478135419338292e-05,
+      "loss": 0.5501,
+      "step": 23785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478071300333419e-05,
+      "loss": 0.5363,
+      "step": 23786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478007181328546e-05,
+      "loss": 0.4118,
+      "step": 23787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.477943062323673e-05,
+      "loss": 0.3233,
+      "step": 23788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4778789433187994e-05,
+      "loss": 0.4395,
+      "step": 23789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.477814824313927e-05,
+      "loss": 0.5264,
+      "step": 23790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.477750705309054e-05,
+      "loss": 0.5187,
+      "step": 23791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.477686586304181e-05,
+      "loss": 0.4363,
+      "step": 23792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4776224672993074e-05,
+      "loss": 0.4741,
+      "step": 23793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4775583482944346e-05,
+      "loss": 0.5307,
+      "step": 23794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.477494229289562e-05,
+      "loss": 0.4913,
+      "step": 23795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.477430110284688e-05,
+      "loss": 0.4418,
+      "step": 23796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4773659912798154e-05,
+      "loss": 0.5231,
+      "step": 23797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4773018722749426e-05,
+      "loss": 0.4235,
+      "step": 23798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.47723775327007e-05,
+      "loss": 0.4713,
+      "step": 23799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477173634265196e-05,
+      "loss": 0.4577,
+      "step": 23800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4771095152603234e-05,
+      "loss": 0.4396,
+      "step": 23801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47704539625545e-05,
+      "loss": 0.4258,
+      "step": 23802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476981277250577e-05,
+      "loss": 0.5691,
+      "step": 23803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476917158245704e-05,
+      "loss": 0.5883,
+      "step": 23804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476853039240831e-05,
+      "loss": 0.4568,
+      "step": 23805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4767889202359586e-05,
+      "loss": 0.5739,
+      "step": 23806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476724801231085e-05,
+      "loss": 0.528,
+      "step": 23807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476660682226212e-05,
+      "loss": 0.4871,
+      "step": 23808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476596563221339e-05,
+      "loss": 0.5642,
+      "step": 23809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476532444216466e-05,
+      "loss": 0.3995,
+      "step": 23810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476468325211593e-05,
+      "loss": 0.5209,
+      "step": 23811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4764042062067196e-05,
+      "loss": 0.482,
+      "step": 23812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476340087201847e-05,
+      "loss": 0.4011,
+      "step": 23813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476275968196974e-05,
+      "loss": 0.4798,
+      "step": 23814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476211849192101e-05,
+      "loss": 0.4656,
+      "step": 23815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4761477301872276e-05,
+      "loss": 0.5015,
+      "step": 23816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476083611182355e-05,
+      "loss": 0.3282,
+      "step": 23817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476019492177481e-05,
+      "loss": 0.5088,
+      "step": 23818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4759553731726084e-05,
+      "loss": 0.6152,
+      "step": 23819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4758912541677356e-05,
+      "loss": 0.4754,
+      "step": 23820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475827135162862e-05,
+      "loss": 0.4807,
+      "step": 23821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475763016157989e-05,
+      "loss": 0.5267,
+      "step": 23822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4756988971531164e-05,
+      "loss": 0.5512,
+      "step": 23823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4756347781482436e-05,
+      "loss": 0.4897,
+      "step": 23824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47557065914337e-05,
+      "loss": 0.4891,
+      "step": 23825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475506540138497e-05,
+      "loss": 0.4631,
+      "step": 23826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475442421133624e-05,
+      "loss": 0.5331,
+      "step": 23827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475378302128751e-05,
+      "loss": 0.5388,
+      "step": 23828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475314183123878e-05,
+      "loss": 0.4137,
+      "step": 23829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475250064119005e-05,
+      "loss": 0.4691,
+      "step": 23830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4751859451141325e-05,
+      "loss": 0.4408,
+      "step": 23831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475121826109259e-05,
+      "loss": 0.5153,
+      "step": 23832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475057707104386e-05,
+      "loss": 0.4005,
+      "step": 23833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4749935880995126e-05,
+      "loss": 0.4653,
+      "step": 23834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47492946909464e-05,
+      "loss": 0.5525,
+      "step": 23835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474865350089766e-05,
+      "loss": 0.5515,
+      "step": 23836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4748012310848935e-05,
+      "loss": 0.4553,
+      "step": 23837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4747371120800206e-05,
+      "loss": 0.4347,
+      "step": 23838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474672993075148e-05,
+      "loss": 0.5799,
+      "step": 23839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474608874070275e-05,
+      "loss": 0.5384,
+      "step": 23840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4745447550654015e-05,
+      "loss": 0.4887,
+      "step": 23841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4744806360605286e-05,
+      "loss": 0.5422,
+      "step": 23842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474416517055655e-05,
+      "loss": 0.5305,
+      "step": 23843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474352398050782e-05,
+      "loss": 0.5668,
+      "step": 23844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4742882790459095e-05,
+      "loss": 0.5335,
+      "step": 23845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4742241600410366e-05,
+      "loss": 0.471,
+      "step": 23846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474160041036164e-05,
+      "loss": 0.4879,
+      "step": 23847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47409592203129e-05,
+      "loss": 0.3806,
+      "step": 23848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4740318030264175e-05,
+      "loss": 0.4432,
+      "step": 23849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473967684021544e-05,
+      "loss": 0.4205,
+      "step": 23850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473903565016671e-05,
+      "loss": 0.5874,
+      "step": 23851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4738394460117976e-05,
+      "loss": 0.3771,
+      "step": 23852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473775327006925e-05,
+      "loss": 0.5561,
+      "step": 23853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473711208002052e-05,
+      "loss": 0.5124,
+      "step": 23854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473647088997179e-05,
+      "loss": 0.6006,
+      "step": 23855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473582969992306e-05,
+      "loss": 0.4477,
+      "step": 23856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473518850987433e-05,
+      "loss": 0.5058,
+      "step": 23857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47345473198256e-05,
+      "loss": 0.5192,
+      "step": 23858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4733906129776865e-05,
+      "loss": 0.3717,
+      "step": 23859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4733264939728137e-05,
+      "loss": 0.499,
+      "step": 23860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473262374967941e-05,
+      "loss": 0.5408,
+      "step": 23861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473198255963068e-05,
+      "loss": 0.4454,
+      "step": 23862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4731341369581945e-05,
+      "loss": 0.4005,
+      "step": 23863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473070017953322e-05,
+      "loss": 0.3457,
+      "step": 23864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473005898948449e-05,
+      "loss": 0.6263,
+      "step": 23865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472941779943575e-05,
+      "loss": 0.3797,
+      "step": 23866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4728776609387025e-05,
+      "loss": 0.411,
+      "step": 23867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472813541933829e-05,
+      "loss": 0.4975,
+      "step": 23868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472749422928956e-05,
+      "loss": 0.5013,
+      "step": 23869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4726853039240833e-05,
+      "loss": 0.3722,
+      "step": 23870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4726211849192105e-05,
+      "loss": 0.4372,
+      "step": 23871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472557065914337e-05,
+      "loss": 0.4282,
+      "step": 23872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472492946909464e-05,
+      "loss": 0.3552,
+      "step": 23873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4724288279045914e-05,
+      "loss": 0.3917,
+      "step": 23874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472364708899718e-05,
+      "loss": 0.5283,
+      "step": 23875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472300589894845e-05,
+      "loss": 0.4544,
+      "step": 23876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4722364708899715e-05,
+      "loss": 0.4074,
+      "step": 23877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4721723518850994e-05,
+      "loss": 0.4174,
+      "step": 23878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472108232880226e-05,
+      "loss": 0.3564,
+      "step": 23879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472044113875353e-05,
+      "loss": 0.5082,
+      "step": 23880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4719799948704795e-05,
+      "loss": 0.4081,
+      "step": 23881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471915875865607e-05,
+      "loss": 0.4119,
+      "step": 23882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471851756860734e-05,
+      "loss": 0.353,
+      "step": 23883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4717876378558604e-05,
+      "loss": 0.5637,
+      "step": 23884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4717235188509875e-05,
+      "loss": 0.561,
+      "step": 23885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471659399846115e-05,
+      "loss": 0.4756,
+      "step": 23886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471595280841242e-05,
+      "loss": 0.5436,
+      "step": 23887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4715311618363684e-05,
+      "loss": 0.3691,
+      "step": 23888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4714670428314955e-05,
+      "loss": 0.5126,
+      "step": 23889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471402923826622e-05,
+      "loss": 0.4644,
+      "step": 23890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471338804821749e-05,
+      "loss": 0.5946,
+      "step": 23891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4712746858168764e-05,
+      "loss": 0.4441,
+      "step": 23892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471210566812003e-05,
+      "loss": 0.4473,
+      "step": 23893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471146447807131e-05,
+      "loss": 0.4042,
+      "step": 23894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471082328802257e-05,
+      "loss": 0.4987,
+      "step": 23895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4710182097973844e-05,
+      "loss": 0.4828,
+      "step": 23896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470954090792511e-05,
+      "loss": 0.6446,
+      "step": 23897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470889971787638e-05,
+      "loss": 0.743,
+      "step": 23898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4708258527827645e-05,
+      "loss": 0.3833,
+      "step": 23899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470761733777892e-05,
+      "loss": 0.5885,
+      "step": 23900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470697614773019e-05,
+      "loss": 0.4605,
+      "step": 23901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470633495768146e-05,
+      "loss": 0.3981,
+      "step": 23902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470569376763273e-05,
+      "loss": 0.4181,
+      "step": 23903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4705052577584e-05,
+      "loss": 0.4747,
+      "step": 23904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470441138753527e-05,
+      "loss": 0.4109,
+      "step": 23905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4703770197486534e-05,
+      "loss": 0.4457,
+      "step": 23906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4703129007437806e-05,
+      "loss": 0.4674,
+      "step": 23907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470248781738907e-05,
+      "loss": 0.4582,
+      "step": 23908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470184662734034e-05,
+      "loss": 0.5108,
+      "step": 23909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4701205437291614e-05,
+      "loss": 0.5409,
+      "step": 23910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4700564247242886e-05,
+      "loss": 0.4142,
+      "step": 23911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469992305719416e-05,
+      "loss": 0.5741,
+      "step": 23912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469928186714542e-05,
+      "loss": 0.534,
+      "step": 23913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4698640677096694e-05,
+      "loss": 0.4795,
+      "step": 23914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469799948704796e-05,
+      "loss": 0.5097,
+      "step": 23915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469735829699923e-05,
+      "loss": 0.609,
+      "step": 23916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46967171069505e-05,
+      "loss": 0.439,
+      "step": 23917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4696075916901774e-05,
+      "loss": 0.3886,
+      "step": 23918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4695434726853046e-05,
+      "loss": 0.443,
+      "step": 23919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469479353680431e-05,
+      "loss": 0.4498,
+      "step": 23920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469415234675558e-05,
+      "loss": 0.488,
+      "step": 23921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469351115670685e-05,
+      "loss": 0.5793,
+      "step": 23922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469286996665812e-05,
+      "loss": 0.5069,
+      "step": 23923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4692228776609384e-05,
+      "loss": 0.4649,
+      "step": 23924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4691587586560656e-05,
+      "loss": 0.4783,
+      "step": 23925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469094639651193e-05,
+      "loss": 0.5528,
+      "step": 23926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46903052064632e-05,
+      "loss": 0.5011,
+      "step": 23927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468966401641447e-05,
+      "loss": 0.5465,
+      "step": 23928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4689022826365736e-05,
+      "loss": 0.4707,
+      "step": 23929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468838163631701e-05,
+      "loss": 0.5052,
+      "step": 23930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468774044626827e-05,
+      "loss": 0.4295,
+      "step": 23931
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4687099256219544e-05,
+      "loss": 0.395,
+      "step": 23932
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4686458066170816e-05,
+      "loss": 0.4952,
+      "step": 23933
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468581687612209e-05,
+      "loss": 0.3994,
+      "step": 23934
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468517568607335e-05,
+      "loss": 0.4574,
+      "step": 23935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4684534496024624e-05,
+      "loss": 0.5548,
+      "step": 23936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4683893305975896e-05,
+      "loss": 0.4974,
+      "step": 23937
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468325211592716e-05,
+      "loss": 0.4961,
+      "step": 23938
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468261092587843e-05,
+      "loss": 0.5076,
+      "step": 23939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46819697358297e-05,
+      "loss": 0.5181,
+      "step": 23940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468132854578097e-05,
+      "loss": 0.4396,
+      "step": 23941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468068735573224e-05,
+      "loss": 0.4173,
+      "step": 23942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468004616568351e-05,
+      "loss": 0.4629,
+      "step": 23943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467940497563478e-05,
+      "loss": 0.575,
+      "step": 23944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467876378558605e-05,
+      "loss": 0.5809,
+      "step": 23945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467812259553732e-05,
+      "loss": 0.5249,
+      "step": 23946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4677481405488586e-05,
+      "loss": 0.4787,
+      "step": 23947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467684021543986e-05,
+      "loss": 0.482,
+      "step": 23948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467619902539112e-05,
+      "loss": 0.4245,
+      "step": 23949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46755578353424e-05,
+      "loss": 0.3951,
+      "step": 23950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4674916645293666e-05,
+      "loss": 0.4543,
+      "step": 23951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467427545524494e-05,
+      "loss": 0.3963,
+      "step": 23952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46736342651962e-05,
+      "loss": 0.4769,
+      "step": 23953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4672993075147475e-05,
+      "loss": 0.5059,
+      "step": 23954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4672351885098746e-05,
+      "loss": 0.6343,
+      "step": 23955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467171069505001e-05,
+      "loss": 0.4895,
+      "step": 23956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467106950500128e-05,
+      "loss": 0.4594,
+      "step": 23957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4670428314952555e-05,
+      "loss": 0.4527,
+      "step": 23958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4669787124903826e-05,
+      "loss": 0.4602,
+      "step": 23959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466914593485509e-05,
+      "loss": 0.4176,
+      "step": 23960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466850474480636e-05,
+      "loss": 0.3996,
+      "step": 23961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466786355475763e-05,
+      "loss": 0.5078,
+      "step": 23962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46672223647089e-05,
+      "loss": 0.4652,
+      "step": 23963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466658117466017e-05,
+      "loss": 0.5409,
+      "step": 23964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4665939984611436e-05,
+      "loss": 0.5794,
+      "step": 23965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4665298794562715e-05,
+      "loss": 0.5236,
+      "step": 23966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466465760451398e-05,
+      "loss": 0.626,
+      "step": 23967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466401641446525e-05,
+      "loss": 0.4661,
+      "step": 23968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4663375224416517e-05,
+      "loss": 0.3414,
+      "step": 23969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466273403436779e-05,
+      "loss": 0.5547,
+      "step": 23970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466209284431905e-05,
+      "loss": 0.4622,
+      "step": 23971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4661451654270325e-05,
+      "loss": 0.469,
+      "step": 23972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46608104642216e-05,
+      "loss": 0.503,
+      "step": 23973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466016927417287e-05,
+      "loss": 0.4426,
+      "step": 23974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465952808412414e-05,
+      "loss": 0.328,
+      "step": 23975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4658886894075405e-05,
+      "loss": 0.5163,
+      "step": 23976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465824570402668e-05,
+      "loss": 0.5641,
+      "step": 23977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465760451397794e-05,
+      "loss": 0.4847,
+      "step": 23978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465696332392921e-05,
+      "loss": 0.4658,
+      "step": 23979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465632213388048e-05,
+      "loss": 0.5126,
+      "step": 23980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465568094383175e-05,
+      "loss": 0.4284,
+      "step": 23981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465503975378303e-05,
+      "loss": 0.492,
+      "step": 23982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4654398563734293e-05,
+      "loss": 0.3962,
+      "step": 23983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4653757373685565e-05,
+      "loss": 0.4767,
+      "step": 23984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465311618363683e-05,
+      "loss": 0.5224,
+      "step": 23985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46524749935881e-05,
+      "loss": 0.4934,
+      "step": 23986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465183380353937e-05,
+      "loss": 0.4015,
+      "step": 23987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465119261349064e-05,
+      "loss": 0.5895,
+      "step": 23988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465055142344191e-05,
+      "loss": 0.4356,
+      "step": 23989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464991023339318e-05,
+      "loss": 0.5813,
+      "step": 23990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4649269043344454e-05,
+      "loss": 0.4627,
+      "step": 23991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464862785329572e-05,
+      "loss": 0.5327,
+      "step": 23992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464798666324699e-05,
+      "loss": 0.6922,
+      "step": 23993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4647345473198255e-05,
+      "loss": 0.6458,
+      "step": 23994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464670428314953e-05,
+      "loss": 0.4702,
+      "step": 23995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464606309310079e-05,
+      "loss": 0.5032,
+      "step": 23996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4645421903052064e-05,
+      "loss": 0.4393,
+      "step": 23997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4644780713003335e-05,
+      "loss": 0.4459,
+      "step": 23998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464413952295461e-05,
+      "loss": 0.4747,
+      "step": 23999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464349833290588e-05,
+      "loss": 0.5039,
+      "step": 24000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4642857142857144e-05,
+      "loss": 0.3746,
+      "step": 24001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4642215952808415e-05,
+      "loss": 0.4505,
+      "step": 24002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464157476275968e-05,
+      "loss": 0.4402,
+      "step": 24003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464093357271095e-05,
+      "loss": 0.527,
+      "step": 24004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4640292382662224e-05,
+      "loss": 0.5296,
+      "step": 24005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4639651192613496e-05,
+      "loss": 0.4339,
+      "step": 24006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463901000256477e-05,
+      "loss": 0.5531,
+      "step": 24007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463836881251603e-05,
+      "loss": 0.6758,
+      "step": 24008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4637727622467304e-05,
+      "loss": 0.4041,
+      "step": 24009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463708643241857e-05,
+      "loss": 0.574,
+      "step": 24010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463644524236984e-05,
+      "loss": 0.5369,
+      "step": 24011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4635804052321105e-05,
+      "loss": 0.4653,
+      "step": 24012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463516286227238e-05,
+      "loss": 0.4954,
+      "step": 24013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463452167222365e-05,
+      "loss": 0.6168,
+      "step": 24014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463388048217492e-05,
+      "loss": 0.4127,
+      "step": 24015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463323929212619e-05,
+      "loss": 0.4895,
+      "step": 24016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463259810207746e-05,
+      "loss": 0.4064,
+      "step": 24017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463195691202873e-05,
+      "loss": 0.4482,
+      "step": 24018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4631315721979994e-05,
+      "loss": 0.5728,
+      "step": 24019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4630674531931266e-05,
+      "loss": 0.5266,
+      "step": 24020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463003334188254e-05,
+      "loss": 0.6015,
+      "step": 24021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462939215183381e-05,
+      "loss": 0.4306,
+      "step": 24022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4628750961785074e-05,
+      "loss": 0.5749,
+      "step": 24023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4628109771736346e-05,
+      "loss": 0.3833,
+      "step": 24024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462746858168762e-05,
+      "loss": 0.4863,
+      "step": 24025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462682739163888e-05,
+      "loss": 0.5691,
+      "step": 24026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4626186201590154e-05,
+      "loss": 0.5403,
+      "step": 24027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462554501154142e-05,
+      "loss": 0.4186,
+      "step": 24028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462490382149269e-05,
+      "loss": 0.4423,
+      "step": 24029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462426263144396e-05,
+      "loss": 0.5798,
+      "step": 24030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4623621441395234e-05,
+      "loss": 0.6509,
+      "step": 24031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46229802513465e-05,
+      "loss": 0.4826,
+      "step": 24032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462233906129777e-05,
+      "loss": 0.5344,
+      "step": 24033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462169787124904e-05,
+      "loss": 0.5106,
+      "step": 24034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462105668120031e-05,
+      "loss": 0.5415,
+      "step": 24035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462041549115158e-05,
+      "loss": 0.5619,
+      "step": 24036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4619774301102844e-05,
+      "loss": 0.5103,
+      "step": 24037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461913311105412e-05,
+      "loss": 0.5316,
+      "step": 24038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461849192100539e-05,
+      "loss": 0.4004,
+      "step": 24039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461785073095666e-05,
+      "loss": 0.6056,
+      "step": 24040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4617209540907924e-05,
+      "loss": 0.4219,
+      "step": 24041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4616568350859196e-05,
+      "loss": 0.5405,
+      "step": 24042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461592716081047e-05,
+      "loss": 0.5468,
+      "step": 24043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461528597076173e-05,
+      "loss": 0.4137,
+      "step": 24044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4614644780713004e-05,
+      "loss": 0.4554,
+      "step": 24045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4614003590664276e-05,
+      "loss": 0.4623,
+      "step": 24046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461336240061555e-05,
+      "loss": 0.5304,
+      "step": 24047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461272121056681e-05,
+      "loss": 0.4322,
+      "step": 24048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4612080020518084e-05,
+      "loss": 0.5461,
+      "step": 24049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461143883046935e-05,
+      "loss": 0.453,
+      "step": 24050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461079764042062e-05,
+      "loss": 0.4666,
+      "step": 24051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461015645037189e-05,
+      "loss": 0.3882,
+      "step": 24052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.460951526032316e-05,
+      "loss": 0.5523,
+      "step": 24053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4608874070274436e-05,
+      "loss": 0.5164,
+      "step": 24054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46082328802257e-05,
+      "loss": 0.4865,
+      "step": 24055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.460759169017697e-05,
+      "loss": 0.4788,
+      "step": 24056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.460695050012824e-05,
+      "loss": 0.6477,
+      "step": 24057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.460630931007951e-05,
+      "loss": 0.5276,
+      "step": 24058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4605668120030775e-05,
+      "loss": 0.4526,
+      "step": 24059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4605026929982046e-05,
+      "loss": 0.5302,
+      "step": 24060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460438573993332e-05,
+      "loss": 0.5605,
+      "step": 24061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460374454988459e-05,
+      "loss": 0.4483,
+      "step": 24062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460310335983586e-05,
+      "loss": 0.4683,
+      "step": 24063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4602462169787126e-05,
+      "loss": 0.4563,
+      "step": 24064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.46018209797384e-05,
+      "loss": 0.4364,
+      "step": 24065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460117978968966e-05,
+      "loss": 0.4245,
+      "step": 24066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4600538599640935e-05,
+      "loss": 0.4681,
+      "step": 24067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45998974095922e-05,
+      "loss": 0.4258,
+      "step": 24068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459925621954347e-05,
+      "loss": 0.5881,
+      "step": 24069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459861502949474e-05,
+      "loss": 0.4773,
+      "step": 24070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4597973839446015e-05,
+      "loss": 0.4413,
+      "step": 24071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4597332649397286e-05,
+      "loss": 0.584,
+      "step": 24072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459669145934855e-05,
+      "loss": 0.4467,
+      "step": 24073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459605026929982e-05,
+      "loss": 0.4748,
+      "step": 24074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459540907925109e-05,
+      "loss": 0.4617,
+      "step": 24075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459476788920236e-05,
+      "loss": 0.5442,
+      "step": 24076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459412669915363e-05,
+      "loss": 0.4809,
+      "step": 24077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45934855091049e-05,
+      "loss": 0.4847,
+      "step": 24078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4592844319056175e-05,
+      "loss": 0.5163,
+      "step": 24079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459220312900744e-05,
+      "loss": 0.4512,
+      "step": 24080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459156193895871e-05,
+      "loss": 0.4852,
+      "step": 24081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4590920748909977e-05,
+      "loss": 0.3235,
+      "step": 24082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459027955886125e-05,
+      "loss": 0.4694,
+      "step": 24083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458963836881251e-05,
+      "loss": 0.5161,
+      "step": 24084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4588997178763785e-05,
+      "loss": 0.4753,
+      "step": 24085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458835598871506e-05,
+      "loss": 0.5208,
+      "step": 24086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458771479866633e-05,
+      "loss": 0.5328,
+      "step": 24087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45870736086176e-05,
+      "loss": 0.4339,
+      "step": 24088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4586432418568865e-05,
+      "loss": 0.4657,
+      "step": 24089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458579122852014e-05,
+      "loss": 0.5833,
+      "step": 24090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45851500384714e-05,
+      "loss": 0.4443,
+      "step": 24091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4584508848422673e-05,
+      "loss": 0.4278,
+      "step": 24092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4583867658373945e-05,
+      "loss": 0.4677,
+      "step": 24093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458322646832522e-05,
+      "loss": 0.6457,
+      "step": 24094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458258527827648e-05,
+      "loss": 0.4486,
+      "step": 24095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4581944088227753e-05,
+      "loss": 0.5142,
+      "step": 24096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4581302898179025e-05,
+      "loss": 0.4276,
+      "step": 24097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458066170813029e-05,
+      "loss": 0.5134,
+      "step": 24098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458002051808156e-05,
+      "loss": 0.5554,
+      "step": 24099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457937932803283e-05,
+      "loss": 0.4567,
+      "step": 24100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45787381379841e-05,
+      "loss": 0.4922,
+      "step": 24101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457809694793537e-05,
+      "loss": 0.539,
+      "step": 24102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457745575788664e-05,
+      "loss": 0.5896,
+      "step": 24103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457681456783791e-05,
+      "loss": 0.5175,
+      "step": 24104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457617337778918e-05,
+      "loss": 0.4867,
+      "step": 24105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457553218774045e-05,
+      "loss": 0.4613,
+      "step": 24106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4574890997691715e-05,
+      "loss": 0.4208,
+      "step": 24107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457424980764299e-05,
+      "loss": 0.4575,
+      "step": 24108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457360861759425e-05,
+      "loss": 0.3985,
+      "step": 24109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457296742754553e-05,
+      "loss": 0.4241,
+      "step": 24110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4572326237496795e-05,
+      "loss": 0.5204,
+      "step": 24111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457168504744807e-05,
+      "loss": 0.42,
+      "step": 24112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457104385739933e-05,
+      "loss": 0.4825,
+      "step": 24113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4570402667350604e-05,
+      "loss": 0.5606,
+      "step": 24114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4569761477301875e-05,
+      "loss": 0.534,
+      "step": 24115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456912028725314e-05,
+      "loss": 0.6644,
+      "step": 24116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456847909720441e-05,
+      "loss": 0.6599,
+      "step": 24117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4567837907155684e-05,
+      "loss": 0.4515,
+      "step": 24118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4567196717106956e-05,
+      "loss": 0.4685,
+      "step": 24119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456655552705822e-05,
+      "loss": 0.566,
+      "step": 24120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456591433700949e-05,
+      "loss": 0.6215,
+      "step": 24121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456527314696076e-05,
+      "loss": 0.5136,
+      "step": 24122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456463195691203e-05,
+      "loss": 0.4703,
+      "step": 24123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45639907668633e-05,
+      "loss": 0.5105,
+      "step": 24124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4563349576814566e-05,
+      "loss": 0.5074,
+      "step": 24125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4562708386765844e-05,
+      "loss": 0.4581,
+      "step": 24126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456206719671711e-05,
+      "loss": 0.4407,
+      "step": 24127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456142600666838e-05,
+      "loss": 0.4492,
+      "step": 24128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4560784816619646e-05,
+      "loss": 0.4949,
+      "step": 24129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456014362657092e-05,
+      "loss": 0.5483,
+      "step": 24130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455950243652218e-05,
+      "loss": 0.4512,
+      "step": 24131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4558861246473454e-05,
+      "loss": 0.4468,
+      "step": 24132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4558220056424726e-05,
+      "loss": 0.5238,
+      "step": 24133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4557578866376e-05,
+      "loss": 0.6196,
+      "step": 24134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455693767632727e-05,
+      "loss": 0.5494,
+      "step": 24135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4556296486278534e-05,
+      "loss": 0.525,
+      "step": 24136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4555655296229806e-05,
+      "loss": 0.4896,
+      "step": 24137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455501410618107e-05,
+      "loss": 0.4467,
+      "step": 24138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455437291613234e-05,
+      "loss": 0.4443,
+      "step": 24139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455373172608361e-05,
+      "loss": 0.4327,
+      "step": 24140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455309053603488e-05,
+      "loss": 0.5366,
+      "step": 24141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455244934598616e-05,
+      "loss": 0.3717,
+      "step": 24142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455180815593742e-05,
+      "loss": 0.3668,
+      "step": 24143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4551166965888694e-05,
+      "loss": 0.5517,
+      "step": 24144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455052577583996e-05,
+      "loss": 0.5814,
+      "step": 24145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454988458579123e-05,
+      "loss": 0.5195,
+      "step": 24146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4549243395742496e-05,
+      "loss": 0.5306,
+      "step": 24147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454860220569377e-05,
+      "loss": 0.4418,
+      "step": 24148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454796101564504e-05,
+      "loss": 0.4473,
+      "step": 24149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454731982559631e-05,
+      "loss": 0.5156,
+      "step": 24150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454667863554758e-05,
+      "loss": 0.4748,
+      "step": 24151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454603744549885e-05,
+      "loss": 0.6099,
+      "step": 24152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454539625545012e-05,
+      "loss": 0.4567,
+      "step": 24153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4544755065401384e-05,
+      "loss": 0.3817,
+      "step": 24154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4544113875352656e-05,
+      "loss": 0.5129,
+      "step": 24155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454347268530392e-05,
+      "loss": 0.4328,
+      "step": 24156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454283149525519e-05,
+      "loss": 0.4876,
+      "step": 24157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4542190305206464e-05,
+      "loss": 0.4498,
+      "step": 24158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4541549115157736e-05,
+      "loss": 0.3735,
+      "step": 24159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454090792510901e-05,
+      "loss": 0.4906,
+      "step": 24160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454026673506027e-05,
+      "loss": 0.569,
+      "step": 24161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4539625545011544e-05,
+      "loss": 0.4557,
+      "step": 24162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453898435496281e-05,
+      "loss": 0.6386,
+      "step": 24163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453834316491408e-05,
+      "loss": 0.5482,
+      "step": 24164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453770197486535e-05,
+      "loss": 0.5136,
+      "step": 24165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4537060784816625e-05,
+      "loss": 0.4759,
+      "step": 24166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453641959476789e-05,
+      "loss": 0.4398,
+      "step": 24167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453577840471916e-05,
+      "loss": 0.5377,
+      "step": 24168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453513721467043e-05,
+      "loss": 0.432,
+      "step": 24169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45344960246217e-05,
+      "loss": 0.5139,
+      "step": 24170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453385483457297e-05,
+      "loss": 0.4798,
+      "step": 24171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4533213644524235e-05,
+      "loss": 0.5798,
+      "step": 24172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4532572454475506e-05,
+      "loss": 0.483,
+      "step": 24173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453193126442678e-05,
+      "loss": 0.3694,
+      "step": 24174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453129007437805e-05,
+      "loss": 0.4654,
+      "step": 24175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4530648884329315e-05,
+      "loss": 0.5249,
+      "step": 24176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4530007694280586e-05,
+      "loss": 0.4788,
+      "step": 24177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452936650423186e-05,
+      "loss": 0.4287,
+      "step": 24178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452872531418312e-05,
+      "loss": 0.4848,
+      "step": 24179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4528084124134395e-05,
+      "loss": 0.461,
+      "step": 24180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4527442934085666e-05,
+      "loss": 0.4684,
+      "step": 24181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452680174403694e-05,
+      "loss": 0.4531,
+      "step": 24182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45261605539882e-05,
+      "loss": 0.4011,
+      "step": 24183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4525519363939475e-05,
+      "loss": 0.4754,
+      "step": 24184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4524878173890747e-05,
+      "loss": 0.5599,
+      "step": 24185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452423698384201e-05,
+      "loss": 0.5425,
+      "step": 24186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452359579379328e-05,
+      "loss": 0.3867,
+      "step": 24187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452295460374455e-05,
+      "loss": 0.4711,
+      "step": 24188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452231341369582e-05,
+      "loss": 0.6084,
+      "step": 24189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452167222364709e-05,
+      "loss": 0.525,
+      "step": 24190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452103103359836e-05,
+      "loss": 0.4992,
+      "step": 24191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452038984354963e-05,
+      "loss": 0.4446,
+      "step": 24192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45197486535009e-05,
+      "loss": 0.5205,
+      "step": 24193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451910746345217e-05,
+      "loss": 0.5429,
+      "step": 24194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4518466273403437e-05,
+      "loss": 0.5649,
+      "step": 24195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451782508335471e-05,
+      "loss": 0.4681,
+      "step": 24196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451718389330597e-05,
+      "loss": 0.5891,
+      "step": 24197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451654270325725e-05,
+      "loss": 0.5621,
+      "step": 24198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451590151320852e-05,
+      "loss": 0.4291,
+      "step": 24199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451526032315979e-05,
+      "loss": 0.4444,
+      "step": 24200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451461913311105e-05,
+      "loss": 0.5175,
+      "step": 24201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4513977943062325e-05,
+      "loss": 0.4619,
+      "step": 24202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45133367530136e-05,
+      "loss": 0.5167,
+      "step": 24203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451269556296486e-05,
+      "loss": 0.5825,
+      "step": 24204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4512054372916133e-05,
+      "loss": 0.4656,
+      "step": 24205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4511413182867405e-05,
+      "loss": 0.4342,
+      "step": 24206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451077199281868e-05,
+      "loss": 0.413,
+      "step": 24207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451013080276994e-05,
+      "loss": 0.5395,
+      "step": 24208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4509489612721214e-05,
+      "loss": 0.4356,
+      "step": 24209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450884842267248e-05,
+      "loss": 0.3698,
+      "step": 24210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450820723262375e-05,
+      "loss": 0.5114,
+      "step": 24211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450756604257502e-05,
+      "loss": 0.5249,
+      "step": 24212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450692485252629e-05,
+      "loss": 0.456,
+      "step": 24213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4506283662477565e-05,
+      "loss": 0.5863,
+      "step": 24214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450564247242883e-05,
+      "loss": 0.4951,
+      "step": 24215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45050012823801e-05,
+      "loss": 0.4346,
+      "step": 24216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450436009233137e-05,
+      "loss": 0.3242,
+      "step": 24217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450371890228264e-05,
+      "loss": 0.4507,
+      "step": 24218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4503077712233904e-05,
+      "loss": 0.5257,
+      "step": 24219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4502436522185175e-05,
+      "loss": 0.5134,
+      "step": 24220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450179533213645e-05,
+      "loss": 0.4351,
+      "step": 24221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450115414208772e-05,
+      "loss": 0.4102,
+      "step": 24222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450051295203899e-05,
+      "loss": 0.5202,
+      "step": 24223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4499871761990255e-05,
+      "loss": 0.4747,
+      "step": 24224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449923057194153e-05,
+      "loss": 0.3819,
+      "step": 24225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449858938189279e-05,
+      "loss": 0.5514,
+      "step": 24226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4497948191844064e-05,
+      "loss": 0.4738,
+      "step": 24227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449730700179533e-05,
+      "loss": 0.4153,
+      "step": 24228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.44966658117466e-05,
+      "loss": 0.615,
+      "step": 24229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449602462169787e-05,
+      "loss": 0.4059,
+      "step": 24230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4495383431649144e-05,
+      "loss": 0.5058,
+      "step": 24231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4494742241600416e-05,
+      "loss": 0.452,
+      "step": 24232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449410105155168e-05,
+      "loss": 0.428,
+      "step": 24233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449345986150295e-05,
+      "loss": 0.5157,
+      "step": 24234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449281867145422e-05,
+      "loss": 0.4031,
+      "step": 24235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449217748140549e-05,
+      "loss": 0.3954,
+      "step": 24236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449153629135676e-05,
+      "loss": 0.4551,
+      "step": 24237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449089510130803e-05,
+      "loss": 0.4215,
+      "step": 24238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4490253911259304e-05,
+      "loss": 0.4716,
+      "step": 24239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448961272121057e-05,
+      "loss": 0.355,
+      "step": 24240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448897153116184e-05,
+      "loss": 0.4243,
+      "step": 24241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4488330341113106e-05,
+      "loss": 0.4992,
+      "step": 24242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448768915106438e-05,
+      "loss": 0.3609,
+      "step": 24243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448704796101564e-05,
+      "loss": 0.4508,
+      "step": 24244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4486406770966914e-05,
+      "loss": 0.5274,
+      "step": 24245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4485765580918186e-05,
+      "loss": 0.6148,
+      "step": 24246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448512439086946e-05,
+      "loss": 0.3755,
+      "step": 24247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448448320082073e-05,
+      "loss": 0.4449,
+      "step": 24248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4483842010771994e-05,
+      "loss": 0.5478,
+      "step": 24249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4483200820723266e-05,
+      "loss": 0.5618,
+      "step": 24250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448255963067453e-05,
+      "loss": 0.4717,
+      "step": 24251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.44819184406258e-05,
+      "loss": 0.4922,
+      "step": 24252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4481277250577074e-05,
+      "loss": 0.5427,
+      "step": 24253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4480636060528346e-05,
+      "loss": 0.601,
+      "step": 24254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447999487047961e-05,
+      "loss": 0.5504,
+      "step": 24255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447935368043088e-05,
+      "loss": 0.5322,
+      "step": 24256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4478712490382154e-05,
+      "loss": 0.5618,
+      "step": 24257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447807130033342e-05,
+      "loss": 0.6077,
+      "step": 24258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447743011028469e-05,
+      "loss": 0.4589,
+      "step": 24259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4476788920235956e-05,
+      "loss": 0.4674,
+      "step": 24260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447614773018723e-05,
+      "loss": 0.4696,
+      "step": 24261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.44755065401385e-05,
+      "loss": 0.5377,
+      "step": 24262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447486535008977e-05,
+      "loss": 0.4471,
+      "step": 24263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4474224160041036e-05,
+      "loss": 0.4882,
+      "step": 24264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447358296999231e-05,
+      "loss": 0.4399,
+      "step": 24265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447294177994358e-05,
+      "loss": 0.5039,
+      "step": 24266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4472300589894844e-05,
+      "loss": 0.4097,
+      "step": 24267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4471659399846116e-05,
+      "loss": 0.4853,
+      "step": 24268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447101820979739e-05,
+      "loss": 0.4283,
+      "step": 24269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447037701974866e-05,
+      "loss": 0.5411,
+      "step": 24270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4469735829699924e-05,
+      "loss": 0.3788,
+      "step": 24271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4469094639651196e-05,
+      "loss": 0.5625,
+      "step": 24272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446845344960246e-05,
+      "loss": 0.4227,
+      "step": 24273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446781225955373e-05,
+      "loss": 0.5459,
+      "step": 24274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4467171069505005e-05,
+      "loss": 0.4586,
+      "step": 24275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446652987945627e-05,
+      "loss": 0.3891,
+      "step": 24276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446588868940754e-05,
+      "loss": 0.3844,
+      "step": 24277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446524749935881e-05,
+      "loss": 0.4405,
+      "step": 24278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4464606309310085e-05,
+      "loss": 0.4295,
+      "step": 24279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446396511926135e-05,
+      "loss": 0.4336,
+      "step": 24280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446332392921262e-05,
+      "loss": 0.3923,
+      "step": 24281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4462682739163886e-05,
+      "loss": 0.5133,
+      "step": 24282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446204154911516e-05,
+      "loss": 0.5556,
+      "step": 24283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446140035906643e-05,
+      "loss": 0.4971,
+      "step": 24284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4460759169017695e-05,
+      "loss": 0.3648,
+      "step": 24285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446011797896897e-05,
+      "loss": 0.5542,
+      "step": 24286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445947678892024e-05,
+      "loss": 0.4446,
+      "step": 24287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445883559887151e-05,
+      "loss": 0.475,
+      "step": 24288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4458194408822775e-05,
+      "loss": 0.4021,
+      "step": 24289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4457553218774046e-05,
+      "loss": 0.4679,
+      "step": 24290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445691202872531e-05,
+      "loss": 0.3844,
+      "step": 24291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445627083867658e-05,
+      "loss": 0.4873,
+      "step": 24292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4455629648627855e-05,
+      "loss": 0.5075,
+      "step": 24293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4454988458579126e-05,
+      "loss": 0.523,
+      "step": 24294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.44543472685304e-05,
+      "loss": 0.4817,
+      "step": 24295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445370607848166e-05,
+      "loss": 0.4472,
+      "step": 24296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4453064888432935e-05,
+      "loss": 0.4738,
+      "step": 24297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.44524236983842e-05,
+      "loss": 0.5519,
+      "step": 24298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445178250833547e-05,
+      "loss": 0.534,
+      "step": 24299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4451141318286736e-05,
+      "loss": 0.4881,
+      "step": 24300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445050012823801e-05,
+      "loss": 0.2966,
+      "step": 24301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444985893818929e-05,
+      "loss": 0.5174,
+      "step": 24302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444921774814055e-05,
+      "loss": 0.5151,
+      "step": 24303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444857655809182e-05,
+      "loss": 0.5041,
+      "step": 24304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444793536804309e-05,
+      "loss": 0.4855,
+      "step": 24305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444729417799436e-05,
+      "loss": 0.4081,
+      "step": 24306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4446652987945625e-05,
+      "loss": 0.5671,
+      "step": 24307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.44460117978969e-05,
+      "loss": 0.3884,
+      "step": 24308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444537060784817e-05,
+      "loss": 0.363,
+      "step": 24309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444472941779944e-05,
+      "loss": 0.5739,
+      "step": 24310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444408822775071e-05,
+      "loss": 0.5054,
+      "step": 24311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444344703770198e-05,
+      "loss": 0.5038,
+      "step": 24312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444280584765325e-05,
+      "loss": 0.391,
+      "step": 24313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444216465760451e-05,
+      "loss": 0.4924,
+      "step": 24314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4441523467555785e-05,
+      "loss": 0.4158,
+      "step": 24315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444088227750705e-05,
+      "loss": 0.5759,
+      "step": 24316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444024108745832e-05,
+      "loss": 0.5032,
+      "step": 24317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4439599897409593e-05,
+      "loss": 0.4754,
+      "step": 24318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4438958707360865e-05,
+      "loss": 0.3681,
+      "step": 24319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443831751731214e-05,
+      "loss": 0.5007,
+      "step": 24320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.44376763272634e-05,
+      "loss": 0.4971,
+      "step": 24321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4437035137214674e-05,
+      "loss": 0.5395,
+      "step": 24322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443639394716594e-05,
+      "loss": 0.4211,
+      "step": 24323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443575275711721e-05,
+      "loss": 0.5376,
+      "step": 24324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443511156706848e-05,
+      "loss": 0.4982,
+      "step": 24325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4434470377019754e-05,
+      "loss": 0.5468,
+      "step": 24326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443382918697102e-05,
+      "loss": 0.3912,
+      "step": 24327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443318799692229e-05,
+      "loss": 0.5094,
+      "step": 24328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443254680687356e-05,
+      "loss": 0.4012,
+      "step": 24329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443190561682483e-05,
+      "loss": 0.4473,
+      "step": 24330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.44312644267761e-05,
+      "loss": 0.5017,
+      "step": 24331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4430623236727364e-05,
+      "loss": 0.5148,
+      "step": 24332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4429982046678635e-05,
+      "loss": 0.5105,
+      "step": 24333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442934085662991e-05,
+      "loss": 0.5294,
+      "step": 24334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442869966658118e-05,
+      "loss": 0.43,
+      "step": 24335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4428058476532444e-05,
+      "loss": 0.5302,
+      "step": 24336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4427417286483715e-05,
+      "loss": 0.5161,
+      "step": 24337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442677609643499e-05,
+      "loss": 0.4694,
+      "step": 24338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442613490638625e-05,
+      "loss": 0.4508,
+      "step": 24339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4425493716337524e-05,
+      "loss": 0.5667,
+      "step": 24340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4424852526288796e-05,
+      "loss": 0.4597,
+      "step": 24341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442421133624007e-05,
+      "loss": 0.5003,
+      "step": 24342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442357014619133e-05,
+      "loss": 0.4356,
+      "step": 24343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4422928956142604e-05,
+      "loss": 0.4649,
+      "step": 24344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442228776609387e-05,
+      "loss": 0.4219,
+      "step": 24345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442164657604514e-05,
+      "loss": 0.4987,
+      "step": 24346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442100538599641e-05,
+      "loss": 0.6608,
+      "step": 24347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442036419594768e-05,
+      "loss": 0.4901,
+      "step": 24348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441972300589895e-05,
+      "loss": 0.6103,
+      "step": 24349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441908181585022e-05,
+      "loss": 0.4341,
+      "step": 24350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441844062580149e-05,
+      "loss": 0.6226,
+      "step": 24351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441779943575276e-05,
+      "loss": 0.4055,
+      "step": 24352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441715824570403e-05,
+      "loss": 0.5485,
+      "step": 24353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4416517055655294e-05,
+      "loss": 0.4444,
+      "step": 24354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4415875865606566e-05,
+      "loss": 0.4858,
+      "step": 24355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441523467555784e-05,
+      "loss": 0.4804,
+      "step": 24356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.44145934855091e-05,
+      "loss": 0.4452,
+      "step": 24357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441395229546038e-05,
+      "loss": 0.4069,
+      "step": 24358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4413311105411646e-05,
+      "loss": 0.5386,
+      "step": 24359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441266991536292e-05,
+      "loss": 0.5388,
+      "step": 24360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441202872531418e-05,
+      "loss": 0.4009,
+      "step": 24361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4411387535265454e-05,
+      "loss": 0.4509,
+      "step": 24362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441074634521672e-05,
+      "loss": 0.4547,
+      "step": 24363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441010515516799e-05,
+      "loss": 0.558,
+      "step": 24364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440946396511926e-05,
+      "loss": 0.5241,
+      "step": 24365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4408822775070534e-05,
+      "loss": 0.4129,
+      "step": 24366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4408181585021806e-05,
+      "loss": 0.5351,
+      "step": 24367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440754039497307e-05,
+      "loss": 0.4258,
+      "step": 24368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440689920492434e-05,
+      "loss": 0.4347,
+      "step": 24369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440625801487561e-05,
+      "loss": 0.5238,
+      "step": 24370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440561682482688e-05,
+      "loss": 0.5234,
+      "step": 24371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440497563477815e-05,
+      "loss": 0.5275,
+      "step": 24372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4404334444729416e-05,
+      "loss": 0.5351,
+      "step": 24373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4403693254680694e-05,
+      "loss": 0.5333,
+      "step": 24374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440305206463196e-05,
+      "loss": 0.518,
+      "step": 24375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440241087458323e-05,
+      "loss": 0.5659,
+      "step": 24376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4401769684534496e-05,
+      "loss": 0.4401,
+      "step": 24377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440112849448577e-05,
+      "loss": 0.493,
+      "step": 24378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440048730443703e-05,
+      "loss": 0.4549,
+      "step": 24379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4399846114388304e-05,
+      "loss": 0.5148,
+      "step": 24380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4399204924339576e-05,
+      "loss": 0.4566,
+      "step": 24381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439856373429085e-05,
+      "loss": 0.4426,
+      "step": 24382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439792254424212e-05,
+      "loss": 0.5124,
+      "step": 24383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4397281354193384e-05,
+      "loss": 0.5901,
+      "step": 24384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4396640164144656e-05,
+      "loss": 0.4182,
+      "step": 24385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439599897409592e-05,
+      "loss": 0.5629,
+      "step": 24386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439535778404719e-05,
+      "loss": 0.3867,
+      "step": 24387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439471659399846e-05,
+      "loss": 0.4448,
+      "step": 24388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439407540394973e-05,
+      "loss": 0.529,
+      "step": 24389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439343421390101e-05,
+      "loss": 0.5251,
+      "step": 24390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439279302385227e-05,
+      "loss": 0.3995,
+      "step": 24391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4392151833803545e-05,
+      "loss": 0.5438,
+      "step": 24392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439151064375481e-05,
+      "loss": 0.4678,
+      "step": 24393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439086945370608e-05,
+      "loss": 0.4697,
+      "step": 24394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4390228263657346e-05,
+      "loss": 0.4284,
+      "step": 24395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438958707360862e-05,
+      "loss": 0.3841,
+      "step": 24396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438894588355989e-05,
+      "loss": 0.6362,
+      "step": 24397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438830469351116e-05,
+      "loss": 0.4807,
+      "step": 24398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438766350346243e-05,
+      "loss": 0.6555,
+      "step": 24399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43870223134137e-05,
+      "loss": 0.4414,
+      "step": 24400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438638112336497e-05,
+      "loss": 0.566,
+      "step": 24401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4385739933316235e-05,
+      "loss": 0.5167,
+      "step": 24402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4385098743267506e-05,
+      "loss": 0.5942,
+      "step": 24403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438445755321877e-05,
+      "loss": 0.4711,
+      "step": 24404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438381636317004e-05,
+      "loss": 0.5001,
+      "step": 24405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4383175173121315e-05,
+      "loss": 0.4232,
+      "step": 24406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4382533983072587e-05,
+      "loss": 0.6175,
+      "step": 24407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438189279302386e-05,
+      "loss": 0.4493,
+      "step": 24408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438125160297512e-05,
+      "loss": 0.5522,
+      "step": 24409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4380610412926395e-05,
+      "loss": 0.4976,
+      "step": 24410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437996922287766e-05,
+      "loss": 0.4436,
+      "step": 24411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437932803282893e-05,
+      "loss": 0.5464,
+      "step": 24412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43786868427802e-05,
+      "loss": 0.4929,
+      "step": 24413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4378045652731475e-05,
+      "loss": 0.5176,
+      "step": 24414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437740446268274e-05,
+      "loss": 0.5302,
+      "step": 24415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437676327263401e-05,
+      "loss": 0.4594,
+      "step": 24416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437612208258528e-05,
+      "loss": 0.5347,
+      "step": 24417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437548089253655e-05,
+      "loss": 0.4439,
+      "step": 24418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437483970248782e-05,
+      "loss": 0.5052,
+      "step": 24419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4374198512439085e-05,
+      "loss": 0.4251,
+      "step": 24420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437355732239036e-05,
+      "loss": 0.4905,
+      "step": 24421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437291613234163e-05,
+      "loss": 0.5133,
+      "step": 24422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43722749422929e-05,
+      "loss": 0.3589,
+      "step": 24423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4371633752244165e-05,
+      "loss": 0.5353,
+      "step": 24424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437099256219544e-05,
+      "loss": 0.4603,
+      "step": 24425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437035137214671e-05,
+      "loss": 0.5635,
+      "step": 24426
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4369710182097973e-05,
+      "loss": 0.4129,
+      "step": 24427
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4369068992049245e-05,
+      "loss": 0.5184,
+      "step": 24428
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436842780200052e-05,
+      "loss": 0.4491,
+      "step": 24429
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436778661195179e-05,
+      "loss": 0.3643,
+      "step": 24430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4367145421903053e-05,
+      "loss": 0.5556,
+      "step": 24431
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4366504231854325e-05,
+      "loss": 0.4425,
+      "step": 24432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436586304180559e-05,
+      "loss": 0.5845,
+      "step": 24433
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436522185175686e-05,
+      "loss": 0.5408,
+      "step": 24434
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4364580661708134e-05,
+      "loss": 0.3838,
+      "step": 24435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43639394716594e-05,
+      "loss": 0.4937,
+      "step": 24436
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436329828161067e-05,
+      "loss": 0.447,
+      "step": 24437
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436265709156194e-05,
+      "loss": 0.6174,
+      "step": 24438
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4362015901513214e-05,
+      "loss": 0.5896,
+      "step": 24439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436137471146448e-05,
+      "loss": 0.417,
+      "step": 24440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436073352141575e-05,
+      "loss": 0.5038,
+      "step": 24441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4360092331367015e-05,
+      "loss": 0.5408,
+      "step": 24442
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435945114131829e-05,
+      "loss": 0.4442,
+      "step": 24443
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435880995126956e-05,
+      "loss": 0.4717,
+      "step": 24444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4358168761220824e-05,
+      "loss": 0.4197,
+      "step": 24445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43575275711721e-05,
+      "loss": 0.4616,
+      "step": 24446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435688638112337e-05,
+      "loss": 0.2978,
+      "step": 24447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435624519107464e-05,
+      "loss": 0.6055,
+      "step": 24448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4355604001025904e-05,
+      "loss": 0.4379,
+      "step": 24449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4354962810977175e-05,
+      "loss": 0.438,
+      "step": 24450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435432162092844e-05,
+      "loss": 0.5742,
+      "step": 24451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435368043087971e-05,
+      "loss": 0.4547,
+      "step": 24452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4353039240830984e-05,
+      "loss": 0.3947,
+      "step": 24453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4352398050782256e-05,
+      "loss": 0.5393,
+      "step": 24454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435175686073353e-05,
+      "loss": 0.5121,
+      "step": 24455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435111567068479e-05,
+      "loss": 0.4744,
+      "step": 24456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4350474480636064e-05,
+      "loss": 0.6152,
+      "step": 24457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434983329058733e-05,
+      "loss": 0.6814,
+      "step": 24458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43491921005386e-05,
+      "loss": 0.4988,
+      "step": 24459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4348550910489866e-05,
+      "loss": 0.3639,
+      "step": 24460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434790972044114e-05,
+      "loss": 0.5041,
+      "step": 24461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4347268530392416e-05,
+      "loss": 0.451,
+      "step": 24462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434662734034368e-05,
+      "loss": 0.419,
+      "step": 24463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434598615029495e-05,
+      "loss": 0.5324,
+      "step": 24464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434534496024622e-05,
+      "loss": 0.4512,
+      "step": 24465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434470377019749e-05,
+      "loss": 0.3754,
+      "step": 24466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4344062580148754e-05,
+      "loss": 0.4942,
+      "step": 24467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4343421390100026e-05,
+      "loss": 0.4976,
+      "step": 24468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43427802000513e-05,
+      "loss": 0.4325,
+      "step": 24469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434213901000257e-05,
+      "loss": 0.578,
+      "step": 24470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434149781995384e-05,
+      "loss": 0.4159,
+      "step": 24471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4340856629905106e-05,
+      "loss": 0.4459,
+      "step": 24472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434021543985638e-05,
+      "loss": 0.4798,
+      "step": 24473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433957424980764e-05,
+      "loss": 0.5264,
+      "step": 24474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4338933059758914e-05,
+      "loss": 0.4508,
+      "step": 24475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433829186971018e-05,
+      "loss": 0.5528,
+      "step": 24476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433765067966145e-05,
+      "loss": 0.668,
+      "step": 24477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433700948961272e-05,
+      "loss": 0.5433,
+      "step": 24478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4336368299563994e-05,
+      "loss": 0.5067,
+      "step": 24479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4335727109515266e-05,
+      "loss": 0.4979,
+      "step": 24480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433508591946653e-05,
+      "loss": 0.4891,
+      "step": 24481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43344447294178e-05,
+      "loss": 0.4585,
+      "step": 24482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433380353936907e-05,
+      "loss": 0.4349,
+      "step": 24483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433316234932034e-05,
+      "loss": 0.375,
+      "step": 24484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433252115927161e-05,
+      "loss": 0.4977,
+      "step": 24485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433187996922288e-05,
+      "loss": 0.5186,
+      "step": 24486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433123877917415e-05,
+      "loss": 0.4424,
+      "step": 24487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433059758912542e-05,
+      "loss": 0.461,
+      "step": 24488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432995639907669e-05,
+      "loss": 0.4324,
+      "step": 24489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4329315209027956e-05,
+      "loss": 0.4605,
+      "step": 24490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432867401897923e-05,
+      "loss": 0.4684,
+      "step": 24491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432803282893049e-05,
+      "loss": 0.4289,
+      "step": 24492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4327391638881764e-05,
+      "loss": 0.5636,
+      "step": 24493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4326750448833036e-05,
+      "loss": 0.4331,
+      "step": 24494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432610925878431e-05,
+      "loss": 0.4913,
+      "step": 24495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432546806873557e-05,
+      "loss": 0.575,
+      "step": 24496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4324826878686844e-05,
+      "loss": 0.5013,
+      "step": 24497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4324185688638116e-05,
+      "loss": 0.4956,
+      "step": 24498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432354449858938e-05,
+      "loss": 0.5713,
+      "step": 24499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432290330854065e-05,
+      "loss": 0.4901,
+      "step": 24500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4322262118491925e-05,
+      "loss": 0.4547,
+      "step": 24501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4321620928443196e-05,
+      "loss": 0.4985,
+      "step": 24502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432097973839446e-05,
+      "loss": 0.4491,
+      "step": 24503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.432033854834573e-05,
+      "loss": 0.5295,
+      "step": 24504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4319697358297e-05,
+      "loss": 0.5141,
+      "step": 24505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431905616824827e-05,
+      "loss": 0.489,
+      "step": 24506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431841497819954e-05,
+      "loss": 0.5145,
+      "step": 24507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4317773788150806e-05,
+      "loss": 0.5503,
+      "step": 24508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431713259810208e-05,
+      "loss": 0.5527,
+      "step": 24509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431649140805335e-05,
+      "loss": 0.61,
+      "step": 24510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431585021800462e-05,
+      "loss": 0.4675,
+      "step": 24511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4315209027955886e-05,
+      "loss": 0.3957,
+      "step": 24512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431456783790716e-05,
+      "loss": 0.4604,
+      "step": 24513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431392664785842e-05,
+      "loss": 0.5188,
+      "step": 24514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4313285457809695e-05,
+      "loss": 0.4927,
+      "step": 24515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4312644267760966e-05,
+      "loss": 0.505,
+      "step": 24516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431200307771223e-05,
+      "loss": 0.4733,
+      "step": 24517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431136188766351e-05,
+      "loss": 0.4308,
+      "step": 24518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4310720697614775e-05,
+      "loss": 0.4704,
+      "step": 24519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4310079507566047e-05,
+      "loss": 0.4622,
+      "step": 24520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430943831751731e-05,
+      "loss": 0.4223,
+      "step": 24521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430879712746858e-05,
+      "loss": 0.4148,
+      "step": 24522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430815593741985e-05,
+      "loss": 0.4953,
+      "step": 24523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430751474737112e-05,
+      "loss": 0.4782,
+      "step": 24524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430687355732239e-05,
+      "loss": 0.4419,
+      "step": 24525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430623236727366e-05,
+      "loss": 0.4414,
+      "step": 24526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4305591177224935e-05,
+      "loss": 0.6034,
+      "step": 24527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43049499871762e-05,
+      "loss": 0.4244,
+      "step": 24528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430430879712747e-05,
+      "loss": 0.6395,
+      "step": 24529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4303667607078737e-05,
+      "loss": 0.4362,
+      "step": 24530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430302641703001e-05,
+      "loss": 0.4983,
+      "step": 24531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430238522698127e-05,
+      "loss": 0.4619,
+      "step": 24532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4301744036932545e-05,
+      "loss": 0.4912,
+      "step": 24533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4301102846883823e-05,
+      "loss": 0.4334,
+      "step": 24534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430046165683509e-05,
+      "loss": 0.5174,
+      "step": 24535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429982046678636e-05,
+      "loss": 0.4811,
+      "step": 24536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4299179276737625e-05,
+      "loss": 0.5194,
+      "step": 24537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.42985380866889e-05,
+      "loss": 0.3833,
+      "step": 24538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429789689664016e-05,
+      "loss": 0.498,
+      "step": 24539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4297255706591433e-05,
+      "loss": 0.5763,
+      "step": 24540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4296614516542705e-05,
+      "loss": 0.3635,
+      "step": 24541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429597332649398e-05,
+      "loss": 0.4598,
+      "step": 24542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429533213644525e-05,
+      "loss": 0.5288,
+      "step": 24543
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4294690946396514e-05,
+      "loss": 0.4576,
+      "step": 24544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4294049756347785e-05,
+      "loss": 0.4096,
+      "step": 24545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429340856629905e-05,
+      "loss": 0.5667,
+      "step": 24546
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429276737625032e-05,
+      "loss": 0.461,
+      "step": 24547
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429212618620159e-05,
+      "loss": 0.3966,
+      "step": 24548
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429148499615286e-05,
+      "loss": 0.4576,
+      "step": 24549
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.429084380610413e-05,
+      "loss": 0.4686,
+      "step": 24550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.42902026160554e-05,
+      "loss": 0.5482,
+      "step": 24551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4289561426006674e-05,
+      "loss": 0.4802,
+      "step": 24552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.428892023595794e-05,
+      "loss": 0.4142,
+      "step": 24553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.428827904590921e-05,
+      "loss": 0.4329,
+      "step": 24554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4287637855860475e-05,
+      "loss": 0.4261,
+      "step": 24555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.428699666581175e-05,
+      "loss": 0.4439,
+      "step": 24556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.428635547576302e-05,
+      "loss": 0.5403,
+      "step": 24557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.428571428571429e-05,
+      "loss": 0.5838,
+      "step": 24558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4285073095665555e-05,
+      "loss": 0.3696,
+      "step": 24559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.428443190561683e-05,
+      "loss": 0.4454,
+      "step": 24560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.42837907155681e-05,
+      "loss": 0.465,
+      "step": 24561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4283149525519364e-05,
+      "loss": 0.5764,
+      "step": 24562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4282508335470635e-05,
+      "loss": 0.4063,
+      "step": 24563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.42818671454219e-05,
+      "loss": 0.5652,
+      "step": 24564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.428122595537317e-05,
+      "loss": 0.3755,
+      "step": 24565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4280584765324444e-05,
+      "loss": 0.3981,
+      "step": 24566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4279943575275716e-05,
+      "loss": 0.4142,
+      "step": 24567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427930238522699e-05,
+      "loss": 0.5017,
+      "step": 24568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427866119517825e-05,
+      "loss": 0.5568,
+      "step": 24569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4278020005129524e-05,
+      "loss": 0.6094,
+      "step": 24570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427737881508079e-05,
+      "loss": 0.5229,
+      "step": 24571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427673762503206e-05,
+      "loss": 0.443,
+      "step": 24572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427609643498333e-05,
+      "loss": 0.4001,
+      "step": 24573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4275455244934604e-05,
+      "loss": 0.4777,
+      "step": 24574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427481405488587e-05,
+      "loss": 0.4659,
+      "step": 24575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427417286483714e-05,
+      "loss": 0.5115,
+      "step": 24576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427353167478841e-05,
+      "loss": 0.4666,
+      "step": 24577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427289048473968e-05,
+      "loss": 0.5298,
+      "step": 24578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427224929469095e-05,
+      "loss": 0.3734,
+      "step": 24579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4271608104642214e-05,
+      "loss": 0.5029,
+      "step": 24580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4270966914593486e-05,
+      "loss": 0.4415,
+      "step": 24581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.427032572454476e-05,
+      "loss": 0.3732,
+      "step": 24582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426968453449603e-05,
+      "loss": 0.4307,
+      "step": 24583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4269043344447294e-05,
+      "loss": 0.5175,
+      "step": 24584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4268402154398566e-05,
+      "loss": 0.3734,
+      "step": 24585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426776096434984e-05,
+      "loss": 0.526,
+      "step": 24586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42671197743011e-05,
+      "loss": 0.4492,
+      "step": 24587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4266478584252374e-05,
+      "loss": 0.6304,
+      "step": 24588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4265837394203646e-05,
+      "loss": 0.4106,
+      "step": 24589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426519620415492e-05,
+      "loss": 0.3923,
+      "step": 24590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426455501410618e-05,
+      "loss": 0.4072,
+      "step": 24591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4263913824057454e-05,
+      "loss": 0.4809,
+      "step": 24592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426327263400872e-05,
+      "loss": 0.5257,
+      "step": 24593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426263144395999e-05,
+      "loss": 0.4814,
+      "step": 24594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426199025391126e-05,
+      "loss": 0.4648,
+      "step": 24595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426134906386253e-05,
+      "loss": 0.4126,
+      "step": 24596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42607078738138e-05,
+      "loss": 0.5656,
+      "step": 24597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426006668376507e-05,
+      "loss": 0.6817,
+      "step": 24598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425942549371634e-05,
+      "loss": 0.4371,
+      "step": 24599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425878430366761e-05,
+      "loss": 0.4633,
+      "step": 24600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425814311361888e-05,
+      "loss": 0.4943,
+      "step": 24601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4257501923570144e-05,
+      "loss": 0.5477,
+      "step": 24602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4256860733521416e-05,
+      "loss": 0.4609,
+      "step": 24603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425621954347269e-05,
+      "loss": 0.4417,
+      "step": 24604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425557835342395e-05,
+      "loss": 0.4528,
+      "step": 24605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425493716337523e-05,
+      "loss": 0.618,
+      "step": 24606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4254295973326496e-05,
+      "loss": 0.4497,
+      "step": 24607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425365478327777e-05,
+      "loss": 0.498,
+      "step": 24608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425301359322903e-05,
+      "loss": 0.4217,
+      "step": 24609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4252372403180305e-05,
+      "loss": 0.5973,
+      "step": 24610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425173121313157e-05,
+      "loss": 0.4136,
+      "step": 24611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425109002308284e-05,
+      "loss": 0.5005,
+      "step": 24612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425044883303411e-05,
+      "loss": 0.5311,
+      "step": 24613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4249807642985385e-05,
+      "loss": 0.4541,
+      "step": 24614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4249166452936656e-05,
+      "loss": 0.5373,
+      "step": 24615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424852526288792e-05,
+      "loss": 0.484,
+      "step": 24616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424788407283919e-05,
+      "loss": 0.4771,
+      "step": 24617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424724288279046e-05,
+      "loss": 0.5751,
+      "step": 24618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424660169274173e-05,
+      "loss": 0.6178,
+      "step": 24619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4245960502692995e-05,
+      "loss": 0.4675,
+      "step": 24620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4245319312644266e-05,
+      "loss": 0.5171,
+      "step": 24621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4244678122595545e-05,
+      "loss": 0.4678,
+      "step": 24622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424403693254681e-05,
+      "loss": 0.4089,
+      "step": 24623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424339574249808e-05,
+      "loss": 0.454,
+      "step": 24624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4242754552449346e-05,
+      "loss": 0.4222,
+      "step": 24625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424211336240062e-05,
+      "loss": 0.4342,
+      "step": 24626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424147217235188e-05,
+      "loss": 0.4078,
+      "step": 24627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4240830982303155e-05,
+      "loss": 0.4309,
+      "step": 24628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4240189792254426e-05,
+      "loss": 0.4014,
+      "step": 24629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42395486022057e-05,
+      "loss": 0.4585,
+      "step": 24630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423890741215697e-05,
+      "loss": 0.4266,
+      "step": 24631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4238266222108235e-05,
+      "loss": 0.4952,
+      "step": 24632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4237625032059507e-05,
+      "loss": 0.5663,
+      "step": 24633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423698384201077e-05,
+      "loss": 0.4828,
+      "step": 24634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423634265196204e-05,
+      "loss": 0.4133,
+      "step": 24635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423570146191331e-05,
+      "loss": 0.5698,
+      "step": 24636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423506027186458e-05,
+      "loss": 0.4637,
+      "step": 24637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423441908181585e-05,
+      "loss": 0.5356,
+      "step": 24638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423377789176712e-05,
+      "loss": 0.5026,
+      "step": 24639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4233136701718395e-05,
+      "loss": 0.4272,
+      "step": 24640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423249551166966e-05,
+      "loss": 0.4694,
+      "step": 24641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423185432162093e-05,
+      "loss": 0.5806,
+      "step": 24642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42312131315722e-05,
+      "loss": 0.4483,
+      "step": 24643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423057194152347e-05,
+      "loss": 0.4544,
+      "step": 24644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422993075147474e-05,
+      "loss": 0.443,
+      "step": 24645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422928956142601e-05,
+      "loss": 0.4799,
+      "step": 24646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422864837137728e-05,
+      "loss": 0.39,
+      "step": 24647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422800718132855e-05,
+      "loss": 0.433,
+      "step": 24648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422736599127982e-05,
+      "loss": 0.6093,
+      "step": 24649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4226724801231085e-05,
+      "loss": 0.3986,
+      "step": 24650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422608361118236e-05,
+      "loss": 0.4981,
+      "step": 24651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422544242113362e-05,
+      "loss": 0.521,
+      "step": 24652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4224801231084893e-05,
+      "loss": 0.5311,
+      "step": 24653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4224160041036165e-05,
+      "loss": 0.3946,
+      "step": 24654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422351885098744e-05,
+      "loss": 0.3782,
+      "step": 24655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42228776609387e-05,
+      "loss": 0.489,
+      "step": 24656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4222236470889974e-05,
+      "loss": 0.4855,
+      "step": 24657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4221595280841245e-05,
+      "loss": 0.474,
+      "step": 24658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422095409079251e-05,
+      "loss": 0.4483,
+      "step": 24659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422031290074378e-05,
+      "loss": 0.3965,
+      "step": 24660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4219671710695054e-05,
+      "loss": 0.4908,
+      "step": 24661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4219030520646325e-05,
+      "loss": 0.4705,
+      "step": 24662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421838933059759e-05,
+      "loss": 0.4041,
+      "step": 24663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421774814054886e-05,
+      "loss": 0.5353,
+      "step": 24664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421710695050013e-05,
+      "loss": 0.4804,
+      "step": 24665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42164657604514e-05,
+      "loss": 0.3543,
+      "step": 24666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421582457040267e-05,
+      "loss": 0.4151,
+      "step": 24667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4215183380353935e-05,
+      "loss": 0.5362,
+      "step": 24668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421454219030521e-05,
+      "loss": 0.4226,
+      "step": 24669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421390100025648e-05,
+      "loss": 0.408,
+      "step": 24670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421325981020775e-05,
+      "loss": 0.5514,
+      "step": 24671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4212618620159015e-05,
+      "loss": 0.4618,
+      "step": 24672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421197743011029e-05,
+      "loss": 0.5663,
+      "step": 24673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421133624006155e-05,
+      "loss": 0.5103,
+      "step": 24674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4210695050012824e-05,
+      "loss": 0.4907,
+      "step": 24675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4210053859964096e-05,
+      "loss": 0.5557,
+      "step": 24676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420941266991537e-05,
+      "loss": 0.4722,
+      "step": 24677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420877147986664e-05,
+      "loss": 0.659,
+      "step": 24678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4208130289817904e-05,
+      "loss": 0.5417,
+      "step": 24679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4207489099769176e-05,
+      "loss": 0.3023,
+      "step": 24680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420684790972044e-05,
+      "loss": 0.438,
+      "step": 24681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420620671967171e-05,
+      "loss": 0.5718,
+      "step": 24682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420556552962298e-05,
+      "loss": 0.447,
+      "step": 24683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420492433957425e-05,
+      "loss": 0.5173,
+      "step": 24684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420428314952552e-05,
+      "loss": 0.4291,
+      "step": 24685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420364195947679e-05,
+      "loss": 0.4562,
+      "step": 24686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4203000769428064e-05,
+      "loss": 0.4974,
+      "step": 24687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420235957937933e-05,
+      "loss": 0.5427,
+      "step": 24688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42017183893306e-05,
+      "loss": 0.4236,
+      "step": 24689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4201077199281866e-05,
+      "loss": 0.5936,
+      "step": 24690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420043600923314e-05,
+      "loss": 0.3464,
+      "step": 24691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41997948191844e-05,
+      "loss": 0.5255,
+      "step": 24692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4199153629135674e-05,
+      "loss": 0.5583,
+      "step": 24693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419851243908695e-05,
+      "loss": 0.5105,
+      "step": 24694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419787124903822e-05,
+      "loss": 0.5295,
+      "step": 24695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419723005898949e-05,
+      "loss": 0.4613,
+      "step": 24696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4196588868940754e-05,
+      "loss": 0.457,
+      "step": 24697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4195947678892026e-05,
+      "loss": 0.4795,
+      "step": 24698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419530648884329e-05,
+      "loss": 0.5036,
+      "step": 24699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419466529879456e-05,
+      "loss": 0.4482,
+      "step": 24700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4194024108745834e-05,
+      "loss": 0.4222,
+      "step": 24701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4193382918697106e-05,
+      "loss": 0.4227,
+      "step": 24702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419274172864838e-05,
+      "loss": 0.4907,
+      "step": 24703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419210053859964e-05,
+      "loss": 0.5888,
+      "step": 24704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4191459348550914e-05,
+      "loss": 0.3728,
+      "step": 24705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419081815850218e-05,
+      "loss": 0.4011,
+      "step": 24706
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419017696845345e-05,
+      "loss": 0.4337,
+      "step": 24707
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4189535778404716e-05,
+      "loss": 0.4612,
+      "step": 24708
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418889458835599e-05,
+      "loss": 0.4243,
+      "step": 24709
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418825339830726e-05,
+      "loss": 0.4822,
+      "step": 24710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418761220825853e-05,
+      "loss": 0.5241,
+      "step": 24711
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41869710182098e-05,
+      "loss": 0.6086,
+      "step": 24712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418632982816107e-05,
+      "loss": 0.4495,
+      "step": 24713
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418568863811234e-05,
+      "loss": 0.4382,
+      "step": 24714
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4185047448063604e-05,
+      "loss": 0.4993,
+      "step": 24715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4184406258014876e-05,
+      "loss": 0.4782,
+      "step": 24716
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418376506796615e-05,
+      "loss": 0.4236,
+      "step": 24717
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418312387791742e-05,
+      "loss": 0.4814,
+      "step": 24718
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4182482687868684e-05,
+      "loss": 0.5086,
+      "step": 24719
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4181841497819956e-05,
+      "loss": 0.6608,
+      "step": 24720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418120030777123e-05,
+      "loss": 0.535,
+      "step": 24721
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418055911772249e-05,
+      "loss": 0.3857,
+      "step": 24722
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4179917927673765e-05,
+      "loss": 0.4764,
+      "step": 24723
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417927673762503e-05,
+      "loss": 0.5054,
+      "step": 24724
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41786355475763e-05,
+      "loss": 0.5222,
+      "step": 24725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417799435752757e-05,
+      "loss": 0.4238,
+      "step": 24726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4177353167478845e-05,
+      "loss": 0.4256,
+      "step": 24727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417671197743011e-05,
+      "loss": 0.6077,
+      "step": 24728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417607078738138e-05,
+      "loss": 0.5389,
+      "step": 24729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417542959733265e-05,
+      "loss": 0.5731,
+      "step": 24730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417478840728392e-05,
+      "loss": 0.5435,
+      "step": 24731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417414721723519e-05,
+      "loss": 0.4777,
+      "step": 24732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417350602718646e-05,
+      "loss": 0.5447,
+      "step": 24733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417286483713773e-05,
+      "loss": 0.4152,
+      "step": 24734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4172223647089e-05,
+      "loss": 0.462,
+      "step": 24735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417158245704027e-05,
+      "loss": 0.4158,
+      "step": 24736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4170941266991535e-05,
+      "loss": 0.4119,
+      "step": 24737
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4170300076942806e-05,
+      "loss": 0.4036,
+      "step": 24738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416965888689408e-05,
+      "loss": 0.408,
+      "step": 24739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416901769684534e-05,
+      "loss": 0.5046,
+      "step": 24740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4168376506796615e-05,
+      "loss": 0.5031,
+      "step": 24741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4167735316747887e-05,
+      "loss": 0.5227,
+      "step": 24742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416709412669916e-05,
+      "loss": 0.4455,
+      "step": 24743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416645293665042e-05,
+      "loss": 0.4728,
+      "step": 24744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4165811746601695e-05,
+      "loss": 0.4874,
+      "step": 24745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416517055655296e-05,
+      "loss": 0.5303,
+      "step": 24746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416452936650423e-05,
+      "loss": 0.5612,
+      "step": 24747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41638881764555e-05,
+      "loss": 0.4803,
+      "step": 24748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4163246986406775e-05,
+      "loss": 0.515,
+      "step": 24749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416260579635805e-05,
+      "loss": 0.4589,
+      "step": 24750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416196460630931e-05,
+      "loss": 0.4636,
+      "step": 24751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416132341626058e-05,
+      "loss": 0.5055,
+      "step": 24752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416068222621185e-05,
+      "loss": 0.4998,
+      "step": 24753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416004103616312e-05,
+      "loss": 0.4721,
+      "step": 24754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415939984611439e-05,
+      "loss": 0.4458,
+      "step": 24755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415875865606566e-05,
+      "loss": 0.4663,
+      "step": 24756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415811746601693e-05,
+      "loss": 0.5299,
+      "step": 24757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41574762759682e-05,
+      "loss": 0.6337,
+      "step": 24758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415683508591947e-05,
+      "loss": 0.4255,
+      "step": 24759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415619389587074e-05,
+      "loss": 0.5852,
+      "step": 24760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415555270582201e-05,
+      "loss": 0.5338,
+      "step": 24761
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4154911515773273e-05,
+      "loss": 0.5263,
+      "step": 24762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4154270325724545e-05,
+      "loss": 0.4505,
+      "step": 24763
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415362913567582e-05,
+      "loss": 0.5118,
+      "step": 24764
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415298794562708e-05,
+      "loss": 0.4974,
+      "step": 24765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415234675557836e-05,
+      "loss": 0.496,
+      "step": 24766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4151705565529625e-05,
+      "loss": 0.4258,
+      "step": 24767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41510643754809e-05,
+      "loss": 0.4431,
+      "step": 24768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415042318543216e-05,
+      "loss": 0.5018,
+      "step": 24769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4149781995383434e-05,
+      "loss": 0.4995,
+      "step": 24770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41491408053347e-05,
+      "loss": 0.5254,
+      "step": 24771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414849961528597e-05,
+      "loss": 0.4855,
+      "step": 24772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414785842523724e-05,
+      "loss": 0.4706,
+      "step": 24773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4147217235188514e-05,
+      "loss": 0.5374,
+      "step": 24774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4146576045139785e-05,
+      "loss": 0.5199,
+      "step": 24775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414593485509105e-05,
+      "loss": 0.4707,
+      "step": 24776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414529366504232e-05,
+      "loss": 0.4473,
+      "step": 24777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414465247499359e-05,
+      "loss": 0.431,
+      "step": 24778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414401128494486e-05,
+      "loss": 0.4343,
+      "step": 24779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4143370094896124e-05,
+      "loss": 0.5068,
+      "step": 24780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4142728904847395e-05,
+      "loss": 0.4566,
+      "step": 24781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4142087714798674e-05,
+      "loss": 0.4835,
+      "step": 24782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414144652474994e-05,
+      "loss": 0.4914,
+      "step": 24783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414080533470121e-05,
+      "loss": 0.3703,
+      "step": 24784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4140164144652475e-05,
+      "loss": 0.3869,
+      "step": 24785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413952295460375e-05,
+      "loss": 0.4571,
+      "step": 24786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413888176455501e-05,
+      "loss": 0.4587,
+      "step": 24787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4138240574506284e-05,
+      "loss": 0.4773,
+      "step": 24788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4137599384457556e-05,
+      "loss": 0.4695,
+      "step": 24789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413695819440883e-05,
+      "loss": 0.4403,
+      "step": 24790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41363170043601e-05,
+      "loss": 0.406,
+      "step": 24791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4135675814311364e-05,
+      "loss": 0.4862,
+      "step": 24792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4135034624262636e-05,
+      "loss": 0.4283,
+      "step": 24793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41343934342139e-05,
+      "loss": 0.5227,
+      "step": 24794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413375224416517e-05,
+      "loss": 0.468,
+      "step": 24795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413311105411644e-05,
+      "loss": 0.4954,
+      "step": 24796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413246986406771e-05,
+      "loss": 0.4702,
+      "step": 24797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413182867401898e-05,
+      "loss": 0.5805,
+      "step": 24798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413118748397025e-05,
+      "loss": 0.4541,
+      "step": 24799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4130546293921524e-05,
+      "loss": 0.4106,
+      "step": 24800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412990510387279e-05,
+      "loss": 0.6173,
+      "step": 24801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412926391382406e-05,
+      "loss": 0.5313,
+      "step": 24802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4128622723775326e-05,
+      "loss": 0.5171,
+      "step": 24803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41279815337266e-05,
+      "loss": 0.574,
+      "step": 24804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412734034367787e-05,
+      "loss": 0.5684,
+      "step": 24805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412669915362914e-05,
+      "loss": 0.4229,
+      "step": 24806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4126057963580406e-05,
+      "loss": 0.4646,
+      "step": 24807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412541677353168e-05,
+      "loss": 0.3578,
+      "step": 24808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412477558348295e-05,
+      "loss": 0.5918,
+      "step": 24809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4124134393434214e-05,
+      "loss": 0.4529,
+      "step": 24810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4123493203385486e-05,
+      "loss": 0.52,
+      "step": 24811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412285201333675e-05,
+      "loss": 0.5806,
+      "step": 24812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412221082328802e-05,
+      "loss": 0.4449,
+      "step": 24813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4121569633239294e-05,
+      "loss": 0.4547,
+      "step": 24814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4120928443190566e-05,
+      "loss": 0.568,
+      "step": 24815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412028725314183e-05,
+      "loss": 0.4054,
+      "step": 24816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41196460630931e-05,
+      "loss": 0.4439,
+      "step": 24817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4119004873044374e-05,
+      "loss": 0.4202,
+      "step": 24818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411836368299564e-05,
+      "loss": 0.4948,
+      "step": 24819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411772249294691e-05,
+      "loss": 0.4158,
+      "step": 24820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411708130289818e-05,
+      "loss": 0.4486,
+      "step": 24821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4116440112849454e-05,
+      "loss": 0.5005,
+      "step": 24822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411579892280072e-05,
+      "loss": 0.4118,
+      "step": 24823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411515773275199e-05,
+      "loss": 0.5006,
+      "step": 24824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4114516542703256e-05,
+      "loss": 0.5061,
+      "step": 24825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411387535265453e-05,
+      "loss": 0.4511,
+      "step": 24826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41132341626058e-05,
+      "loss": 0.4365,
+      "step": 24827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4112592972557064e-05,
+      "loss": 0.429,
+      "step": 24828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4111951782508336e-05,
+      "loss": 0.398,
+      "step": 24829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411131059245961e-05,
+      "loss": 0.5349,
+      "step": 24830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411066940241088e-05,
+      "loss": 0.4296,
+      "step": 24831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4110028212362144e-05,
+      "loss": 0.3997,
+      "step": 24832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4109387022313416e-05,
+      "loss": 0.406,
+      "step": 24833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.410874583226468e-05,
+      "loss": 0.5112,
+      "step": 24834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.410810464221595e-05,
+      "loss": 0.4918,
+      "step": 24835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4107463452167225e-05,
+      "loss": 0.5948,
+      "step": 24836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4106822262118496e-05,
+      "loss": 0.4802,
+      "step": 24837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.410618107206977e-05,
+      "loss": 0.6003,
+      "step": 24838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.410553988202103e-05,
+      "loss": 0.5843,
+      "step": 24839
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4104898691972305e-05,
+      "loss": 0.4164,
+      "step": 24840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410425750192357e-05,
+      "loss": 0.3841,
+      "step": 24841
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410361631187484e-05,
+      "loss": 0.478,
+      "step": 24842
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4102975121826106e-05,
+      "loss": 0.3682,
+      "step": 24843
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410233393177738e-05,
+      "loss": 0.4552,
+      "step": 24844
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410169274172865e-05,
+      "loss": 0.4815,
+      "step": 24845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410105155167992e-05,
+      "loss": 0.5018,
+      "step": 24846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410041036163119e-05,
+      "loss": 0.3973,
+      "step": 24847
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409976917158246e-05,
+      "loss": 0.5482,
+      "step": 24848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409912798153373e-05,
+      "loss": 0.5461,
+      "step": 24849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4098486791484995e-05,
+      "loss": 0.4907,
+      "step": 24850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4097845601436266e-05,
+      "loss": 0.4909,
+      "step": 24851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409720441138753e-05,
+      "loss": 0.4261,
+      "step": 24852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40965632213388e-05,
+      "loss": 0.4361,
+      "step": 24853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409592203129008e-05,
+      "loss": 0.5266,
+      "step": 24854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4095280841241347e-05,
+      "loss": 0.4854,
+      "step": 24855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409463965119262e-05,
+      "loss": 0.5307,
+      "step": 24856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409399846114388e-05,
+      "loss": 0.473,
+      "step": 24857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4093357271095155e-05,
+      "loss": 0.4743,
+      "step": 24858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409271608104642e-05,
+      "loss": 0.5491,
+      "step": 24859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409207489099769e-05,
+      "loss": 0.5637,
+      "step": 24860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409143370094896e-05,
+      "loss": 0.5103,
+      "step": 24861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4090792510900235e-05,
+      "loss": 0.519,
+      "step": 24862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409015132085151e-05,
+      "loss": 0.532,
+      "step": 24863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408951013080277e-05,
+      "loss": 0.6028,
+      "step": 24864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408886894075404e-05,
+      "loss": 0.5608,
+      "step": 24865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408822775070531e-05,
+      "loss": 0.3851,
+      "step": 24866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408758656065658e-05,
+      "loss": 0.4933,
+      "step": 24867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4086945370607845e-05,
+      "loss": 0.5072,
+      "step": 24868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408630418055912e-05,
+      "loss": 0.4893,
+      "step": 24869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408566299051039e-05,
+      "loss": 0.4253,
+      "step": 24870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408502180046166e-05,
+      "loss": 0.5377,
+      "step": 24871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408438061041293e-05,
+      "loss": 0.4899,
+      "step": 24872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40837394203642e-05,
+      "loss": 0.5091,
+      "step": 24873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408309823031547e-05,
+      "loss": 0.5229,
+      "step": 24874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4082457040266733e-05,
+      "loss": 0.4988,
+      "step": 24875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4081815850218005e-05,
+      "loss": 0.4916,
+      "step": 24876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408117466016928e-05,
+      "loss": 0.4748,
+      "step": 24877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408053347012055e-05,
+      "loss": 0.4816,
+      "step": 24878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4079892280071814e-05,
+      "loss": 0.5318,
+      "step": 24879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4079251090023085e-05,
+      "loss": 0.6076,
+      "step": 24880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407860989997436e-05,
+      "loss": 0.5355,
+      "step": 24881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407796870992562e-05,
+      "loss": 0.4432,
+      "step": 24882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4077327519876894e-05,
+      "loss": 0.4904,
+      "step": 24883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407668632982816e-05,
+      "loss": 0.4446,
+      "step": 24884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407604513977943e-05,
+      "loss": 0.4774,
+      "step": 24885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40754039497307e-05,
+      "loss": 0.4351,
+      "step": 24886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4074762759681974e-05,
+      "loss": 0.4785,
+      "step": 24887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407412156963324e-05,
+      "loss": 0.5235,
+      "step": 24888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407348037958451e-05,
+      "loss": 0.4221,
+      "step": 24889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407283918953578e-05,
+      "loss": 0.4841,
+      "step": 24890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407219799948705e-05,
+      "loss": 0.4717,
+      "step": 24891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407155680943832e-05,
+      "loss": 0.5317,
+      "step": 24892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407091561938959e-05,
+      "loss": 0.3658,
+      "step": 24893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407027442934086e-05,
+      "loss": 0.3831,
+      "step": 24894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406963323929213e-05,
+      "loss": 0.4758,
+      "step": 24895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40689920492434e-05,
+      "loss": 0.5078,
+      "step": 24896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4068350859194664e-05,
+      "loss": 0.4279,
+      "step": 24897
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4067709669145935e-05,
+      "loss": 0.3524,
+      "step": 24898
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406706847909721e-05,
+      "loss": 0.5066,
+      "step": 24899
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406642728904847e-05,
+      "loss": 0.4217,
+      "step": 24900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4065786098999744e-05,
+      "loss": 0.4399,
+      "step": 24901
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4065144908951016e-05,
+      "loss": 0.4739,
+      "step": 24902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406450371890229e-05,
+      "loss": 0.4578,
+      "step": 24903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406386252885355e-05,
+      "loss": 0.4883,
+      "step": 24904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4063221338804824e-05,
+      "loss": 0.6489,
+      "step": 24905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406258014875609e-05,
+      "loss": 0.6378,
+      "step": 24906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406193895870736e-05,
+      "loss": 0.3985,
+      "step": 24907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406129776865863e-05,
+      "loss": 0.6133,
+      "step": 24908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4060656578609904e-05,
+      "loss": 0.3887,
+      "step": 24909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4060015388561176e-05,
+      "loss": 0.4367,
+      "step": 24910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405937419851244e-05,
+      "loss": 0.4096,
+      "step": 24911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405873300846371e-05,
+      "loss": 0.579,
+      "step": 24912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405809181841498e-05,
+      "loss": 0.4869,
+      "step": 24913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405745062836625e-05,
+      "loss": 0.5219,
+      "step": 24914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4056809438317514e-05,
+      "loss": 0.4533,
+      "step": 24915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4056168248268786e-05,
+      "loss": 0.5353,
+      "step": 24916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405552705822006e-05,
+      "loss": 0.5514,
+      "step": 24917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405488586817133e-05,
+      "loss": 0.5024,
+      "step": 24918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40542446781226e-05,
+      "loss": 0.4791,
+      "step": 24919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4053603488073866e-05,
+      "loss": 0.4254,
+      "step": 24920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405296229802514e-05,
+      "loss": 0.5905,
+      "step": 24921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40523211079764e-05,
+      "loss": 0.4331,
+      "step": 24922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4051679917927674e-05,
+      "loss": 0.5721,
+      "step": 24923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405103872787894e-05,
+      "loss": 0.5426,
+      "step": 24924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405039753783022e-05,
+      "loss": 0.467,
+      "step": 24925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404975634778149e-05,
+      "loss": 0.4859,
+      "step": 24926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4049115157732754e-05,
+      "loss": 0.3754,
+      "step": 24927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4048473967684026e-05,
+      "loss": 0.4811,
+      "step": 24928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404783277763529e-05,
+      "loss": 0.5565,
+      "step": 24929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404719158758656e-05,
+      "loss": 0.4996,
+      "step": 24930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404655039753783e-05,
+      "loss": 0.6596,
+      "step": 24931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40459092074891e-05,
+      "loss": 0.4449,
+      "step": 24932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404526801744037e-05,
+      "loss": 0.4635,
+      "step": 24933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404462682739164e-05,
+      "loss": 0.4983,
+      "step": 24934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4043985637342914e-05,
+      "loss": 0.4672,
+      "step": 24935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404334444729418e-05,
+      "loss": 0.3771,
+      "step": 24936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404270325724545e-05,
+      "loss": 0.5187,
+      "step": 24937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4042062067196716e-05,
+      "loss": 0.5169,
+      "step": 24938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404142087714799e-05,
+      "loss": 0.573,
+      "step": 24939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404077968709925e-05,
+      "loss": 0.5252,
+      "step": 24940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4040138497050524e-05,
+      "loss": 0.5663,
+      "step": 24941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40394973070018e-05,
+      "loss": 0.4174,
+      "step": 24942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403885611695307e-05,
+      "loss": 0.4813,
+      "step": 24943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403821492690434e-05,
+      "loss": 0.5663,
+      "step": 24944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4037573736855605e-05,
+      "loss": 0.4704,
+      "step": 24945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4036932546806876e-05,
+      "loss": 0.4283,
+      "step": 24946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403629135675814e-05,
+      "loss": 0.4161,
+      "step": 24947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403565016670941e-05,
+      "loss": 0.528,
+      "step": 24948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4035008976660685e-05,
+      "loss": 0.4732,
+      "step": 24949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4034367786611956e-05,
+      "loss": 0.5448,
+      "step": 24950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403372659656323e-05,
+      "loss": 0.5107,
+      "step": 24951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403308540651449e-05,
+      "loss": 0.4935,
+      "step": 24952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4032444216465765e-05,
+      "loss": 0.3343,
+      "step": 24953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403180302641703e-05,
+      "loss": 0.4351,
+      "step": 24954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40311618363683e-05,
+      "loss": 0.4866,
+      "step": 24955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4030520646319566e-05,
+      "loss": 0.6012,
+      "step": 24956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402987945627084e-05,
+      "loss": 0.5237,
+      "step": 24957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402923826622211e-05,
+      "loss": 0.4824,
+      "step": 24958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402859707617338e-05,
+      "loss": 0.4353,
+      "step": 24959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402795588612465e-05,
+      "loss": 0.4884,
+      "step": 24960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402731469607592e-05,
+      "loss": 0.55,
+      "step": 24961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402667350602719e-05,
+      "loss": 0.4605,
+      "step": 24962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4026032315978455e-05,
+      "loss": 0.3968,
+      "step": 24963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4025391125929726e-05,
+      "loss": 0.4092,
+      "step": 24964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4024749935881e-05,
+      "loss": 0.4839,
+      "step": 24965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402410874583227e-05,
+      "loss": 0.4948,
+      "step": 24966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4023467555783535e-05,
+      "loss": 0.4517,
+      "step": 24967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4022826365734807e-05,
+      "loss": 0.4542,
+      "step": 24968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402218517568608e-05,
+      "loss": 0.5315,
+      "step": 24969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402154398563734e-05,
+      "loss": 0.4513,
+      "step": 24970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4020902795588615e-05,
+      "loss": 0.4757,
+      "step": 24971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402026160553988e-05,
+      "loss": 0.5537,
+      "step": 24972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401962041549115e-05,
+      "loss": 0.5417,
+      "step": 24973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401897922544242e-05,
+      "loss": 0.5131,
+      "step": 24974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4018338035393695e-05,
+      "loss": 0.4734,
+      "step": 24975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401769684534496e-05,
+      "loss": 0.3937,
+      "step": 24976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401705565529623e-05,
+      "loss": 0.4712,
+      "step": 24977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4016414465247503e-05,
+      "loss": 0.3753,
+      "step": 24978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401577327519877e-05,
+      "loss": 0.5621,
+      "step": 24979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401513208515004e-05,
+      "loss": 0.3602,
+      "step": 24980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401449089510131e-05,
+      "loss": 0.4332,
+      "step": 24981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4013849705052584e-05,
+      "loss": 0.4206,
+      "step": 24982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401320851500385e-05,
+      "loss": 0.6004,
+      "step": 24983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401256732495512e-05,
+      "loss": 0.4468,
+      "step": 24984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4011926134906385e-05,
+      "loss": 0.3626,
+      "step": 24985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401128494485766e-05,
+      "loss": 0.4207,
+      "step": 24986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401064375480893e-05,
+      "loss": 0.443,
+      "step": 24987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4010002564760193e-05,
+      "loss": 0.5112,
+      "step": 24988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4009361374711465e-05,
+      "loss": 0.4821,
+      "step": 24989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400872018466274e-05,
+      "loss": 0.5111,
+      "step": 24990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400807899461401e-05,
+      "loss": 0.4566,
+      "step": 24991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4007437804565274e-05,
+      "loss": 0.4702,
+      "step": 24992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4006796614516545e-05,
+      "loss": 0.6974,
+      "step": 24993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400615542446781e-05,
+      "loss": 0.4408,
+      "step": 24994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400551423441908e-05,
+      "loss": 0.6026,
+      "step": 24995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4004873044370354e-05,
+      "loss": 0.4237,
+      "step": 24996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4004231854321625e-05,
+      "loss": 0.4989,
+      "step": 24997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40035906642729e-05,
+      "loss": 0.6686,
+      "step": 24998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400294947422416e-05,
+      "loss": 0.4972,
+      "step": 24999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4002308284175434e-05,
+      "loss": 0.4729,
+      "step": 25000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40016670941267e-05,
+      "loss": 0.341,
+      "step": 25001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400102590407797e-05,
+      "loss": 0.5067,
+      "step": 25002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4000384714029235e-05,
+      "loss": 0.4749,
+      "step": 25003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399974352398051e-05,
+      "loss": 0.4913,
+      "step": 25004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399910233393178e-05,
+      "loss": 0.4161,
+      "step": 25005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399846114388305e-05,
+      "loss": 0.5173,
+      "step": 25006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399781995383432e-05,
+      "loss": 0.5367,
+      "step": 25007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399717876378559e-05,
+      "loss": 0.5398,
+      "step": 25008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399653757373686e-05,
+      "loss": 0.5417,
+      "step": 25009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3995896383688124e-05,
+      "loss": 0.4994,
+      "step": 25010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3995255193639396e-05,
+      "loss": 0.5275,
+      "step": 25011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399461400359066e-05,
+      "loss": 0.5698,
+      "step": 25012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399397281354193e-05,
+      "loss": 0.4471,
+      "step": 25013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399333162349321e-05,
+      "loss": 0.4116,
+      "step": 25014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3992690433444476e-05,
+      "loss": 0.4521,
+      "step": 25015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399204924339575e-05,
+      "loss": 0.4426,
+      "step": 25016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399140805334701e-05,
+      "loss": 0.3506,
+      "step": 25017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3990766863298284e-05,
+      "loss": 0.4788,
+      "step": 25018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399012567324955e-05,
+      "loss": 0.4067,
+      "step": 25019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398948448320082e-05,
+      "loss": 0.4455,
+      "step": 25020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398884329315209e-05,
+      "loss": 0.4711,
+      "step": 25021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3988202103103364e-05,
+      "loss": 0.465,
+      "step": 25022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3987560913054636e-05,
+      "loss": 0.5185,
+      "step": 25023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.39869197230059e-05,
+      "loss": 0.5597,
+      "step": 25024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398627853295717e-05,
+      "loss": 0.4618,
+      "step": 25025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398563734290844e-05,
+      "loss": 0.5254,
+      "step": 25026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398499615285971e-05,
+      "loss": 0.5907,
+      "step": 25027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3984354962810974e-05,
+      "loss": 0.4816,
+      "step": 25028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3983713772762246e-05,
+      "loss": 0.4423,
+      "step": 25029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398307258271352e-05,
+      "loss": 0.4786,
+      "step": 25030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398243139266479e-05,
+      "loss": 0.5156,
+      "step": 25031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398179020261606e-05,
+      "loss": 0.4258,
+      "step": 25032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3981149012567326e-05,
+      "loss": 0.4723,
+      "step": 25033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.39805078225186e-05,
+      "loss": 0.6233,
+      "step": 25034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397986663246986e-05,
+      "loss": 0.5879,
+      "step": 25035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3979225442421134e-05,
+      "loss": 0.5249,
+      "step": 25036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3978584252372406e-05,
+      "loss": 0.424,
+      "step": 25037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397794306232368e-05,
+      "loss": 0.461,
+      "step": 25038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397730187227494e-05,
+      "loss": 0.5147,
+      "step": 25039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3976660682226214e-05,
+      "loss": 0.4194,
+      "step": 25040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3976019492177486e-05,
+      "loss": 0.4621,
+      "step": 25041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397537830212875e-05,
+      "loss": 0.4983,
+      "step": 25042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397473711208002e-05,
+      "loss": 0.3217,
+      "step": 25043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397409592203129e-05,
+      "loss": 0.4009,
+      "step": 25044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397345473198256e-05,
+      "loss": 0.397,
+      "step": 25045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397281354193383e-05,
+      "loss": 0.3175,
+      "step": 25046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.39721723518851e-05,
+      "loss": 0.3941,
+      "step": 25047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397153116183637e-05,
+      "loss": 0.6058,
+      "step": 25048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397088997178764e-05,
+      "loss": 0.5934,
+      "step": 25049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397024878173891e-05,
+      "loss": 0.4645,
+      "step": 25050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3969607591690176e-05,
+      "loss": 0.5373,
+      "step": 25051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396896640164145e-05,
+      "loss": 0.4203,
+      "step": 25052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396832521159272e-05,
+      "loss": 0.5636,
+      "step": 25053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396768402154399e-05,
+      "loss": 0.5464,
+      "step": 25054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3967042831495256e-05,
+      "loss": 0.3711,
+      "step": 25055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396640164144653e-05,
+      "loss": 0.3925,
+      "step": 25056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396576045139779e-05,
+      "loss": 0.4047,
+      "step": 25057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3965119261349065e-05,
+      "loss": 0.5251,
+      "step": 25058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3964478071300336e-05,
+      "loss": 0.453,
+      "step": 25059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.39638368812516e-05,
+      "loss": 0.4062,
+      "step": 25060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396319569120287e-05,
+      "loss": 0.5281,
+      "step": 25061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3962554501154145e-05,
+      "loss": 0.4608,
+      "step": 25062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3961913311105416e-05,
+      "loss": 0.3645,
+      "step": 25063
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396127212105668e-05,
+      "loss": 0.4884,
+      "step": 25064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396063093100795e-05,
+      "loss": 0.6272,
+      "step": 25065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395998974095922e-05,
+      "loss": 0.4994,
+      "step": 25066
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395934855091049e-05,
+      "loss": 0.6017,
+      "step": 25067
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395870736086176e-05,
+      "loss": 0.4551,
+      "step": 25068
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395806617081303e-05,
+      "loss": 0.3934,
+      "step": 25069
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3957424980764305e-05,
+      "loss": 0.412,
+      "step": 25070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395678379071557e-05,
+      "loss": 0.4253,
+      "step": 25071
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395614260066684e-05,
+      "loss": 0.4761,
+      "step": 25072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3955501410618106e-05,
+      "loss": 0.5285,
+      "step": 25073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395486022056938e-05,
+      "loss": 0.4543,
+      "step": 25074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395421903052064e-05,
+      "loss": 0.5602,
+      "step": 25075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3953577840471915e-05,
+      "loss": 0.6188,
+      "step": 25076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3952936650423187e-05,
+      "loss": 0.3898,
+      "step": 25077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395229546037446e-05,
+      "loss": 0.5273,
+      "step": 25078
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395165427032573e-05,
+      "loss": 0.4452,
+      "step": 25079
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3951013080276995e-05,
+      "loss": 0.4911,
+      "step": 25080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395037189022827e-05,
+      "loss": 0.4738,
+      "step": 25081
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394973070017953e-05,
+      "loss": 0.3935,
+      "step": 25082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.39490895101308e-05,
+      "loss": 0.3948,
+      "step": 25083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394844832008207e-05,
+      "loss": 0.3197,
+      "step": 25084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394780713003335e-05,
+      "loss": 0.5157,
+      "step": 25085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394716593998462e-05,
+      "loss": 0.4471,
+      "step": 25086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394652474993588e-05,
+      "loss": 0.4542,
+      "step": 25087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3945883559887155e-05,
+      "loss": 0.6041,
+      "step": 25088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394524236983842e-05,
+      "loss": 0.4037,
+      "step": 25089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394460117978969e-05,
+      "loss": 0.489,
+      "step": 25090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394395998974096e-05,
+      "loss": 0.3736,
+      "step": 25091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394331879969223e-05,
+      "loss": 0.4479,
+      "step": 25092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.39426776096435e-05,
+      "loss": 0.5221,
+      "step": 25093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394203641959477e-05,
+      "loss": 0.3698,
+      "step": 25094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3941395229546044e-05,
+      "loss": 0.4617,
+      "step": 25095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394075403949731e-05,
+      "loss": 0.5438,
+      "step": 25096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394011284944858e-05,
+      "loss": 0.4206,
+      "step": 25097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3939471659399845e-05,
+      "loss": 0.577,
+      "step": 25098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.393883046935112e-05,
+      "loss": 0.5619,
+      "step": 25099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393818927930238e-05,
+      "loss": 0.3596,
+      "step": 25100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3937548089253654e-05,
+      "loss": 0.5355,
+      "step": 25101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3936906899204925e-05,
+      "loss": 0.5098,
+      "step": 25102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39362657091562e-05,
+      "loss": 0.3744,
+      "step": 25103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393562451910747e-05,
+      "loss": 0.6627,
+      "step": 25104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3934983329058734e-05,
+      "loss": 0.6093,
+      "step": 25105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3934342139010005e-05,
+      "loss": 0.479,
+      "step": 25106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393370094896127e-05,
+      "loss": 0.4484,
+      "step": 25107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393305975891254e-05,
+      "loss": 0.4068,
+      "step": 25108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3932418568863814e-05,
+      "loss": 0.6071,
+      "step": 25109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3931777378815085e-05,
+      "loss": 0.5145,
+      "step": 25110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393113618876635e-05,
+      "loss": 0.4393,
+      "step": 25111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393049499871762e-05,
+      "loss": 0.4531,
+      "step": 25112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3929853808668894e-05,
+      "loss": 0.4916,
+      "step": 25113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392921261862016e-05,
+      "loss": 0.5335,
+      "step": 25114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.4386,
+      "step": 25115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3927930238522695e-05,
+      "loss": 0.5333,
+      "step": 25116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392728904847397e-05,
+      "loss": 0.5702,
+      "step": 25117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392664785842524e-05,
+      "loss": 0.4474,
+      "step": 25118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392600666837651e-05,
+      "loss": 0.4629,
+      "step": 25119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3925365478327775e-05,
+      "loss": 0.4069,
+      "step": 25120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392472428827905e-05,
+      "loss": 0.444,
+      "step": 25121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392408309823032e-05,
+      "loss": 0.5803,
+      "step": 25122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3923441908181584e-05,
+      "loss": 0.5108,
+      "step": 25123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3922800718132856e-05,
+      "loss": 0.5384,
+      "step": 25124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392215952808413e-05,
+      "loss": 0.5259,
+      "step": 25125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39215183380354e-05,
+      "loss": 0.4213,
+      "step": 25126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3920877147986664e-05,
+      "loss": 0.4011,
+      "step": 25127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3920235957937936e-05,
+      "loss": 0.3706,
+      "step": 25128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391959476788921e-05,
+      "loss": 0.3903,
+      "step": 25129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391895357784047e-05,
+      "loss": 0.4825,
+      "step": 25130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3918312387791744e-05,
+      "loss": 0.4356,
+      "step": 25131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391767119774301e-05,
+      "loss": 0.354,
+      "step": 25132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391703000769428e-05,
+      "loss": 0.6034,
+      "step": 25133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391638881764555e-05,
+      "loss": 0.4397,
+      "step": 25134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3915747627596824e-05,
+      "loss": 0.447,
+      "step": 25135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391510643754809e-05,
+      "loss": 0.4223,
+      "step": 25136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391446524749936e-05,
+      "loss": 0.5868,
+      "step": 25137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391382405745063e-05,
+      "loss": 0.4472,
+      "step": 25138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39131828674019e-05,
+      "loss": 0.5101,
+      "step": 25139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391254167735317e-05,
+      "loss": 0.5895,
+      "step": 25140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391190048730444e-05,
+      "loss": 0.4721,
+      "step": 25141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391125929725571e-05,
+      "loss": 0.4202,
+      "step": 25142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391061810720698e-05,
+      "loss": 0.6033,
+      "step": 25143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390997691715825e-05,
+      "loss": 0.4493,
+      "step": 25144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3909335727109514e-05,
+      "loss": 0.5889,
+      "step": 25145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3908694537060786e-05,
+      "loss": 0.5457,
+      "step": 25146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390805334701206e-05,
+      "loss": 0.4917,
+      "step": 25147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390741215696332e-05,
+      "loss": 0.3853,
+      "step": 25148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3906770966914594e-05,
+      "loss": 0.4876,
+      "step": 25149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3906129776865866e-05,
+      "loss": 0.455,
+      "step": 25150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390548858681714e-05,
+      "loss": 0.4282,
+      "step": 25151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39048473967684e-05,
+      "loss": 0.4676,
+      "step": 25152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3904206206719674e-05,
+      "loss": 0.5479,
+      "step": 25153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390356501667094e-05,
+      "loss": 0.4878,
+      "step": 25154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390292382662221e-05,
+      "loss": 0.4747,
+      "step": 25155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390228263657348e-05,
+      "loss": 0.4923,
+      "step": 25156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3901641446524754e-05,
+      "loss": 0.5554,
+      "step": 25157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3901000256476026e-05,
+      "loss": 0.493,
+      "step": 25158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390035906642729e-05,
+      "loss": 0.4216,
+      "step": 25159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389971787637856e-05,
+      "loss": 0.4257,
+      "step": 25160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389907668632983e-05,
+      "loss": 0.544,
+      "step": 25161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38984354962811e-05,
+      "loss": 0.4357,
+      "step": 25162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3897794306232364e-05,
+      "loss": 0.5388,
+      "step": 25163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3897153116183636e-05,
+      "loss": 0.4729,
+      "step": 25164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389651192613491e-05,
+      "loss": 0.4985,
+      "step": 25165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389587073608618e-05,
+      "loss": 0.4229,
+      "step": 25166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389522954603745e-05,
+      "loss": 0.6303,
+      "step": 25167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3894588355988716e-05,
+      "loss": 0.5698,
+      "step": 25168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389394716593999e-05,
+      "loss": 0.4908,
+      "step": 25169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389330597589125e-05,
+      "loss": 0.4682,
+      "step": 25170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3892664785842525e-05,
+      "loss": 0.5323,
+      "step": 25171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389202359579379e-05,
+      "loss": 0.6348,
+      "step": 25172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389138240574506e-05,
+      "loss": 0.5301,
+      "step": 25173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389074121569634e-05,
+      "loss": 0.4442,
+      "step": 25174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3890100025647605e-05,
+      "loss": 0.5024,
+      "step": 25175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3889458835598876e-05,
+      "loss": 0.5249,
+      "step": 25176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388881764555014e-05,
+      "loss": 0.5661,
+      "step": 25177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388817645550141e-05,
+      "loss": 0.5405,
+      "step": 25178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388753526545268e-05,
+      "loss": 0.441,
+      "step": 25179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388689407540395e-05,
+      "loss": 0.4705,
+      "step": 25180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388625288535522e-05,
+      "loss": 0.5021,
+      "step": 25181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388561169530649e-05,
+      "loss": 0.4297,
+      "step": 25182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3884970505257765e-05,
+      "loss": 0.5741,
+      "step": 25183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388432931520903e-05,
+      "loss": 0.5745,
+      "step": 25184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38836881251603e-05,
+      "loss": 0.4594,
+      "step": 25185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3883046935111566e-05,
+      "loss": 0.4807,
+      "step": 25186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388240574506284e-05,
+      "loss": 0.5119,
+      "step": 25187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38817645550141e-05,
+      "loss": 0.4815,
+      "step": 25188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3881123364965375e-05,
+      "loss": 0.4149,
+      "step": 25189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3880482174916647e-05,
+      "loss": 0.3345,
+      "step": 25190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387984098486792e-05,
+      "loss": 0.4294,
+      "step": 25191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387919979481919e-05,
+      "loss": 0.5977,
+      "step": 25192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3878558604770455e-05,
+      "loss": 0.4624,
+      "step": 25193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387791741472173e-05,
+      "loss": 0.5103,
+      "step": 25194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387727622467299e-05,
+      "loss": 0.3962,
+      "step": 25195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387663503462426e-05,
+      "loss": 0.5552,
+      "step": 25196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3875993844575535e-05,
+      "loss": 0.4405,
+      "step": 25197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387535265452681e-05,
+      "loss": 0.477,
+      "step": 25198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387471146447807e-05,
+      "loss": 0.5457,
+      "step": 25199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3874070274429343e-05,
+      "loss": 0.4631,
+      "step": 25200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3873429084380615e-05,
+      "loss": 0.5287,
+      "step": 25201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387278789433188e-05,
+      "loss": 0.4833,
+      "step": 25202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387214670428315e-05,
+      "loss": 0.4127,
+      "step": 25203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387150551423442e-05,
+      "loss": 0.536,
+      "step": 25204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387086432418569e-05,
+      "loss": 0.5542,
+      "step": 25205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387022313413696e-05,
+      "loss": 0.4861,
+      "step": 25206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386958194408823e-05,
+      "loss": 0.5243,
+      "step": 25207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38689407540395e-05,
+      "loss": 0.4522,
+      "step": 25208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386829956399077e-05,
+      "loss": 0.5066,
+      "step": 25209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386765837394204e-05,
+      "loss": 0.5254,
+      "step": 25210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3867017183893305e-05,
+      "loss": 0.5026,
+      "step": 25211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386637599384458e-05,
+      "loss": 0.4536,
+      "step": 25212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386573480379585e-05,
+      "loss": 0.456,
+      "step": 25213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386509361374712e-05,
+      "loss": 0.4539,
+      "step": 25214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3864452423698385e-05,
+      "loss": 0.5406,
+      "step": 25215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386381123364966e-05,
+      "loss": 0.5087,
+      "step": 25216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386317004360092e-05,
+      "loss": 0.4285,
+      "step": 25217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3862528853552194e-05,
+      "loss": 0.4417,
+      "step": 25218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3861887663503465e-05,
+      "loss": 0.5412,
+      "step": 25219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386124647345473e-05,
+      "loss": 0.5718,
+      "step": 25220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3860605283406e-05,
+      "loss": 0.4231,
+      "step": 25221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3859964093357274e-05,
+      "loss": 0.5644,
+      "step": 25222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3859322903308545e-05,
+      "loss": 0.4926,
+      "step": 25223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385868171325981e-05,
+      "loss": 0.4772,
+      "step": 25224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385804052321108e-05,
+      "loss": 0.5183,
+      "step": 25225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385739933316235e-05,
+      "loss": 0.497,
+      "step": 25226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385675814311362e-05,
+      "loss": 0.5325,
+      "step": 25227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385611695306489e-05,
+      "loss": 0.4842,
+      "step": 25228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385547576301616e-05,
+      "loss": 0.4243,
+      "step": 25229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3854834572967434e-05,
+      "loss": 0.5239,
+      "step": 25230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38541933829187e-05,
+      "loss": 0.4742,
+      "step": 25231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385355219286997e-05,
+      "loss": 0.3713,
+      "step": 25232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3852911002821236e-05,
+      "loss": 0.4263,
+      "step": 25233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385226981277251e-05,
+      "loss": 0.4609,
+      "step": 25234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385162862272377e-05,
+      "loss": 0.4362,
+      "step": 25235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3850987432675044e-05,
+      "loss": 0.4565,
+      "step": 25236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3850346242626316e-05,
+      "loss": 0.584,
+      "step": 25237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384970505257759e-05,
+      "loss": 0.5129,
+      "step": 25238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384906386252886e-05,
+      "loss": 0.5691,
+      "step": 25239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3848422672480124e-05,
+      "loss": 0.4681,
+      "step": 25240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3847781482431396e-05,
+      "loss": 0.419,
+      "step": 25241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384714029238266e-05,
+      "loss": 0.5552,
+      "step": 25242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384649910233393e-05,
+      "loss": 0.5188,
+      "step": 25243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38458579122852e-05,
+      "loss": 0.4688,
+      "step": 25244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3845216722236476e-05,
+      "loss": 0.4721,
+      "step": 25245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384457553218775e-05,
+      "loss": 0.4803,
+      "step": 25246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384393434213901e-05,
+      "loss": 0.4932,
+      "step": 25247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3843293152090284e-05,
+      "loss": 0.4985,
+      "step": 25248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384265196204155e-05,
+      "loss": 0.5492,
+      "step": 25249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384201077199282e-05,
+      "loss": 0.3906,
+      "step": 25250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3841369581944086e-05,
+      "loss": 0.385,
+      "step": 25251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384072839189536e-05,
+      "loss": 0.4595,
+      "step": 25252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384008720184663e-05,
+      "loss": 0.3967,
+      "step": 25253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38394460117979e-05,
+      "loss": 0.505,
+      "step": 25254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383880482174917e-05,
+      "loss": 0.4904,
+      "step": 25255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383816363170044e-05,
+      "loss": 0.5358,
+      "step": 25256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383752244165171e-05,
+      "loss": 0.4307,
+      "step": 25257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3836881251602974e-05,
+      "loss": 0.4757,
+      "step": 25258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3836240061554246e-05,
+      "loss": 0.5961,
+      "step": 25259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383559887150551e-05,
+      "loss": 0.4477,
+      "step": 25260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383495768145678e-05,
+      "loss": 0.457,
+      "step": 25261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3834316491408054e-05,
+      "loss": 0.4283,
+      "step": 25262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3833675301359326e-05,
+      "loss": 0.3943,
+      "step": 25263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38330341113106e-05,
+      "loss": 0.4964,
+      "step": 25264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383239292126186e-05,
+      "loss": 0.4783,
+      "step": 25265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3831751731213134e-05,
+      "loss": 0.3813,
+      "step": 25266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38311105411644e-05,
+      "loss": 0.356,
+      "step": 25267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383046935111567e-05,
+      "loss": 0.6152,
+      "step": 25268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382982816106694e-05,
+      "loss": 0.5119,
+      "step": 25269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3829186971018214e-05,
+      "loss": 0.4837,
+      "step": 25270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382854578096948e-05,
+      "loss": 0.4434,
+      "step": 25271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382790459092075e-05,
+      "loss": 0.3968,
+      "step": 25272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382726340087202e-05,
+      "loss": 0.4792,
+      "step": 25273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382662221082329e-05,
+      "loss": 0.456,
+      "step": 25274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382598102077456e-05,
+      "loss": 0.4107,
+      "step": 25275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3825339830725824e-05,
+      "loss": 0.4887,
+      "step": 25276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3824698640677096e-05,
+      "loss": 0.4728,
+      "step": 25277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382405745062837e-05,
+      "loss": 0.4272,
+      "step": 25278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382341626057964e-05,
+      "loss": 0.4218,
+      "step": 25279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3822775070530905e-05,
+      "loss": 0.5686,
+      "step": 25280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3822133880482176e-05,
+      "loss": 0.5586,
+      "step": 25281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382149269043345e-05,
+      "loss": 0.5196,
+      "step": 25282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.382085150038471e-05,
+      "loss": 0.3749,
+      "step": 25283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3820210310335985e-05,
+      "loss": 0.3482,
+      "step": 25284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3819569120287256e-05,
+      "loss": 0.399,
+      "step": 25285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381892793023853e-05,
+      "loss": 0.5307,
+      "step": 25286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381828674018979e-05,
+      "loss": 0.4536,
+      "step": 25287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3817645550141065e-05,
+      "loss": 0.6465,
+      "step": 25288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381700436009233e-05,
+      "loss": 0.3267,
+      "step": 25289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38163631700436e-05,
+      "loss": 0.5672,
+      "step": 25290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381572197999487e-05,
+      "loss": 0.5165,
+      "step": 25291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381508078994614e-05,
+      "loss": 0.4415,
+      "step": 25292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381443959989741e-05,
+      "loss": 0.4784,
+      "step": 25293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381379840984868e-05,
+      "loss": 0.545,
+      "step": 25294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381315721979995e-05,
+      "loss": 0.5692,
+      "step": 25295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381251602975122e-05,
+      "loss": 0.395,
+      "step": 25296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381187483970249e-05,
+      "loss": 0.6064,
+      "step": 25297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3811233649653755e-05,
+      "loss": 0.4368,
+      "step": 25298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3810592459605026e-05,
+      "loss": 0.4888,
+      "step": 25299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38099512695563e-05,
+      "loss": 0.5017,
+      "step": 25300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380931007950757e-05,
+      "loss": 0.4386,
+      "step": 25301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380866888945884e-05,
+      "loss": 0.4618,
+      "step": 25302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3808027699410107e-05,
+      "loss": 0.4379,
+      "step": 25303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380738650936138e-05,
+      "loss": 0.4392,
+      "step": 25304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380674531931264e-05,
+      "loss": 0.4467,
+      "step": 25305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3806104129263915e-05,
+      "loss": 0.4916,
+      "step": 25306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380546293921518e-05,
+      "loss": 0.5522,
+      "step": 25307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380482174916645e-05,
+      "loss": 0.4514,
+      "step": 25308
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380418055911772e-05,
+      "loss": 0.4074,
+      "step": 25309
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3803539369068995e-05,
+      "loss": 0.4148,
+      "step": 25310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380289817902027e-05,
+      "loss": 0.4573,
+      "step": 25311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380225698897153e-05,
+      "loss": 0.4839,
+      "step": 25312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3801615798922803e-05,
+      "loss": 0.4053,
+      "step": 25313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380097460887407e-05,
+      "loss": 0.4784,
+      "step": 25314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.380033341882534e-05,
+      "loss": 0.4034,
+      "step": 25315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3799692228776605e-05,
+      "loss": 0.4717,
+      "step": 25316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3799051038727884e-05,
+      "loss": 0.5348,
+      "step": 25317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3798409848679155e-05,
+      "loss": 0.6502,
+      "step": 25318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379776865863042e-05,
+      "loss": 0.5355,
+      "step": 25319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379712746858169e-05,
+      "loss": 0.4059,
+      "step": 25320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379648627853296e-05,
+      "loss": 0.4671,
+      "step": 25321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379584508848423e-05,
+      "loss": 0.5041,
+      "step": 25322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3795203898435493e-05,
+      "loss": 0.5345,
+      "step": 25323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3794562708386765e-05,
+      "loss": 0.4491,
+      "step": 25324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379392151833804e-05,
+      "loss": 0.5334,
+      "step": 25325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379328032828931e-05,
+      "loss": 0.4485,
+      "step": 25326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379263913824058e-05,
+      "loss": 0.4271,
+      "step": 25327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3791997948191845e-05,
+      "loss": 0.4469,
+      "step": 25328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379135675814312e-05,
+      "loss": 0.4587,
+      "step": 25329
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379071556809438e-05,
+      "loss": 0.4609,
+      "step": 25330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3790074378045654e-05,
+      "loss": 0.4993,
+      "step": 25331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378943318799692e-05,
+      "loss": 0.4763,
+      "step": 25332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.37887919979482e-05,
+      "loss": 0.5768,
+      "step": 25333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378815080789947e-05,
+      "loss": 0.5832,
+      "step": 25334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3787509617850734e-05,
+      "loss": 0.4833,
+      "step": 25335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3786868427802005e-05,
+      "loss": 0.4808,
+      "step": 25336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378622723775327e-05,
+      "loss": 0.5759,
+      "step": 25337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378558604770454e-05,
+      "loss": 0.5626,
+      "step": 25338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378494485765581e-05,
+      "loss": 0.5506,
+      "step": 25339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378430366760708e-05,
+      "loss": 0.4425,
+      "step": 25340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378366247755835e-05,
+      "loss": 0.4606,
+      "step": 25341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378302128750962e-05,
+      "loss": 0.4064,
+      "step": 25342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3782380097460894e-05,
+      "loss": 0.4711,
+      "step": 25343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378173890741216e-05,
+      "loss": 0.4904,
+      "step": 25344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378109771736343e-05,
+      "loss": 0.5926,
+      "step": 25345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3780456527314696e-05,
+      "loss": 0.5327,
+      "step": 25346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.377981533726597e-05,
+      "loss": 0.5075,
+      "step": 25347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.377917414721723e-05,
+      "loss": 0.4103,
+      "step": 25348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3778532957168504e-05,
+      "loss": 0.4598,
+      "step": 25349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3777891767119776e-05,
+      "loss": 0.5085,
+      "step": 25350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.377725057707105e-05,
+      "loss": 0.4625,
+      "step": 25351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.377660938702232e-05,
+      "loss": 0.5394,
+      "step": 25352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3775968196973584e-05,
+      "loss": 0.4524,
+      "step": 25353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3775327006924856e-05,
+      "loss": 0.4583,
+      "step": 25354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.377468581687612e-05,
+      "loss": 0.4679,
+      "step": 25355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.377404462682739e-05,
+      "loss": 0.5222,
+      "step": 25356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3773403436778664e-05,
+      "loss": 0.5588,
+      "step": 25357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3772762246729936e-05,
+      "loss": 0.4935,
+      "step": 25358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.37721210566812e-05,
+      "loss": 0.4884,
+      "step": 25359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.377147986663247e-05,
+      "loss": 0.5279,
+      "step": 25360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3770838676583744e-05,
+      "loss": 0.4311,
+      "step": 25361
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.377019748653501e-05,
+      "loss": 0.5083,
+      "step": 25362
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376955629648628e-05,
+      "loss": 0.4228,
+      "step": 25363
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3768915106437546e-05,
+      "loss": 0.5195,
+      "step": 25364
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376827391638882e-05,
+      "loss": 0.4458,
+      "step": 25365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376763272634009e-05,
+      "loss": 0.4763,
+      "step": 25366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376699153629136e-05,
+      "loss": 0.501,
+      "step": 25367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3766350346242626e-05,
+      "loss": 0.4228,
+      "step": 25368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37657091561939e-05,
+      "loss": 0.5558,
+      "step": 25369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376506796614517e-05,
+      "loss": 0.4587,
+      "step": 25370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3764426776096434e-05,
+      "loss": 0.5104,
+      "step": 25371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3763785586047706e-05,
+      "loss": 0.4756,
+      "step": 25372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376314439599898e-05,
+      "loss": 0.4934,
+      "step": 25373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376250320595025e-05,
+      "loss": 0.6275,
+      "step": 25374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3761862015901514e-05,
+      "loss": 0.5245,
+      "step": 25375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3761220825852786e-05,
+      "loss": 0.5933,
+      "step": 25376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376057963580405e-05,
+      "loss": 0.5472,
+      "step": 25377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375993844575532e-05,
+      "loss": 0.5107,
+      "step": 25378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3759297255706594e-05,
+      "loss": 0.4739,
+      "step": 25379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375865606565786e-05,
+      "loss": 0.4444,
+      "step": 25380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375801487560913e-05,
+      "loss": 0.5368,
+      "step": 25381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37573736855604e-05,
+      "loss": 0.419,
+      "step": 25382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3756732495511675e-05,
+      "loss": 0.4551,
+      "step": 25383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375609130546294e-05,
+      "loss": 0.5086,
+      "step": 25384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375545011541421e-05,
+      "loss": 0.5315,
+      "step": 25385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3754808925365476e-05,
+      "loss": 0.4406,
+      "step": 25386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375416773531675e-05,
+      "loss": 0.4799,
+      "step": 25387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375352654526802e-05,
+      "loss": 0.5668,
+      "step": 25388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375288535521929e-05,
+      "loss": 0.3994,
+      "step": 25389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375224416517056e-05,
+      "loss": 0.4764,
+      "step": 25390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375160297512183e-05,
+      "loss": 0.5685,
+      "step": 25391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37509617850731e-05,
+      "loss": 0.5664,
+      "step": 25392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3750320595024365e-05,
+      "loss": 0.4887,
+      "step": 25393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3749679404975636e-05,
+      "loss": 0.4787,
+      "step": 25394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37490382149269e-05,
+      "loss": 0.5318,
+      "step": 25395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374839702487817e-05,
+      "loss": 0.4005,
+      "step": 25396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3747755834829445e-05,
+      "loss": 0.446,
+      "step": 25397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3747114644780716e-05,
+      "loss": 0.4462,
+      "step": 25398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374647345473199e-05,
+      "loss": 0.4268,
+      "step": 25399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374583226468325e-05,
+      "loss": 0.3866,
+      "step": 25400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3745191074634525e-05,
+      "loss": 0.5146,
+      "step": 25401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374454988458579e-05,
+      "loss": 0.512,
+      "step": 25402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374390869453706e-05,
+      "loss": 0.4586,
+      "step": 25403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3743267504488326e-05,
+      "loss": 0.4651,
+      "step": 25404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3742626314439605e-05,
+      "loss": 0.3829,
+      "step": 25405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3741985124390877e-05,
+      "loss": 0.4714,
+      "step": 25406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374134393434214e-05,
+      "loss": 0.4057,
+      "step": 25407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374070274429341e-05,
+      "loss": 0.5046,
+      "step": 25408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374006155424468e-05,
+      "loss": 0.5558,
+      "step": 25409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373942036419595e-05,
+      "loss": 0.4398,
+      "step": 25410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3738779174147215e-05,
+      "loss": 0.524,
+      "step": 25411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3738137984098487e-05,
+      "loss": 0.5236,
+      "step": 25412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373749679404976e-05,
+      "loss": 0.5974,
+      "step": 25413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373685560400103e-05,
+      "loss": 0.5657,
+      "step": 25414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37362144139523e-05,
+      "loss": 0.4586,
+      "step": 25415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373557322390357e-05,
+      "loss": 0.5094,
+      "step": 25416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373493203385484e-05,
+      "loss": 0.4925,
+      "step": 25417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37342908438061e-05,
+      "loss": 0.5191,
+      "step": 25418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3733649653757375e-05,
+      "loss": 0.53,
+      "step": 25419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373300846370864e-05,
+      "loss": 0.5136,
+      "step": 25420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373236727365991e-05,
+      "loss": 0.5409,
+      "step": 25421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373172608361118e-05,
+      "loss": 0.5062,
+      "step": 25422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3731084893562455e-05,
+      "loss": 0.5249,
+      "step": 25423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373044370351373e-05,
+      "loss": 0.4453,
+      "step": 25424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372980251346499e-05,
+      "loss": 0.3787,
+      "step": 25425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3729161323416263e-05,
+      "loss": 0.4543,
+      "step": 25426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372852013336753e-05,
+      "loss": 0.5392,
+      "step": 25427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37278789433188e-05,
+      "loss": 0.4734,
+      "step": 25428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372723775327007e-05,
+      "loss": 0.5566,
+      "step": 25429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3726596563221344e-05,
+      "loss": 0.5016,
+      "step": 25430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372595537317261e-05,
+      "loss": 0.3943,
+      "step": 25431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372531418312388e-05,
+      "loss": 0.4617,
+      "step": 25432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372467299307515e-05,
+      "loss": 0.5303,
+      "step": 25433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372403180302642e-05,
+      "loss": 0.5444,
+      "step": 25434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372339061297769e-05,
+      "loss": 0.6627,
+      "step": 25435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3722749422928954e-05,
+      "loss": 0.4685,
+      "step": 25436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3722108232880225e-05,
+      "loss": 0.4463,
+      "step": 25437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37214670428315e-05,
+      "loss": 0.4848,
+      "step": 25438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372082585278277e-05,
+      "loss": 0.442,
+      "step": 25439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3720184662734034e-05,
+      "loss": 0.4799,
+      "step": 25440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3719543472685305e-05,
+      "loss": 0.4912,
+      "step": 25441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371890228263658e-05,
+      "loss": 0.4708,
+      "step": 25442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371826109258784e-05,
+      "loss": 0.5124,
+      "step": 25443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3717619902539114e-05,
+      "loss": 0.4775,
+      "step": 25444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3716978712490385e-05,
+      "loss": 0.4617,
+      "step": 25445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371633752244166e-05,
+      "loss": 0.4087,
+      "step": 25446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371569633239292e-05,
+      "loss": 0.4569,
+      "step": 25447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3715055142344194e-05,
+      "loss": 0.4862,
+      "step": 25448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371441395229546e-05,
+      "loss": 0.4484,
+      "step": 25449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371377276224673e-05,
+      "loss": 0.5415,
+      "step": 25450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3713131572198e-05,
+      "loss": 0.4199,
+      "step": 25451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371249038214927e-05,
+      "loss": 0.4536,
+      "step": 25452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371184919210054e-05,
+      "loss": 0.488,
+      "step": 25453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371120800205181e-05,
+      "loss": 0.5523,
+      "step": 25454
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371056681200308e-05,
+      "loss": 0.6406,
+      "step": 25455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370992562195435e-05,
+      "loss": 0.4743,
+      "step": 25456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370928443190562e-05,
+      "loss": 0.3712,
+      "step": 25457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3708643241856884e-05,
+      "loss": 0.4662,
+      "step": 25458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3708002051808156e-05,
+      "loss": 0.5588,
+      "step": 25459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370736086175943e-05,
+      "loss": 0.5334,
+      "step": 25460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37067196717107e-05,
+      "loss": 0.4548,
+      "step": 25461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370607848166197e-05,
+      "loss": 0.4714,
+      "step": 25462
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3705437291613236e-05,
+      "loss": 0.4742,
+      "step": 25463
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370479610156451e-05,
+      "loss": 0.5599,
+      "step": 25464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370415491151577e-05,
+      "loss": 0.5721,
+      "step": 25465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3703513721467044e-05,
+      "loss": 0.462,
+      "step": 25466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370287253141831e-05,
+      "loss": 0.486,
+      "step": 25467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370223134136958e-05,
+      "loss": 0.4429,
+      "step": 25468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370159015132085e-05,
+      "loss": 0.381,
+      "step": 25469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3700948961272124e-05,
+      "loss": 0.4593,
+      "step": 25470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3700307771223396e-05,
+      "loss": 0.6097,
+      "step": 25471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369966658117466e-05,
+      "loss": 0.529,
+      "step": 25472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369902539112593e-05,
+      "loss": 0.4584,
+      "step": 25473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36983842010772e-05,
+      "loss": 0.505,
+      "step": 25474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369774301102847e-05,
+      "loss": 0.5279,
+      "step": 25475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3697101820979734e-05,
+      "loss": 0.5228,
+      "step": 25476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369646063093101e-05,
+      "loss": 0.3372,
+      "step": 25477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3695819440882284e-05,
+      "loss": 0.5201,
+      "step": 25478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369517825083355e-05,
+      "loss": 0.4368,
+      "step": 25479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369453706078482e-05,
+      "loss": 0.4225,
+      "step": 25480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3693895870736086e-05,
+      "loss": 0.4649,
+      "step": 25481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369325468068736e-05,
+      "loss": 0.5609,
+      "step": 25482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369261349063862e-05,
+      "loss": 0.6058,
+      "step": 25483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3691972300589894e-05,
+      "loss": 0.599,
+      "step": 25484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3691331110541166e-05,
+      "loss": 0.597,
+      "step": 25485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369068992049244e-05,
+      "loss": 0.5341,
+      "step": 25486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369004873044371e-05,
+      "loss": 0.4191,
+      "step": 25487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3689407540394974e-05,
+      "loss": 0.451,
+      "step": 25488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3688766350346246e-05,
+      "loss": 0.5026,
+      "step": 25489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368812516029751e-05,
+      "loss": 0.4899,
+      "step": 25490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368748397024878e-05,
+      "loss": 0.5148,
+      "step": 25491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368684278020005e-05,
+      "loss": 0.4596,
+      "step": 25492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3686201590151326e-05,
+      "loss": 0.4404,
+      "step": 25493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368556040010259e-05,
+      "loss": 0.4682,
+      "step": 25494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368491921005386e-05,
+      "loss": 0.4767,
+      "step": 25495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3684278020005135e-05,
+      "loss": 0.5865,
+      "step": 25496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36836368299564e-05,
+      "loss": 0.5169,
+      "step": 25497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368299563990767e-05,
+      "loss": 0.7067,
+      "step": 25498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3682354449858936e-05,
+      "loss": 0.5324,
+      "step": 25499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368171325981021e-05,
+      "loss": 0.5146,
+      "step": 25500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368107206976148e-05,
+      "loss": 0.4942,
+      "step": 25501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368043087971275e-05,
+      "loss": 0.4034,
+      "step": 25502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3679789689664016e-05,
+      "loss": 0.5416,
+      "step": 25503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367914849961529e-05,
+      "loss": 0.4258,
+      "step": 25504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367850730956656e-05,
+      "loss": 0.4693,
+      "step": 25505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3677866119517825e-05,
+      "loss": 0.5338,
+      "step": 25506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3677224929469096e-05,
+      "loss": 0.4275,
+      "step": 25507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367658373942036e-05,
+      "loss": 0.443,
+      "step": 25508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367594254937163e-05,
+      "loss": 0.4842,
+      "step": 25509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3675301359322905e-05,
+      "loss": 0.5212,
+      "step": 25510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3674660169274176e-05,
+      "loss": 0.5734,
+      "step": 25511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367401897922545e-05,
+      "loss": 0.4273,
+      "step": 25512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367337778917671e-05,
+      "loss": 0.4256,
+      "step": 25513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3672736599127985e-05,
+      "loss": 0.3936,
+      "step": 25514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367209540907925e-05,
+      "loss": 0.4884,
+      "step": 25515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367145421903052e-05,
+      "loss": 0.5292,
+      "step": 25516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367081302898179e-05,
+      "loss": 0.5954,
+      "step": 25517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3670171838933065e-05,
+      "loss": 0.5151,
+      "step": 25518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366953064888433e-05,
+      "loss": 0.4497,
+      "step": 25519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36688894588356e-05,
+      "loss": 0.5447,
+      "step": 25520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366824826878687e-05,
+      "loss": 0.4679,
+      "step": 25521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366760707873814e-05,
+      "loss": 0.4427,
+      "step": 25522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366696588868941e-05,
+      "loss": 0.5107,
+      "step": 25523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3666324698640675e-05,
+      "loss": 0.3774,
+      "step": 25524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3665683508591947e-05,
+      "loss": 0.5168,
+      "step": 25525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366504231854322e-05,
+      "loss": 0.2846,
+      "step": 25526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366440112849449e-05,
+      "loss": 0.4411,
+      "step": 25527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3663759938445755e-05,
+      "loss": 0.4772,
+      "step": 25528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366311874839703e-05,
+      "loss": 0.4019,
+      "step": 25529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36624775583483e-05,
+      "loss": 0.4985,
+      "step": 25530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366183636829956e-05,
+      "loss": 0.5418,
+      "step": 25531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3661195178250835e-05,
+      "loss": 0.484,
+      "step": 25532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366055398820211e-05,
+      "loss": 0.5226,
+      "step": 25533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365991279815338e-05,
+      "loss": 0.4205,
+      "step": 25534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3659271608104643e-05,
+      "loss": 0.4518,
+      "step": 25535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3658630418055915e-05,
+      "loss": 0.4729,
+      "step": 25536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365798922800718e-05,
+      "loss": 0.4464,
+      "step": 25537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365734803795845e-05,
+      "loss": 0.4582,
+      "step": 25538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3656706847909723e-05,
+      "loss": 0.4576,
+      "step": 25539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365606565786099e-05,
+      "loss": 0.4446,
+      "step": 25540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365542446781226e-05,
+      "loss": 0.429,
+      "step": 25541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365478327776353e-05,
+      "loss": 0.5683,
+      "step": 25542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3654142087714804e-05,
+      "loss": 0.5434,
+      "step": 25543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365350089766607e-05,
+      "loss": 0.4623,
+      "step": 25544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365285970761734e-05,
+      "loss": 0.5163,
+      "step": 25545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3652218517568605e-05,
+      "loss": 0.5724,
+      "step": 25546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365157732751988e-05,
+      "loss": 0.5635,
+      "step": 25547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365093613747115e-05,
+      "loss": 0.4251,
+      "step": 25548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.365029494742242e-05,
+      "loss": 0.5921,
+      "step": 25549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364965375737369e-05,
+      "loss": 0.4159,
+      "step": 25550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364901256732496e-05,
+      "loss": 0.4468,
+      "step": 25551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364837137727623e-05,
+      "loss": 0.5763,
+      "step": 25552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3647730187227494e-05,
+      "loss": 0.5508,
+      "step": 25553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3647088997178765e-05,
+      "loss": 0.4402,
+      "step": 25554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364644780713003e-05,
+      "loss": 0.457,
+      "step": 25555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36458066170813e-05,
+      "loss": 0.471,
+      "step": 25556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3645165427032574e-05,
+      "loss": 0.4318,
+      "step": 25557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3644524236983845e-05,
+      "loss": 0.507,
+      "step": 25558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364388304693512e-05,
+      "loss": 0.4377,
+      "step": 25559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364324185688638e-05,
+      "loss": 0.4863,
+      "step": 25560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3642600666837654e-05,
+      "loss": 0.45,
+      "step": 25561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364195947678892e-05,
+      "loss": 0.418,
+      "step": 25562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364131828674019e-05,
+      "loss": 0.5859,
+      "step": 25563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3640677096691455e-05,
+      "loss": 0.4324,
+      "step": 25564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3640035906642734e-05,
+      "loss": 0.5444,
+      "step": 25565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3639394716594006e-05,
+      "loss": 0.6352,
+      "step": 25566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363875352654527e-05,
+      "loss": 0.4175,
+      "step": 25567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363811233649654e-05,
+      "loss": 0.3633,
+      "step": 25568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363747114644781e-05,
+      "loss": 0.5098,
+      "step": 25569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363682995639908e-05,
+      "loss": 0.4507,
+      "step": 25570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3636188766350344e-05,
+      "loss": 0.5263,
+      "step": 25571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3635547576301616e-05,
+      "loss": 0.4565,
+      "step": 25572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363490638625289e-05,
+      "loss": 0.5589,
+      "step": 25573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363426519620416e-05,
+      "loss": 0.4971,
+      "step": 25574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363362400615543e-05,
+      "loss": 0.4327,
+      "step": 25575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3632982816106696e-05,
+      "loss": 0.5015,
+      "step": 25576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363234162605797e-05,
+      "loss": 0.4862,
+      "step": 25577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363170043600923e-05,
+      "loss": 0.4793,
+      "step": 25578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3631059245960504e-05,
+      "loss": 0.4205,
+      "step": 25579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.363041805591177e-05,
+      "loss": 0.4947,
+      "step": 25580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362977686586304e-05,
+      "loss": 0.4341,
+      "step": 25581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362913567581431e-05,
+      "loss": 0.6476,
+      "step": 25582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3628494485765584e-05,
+      "loss": 0.4912,
+      "step": 25583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3627853295716856e-05,
+      "loss": 0.5042,
+      "step": 25584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362721210566812e-05,
+      "loss": 0.5621,
+      "step": 25585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362657091561939e-05,
+      "loss": 0.5273,
+      "step": 25586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362592972557066e-05,
+      "loss": 0.5329,
+      "step": 25587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362528853552193e-05,
+      "loss": 0.4707,
+      "step": 25588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36246473454732e-05,
+      "loss": 0.3916,
+      "step": 25589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362400615542447e-05,
+      "loss": 0.6777,
+      "step": 25590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362336496537574e-05,
+      "loss": 0.5798,
+      "step": 25591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362272377532701e-05,
+      "loss": 0.4018,
+      "step": 25592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362208258527828e-05,
+      "loss": 0.5091,
+      "step": 25593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3621441395229546e-05,
+      "loss": 0.428,
+      "step": 25594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362080020518082e-05,
+      "loss": 0.4625,
+      "step": 25595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362015901513208e-05,
+      "loss": 0.4196,
+      "step": 25596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3619517825083354e-05,
+      "loss": 0.4801,
+      "step": 25597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3618876635034626e-05,
+      "loss": 0.486,
+      "step": 25598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36182354449859e-05,
+      "loss": 0.3606,
+      "step": 25599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.361759425493716e-05,
+      "loss": 0.5063,
+      "step": 25600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3616953064888434e-05,
+      "loss": 0.5311,
+      "step": 25601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3616311874839706e-05,
+      "loss": 0.3989,
+      "step": 25602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.361567068479097e-05,
+      "loss": 0.5563,
+      "step": 25603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.361502949474224e-05,
+      "loss": 0.5417,
+      "step": 25604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3614388304693514e-05,
+      "loss": 0.5456,
+      "step": 25605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3613747114644786e-05,
+      "loss": 0.405,
+      "step": 25606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.361310592459605e-05,
+      "loss": 0.4937,
+      "step": 25607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.361246473454732e-05,
+      "loss": 0.4618,
+      "step": 25608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.361182354449859e-05,
+      "loss": 0.4866,
+      "step": 25609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.361118235444986e-05,
+      "loss": 0.4363,
+      "step": 25610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.361054116440113e-05,
+      "loss": 0.4467,
+      "step": 25611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3609899974352396e-05,
+      "loss": 0.4358,
+      "step": 25612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.360925878430367e-05,
+      "loss": 0.4238,
+      "step": 25613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.360861759425494e-05,
+      "loss": 0.5656,
+      "step": 25614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.360797640420621e-05,
+      "loss": 0.4136,
+      "step": 25615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3607335214157476e-05,
+      "loss": 0.5106,
+      "step": 25616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.360669402410875e-05,
+      "loss": 0.6268,
+      "step": 25617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.360605283406001e-05,
+      "loss": 0.4967,
+      "step": 25618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3605411644011285e-05,
+      "loss": 0.5864,
+      "step": 25619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3604770453962556e-05,
+      "loss": 0.4804,
+      "step": 25620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.360412926391383e-05,
+      "loss": 0.3961,
+      "step": 25621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.36034880738651e-05,
+      "loss": 0.55,
+      "step": 25622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3602846883816365e-05,
+      "loss": 0.5528,
+      "step": 25623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3602205693767636e-05,
+      "loss": 0.5423,
+      "step": 25624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.36015645037189e-05,
+      "loss": 0.5739,
+      "step": 25625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.360092331367017e-05,
+      "loss": 0.4318,
+      "step": 25626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.360028212362144e-05,
+      "loss": 0.6169,
+      "step": 25627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359964093357271e-05,
+      "loss": 0.4488,
+      "step": 25628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359899974352398e-05,
+      "loss": 0.4338,
+      "step": 25629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359835855347525e-05,
+      "loss": 0.5375,
+      "step": 25630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3597717363426525e-05,
+      "loss": 0.4824,
+      "step": 25631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359707617337779e-05,
+      "loss": 0.4985,
+      "step": 25632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359643498332906e-05,
+      "loss": 0.3921,
+      "step": 25633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3595793793280327e-05,
+      "loss": 0.4297,
+      "step": 25634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35951526032316e-05,
+      "loss": 0.3872,
+      "step": 25635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359451141318286e-05,
+      "loss": 0.5291,
+      "step": 25636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359387022313414e-05,
+      "loss": 0.4381,
+      "step": 25637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359322903308541e-05,
+      "loss": 0.4797,
+      "step": 25638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359258784303668e-05,
+      "loss": 0.5338,
+      "step": 25639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359194665298795e-05,
+      "loss": 0.6074,
+      "step": 25640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3591305462939215e-05,
+      "loss": 0.54,
+      "step": 25641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359066427289049e-05,
+      "loss": 0.4138,
+      "step": 25642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359002308284175e-05,
+      "loss": 0.5542,
+      "step": 25643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358938189279302e-05,
+      "loss": 0.5145,
+      "step": 25644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3588740702744295e-05,
+      "loss": 0.4124,
+      "step": 25645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358809951269557e-05,
+      "loss": 0.619,
+      "step": 25646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358745832264684e-05,
+      "loss": 0.3809,
+      "step": 25647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3586817132598103e-05,
+      "loss": 0.5077,
+      "step": 25648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3586175942549375e-05,
+      "loss": 0.4435,
+      "step": 25649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358553475250064e-05,
+      "loss": 0.4064,
+      "step": 25650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358489356245191e-05,
+      "loss": 0.4916,
+      "step": 25651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358425237240318e-05,
+      "loss": 0.4598,
+      "step": 25652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3583611182354455e-05,
+      "loss": 0.4069,
+      "step": 25653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358296999230572e-05,
+      "loss": 0.396,
+      "step": 25654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358232880225699e-05,
+      "loss": 0.5606,
+      "step": 25655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3581687612208264e-05,
+      "loss": 0.4478,
+      "step": 25656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358104642215953e-05,
+      "loss": 0.4971,
+      "step": 25657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35804052321108e-05,
+      "loss": 0.457,
+      "step": 25658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3579764042062065e-05,
+      "loss": 0.5348,
+      "step": 25659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357912285201334e-05,
+      "loss": 0.4503,
+      "step": 25660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357848166196461e-05,
+      "loss": 0.4318,
+      "step": 25661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357784047191588e-05,
+      "loss": 0.4342,
+      "step": 25662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3577199281867145e-05,
+      "loss": 0.4974,
+      "step": 25663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357655809181842e-05,
+      "loss": 0.4334,
+      "step": 25664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357591690176969e-05,
+      "loss": 0.4704,
+      "step": 25665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3575275711720954e-05,
+      "loss": 0.4695,
+      "step": 25666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3574634521672225e-05,
+      "loss": 0.3999,
+      "step": 25667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357399333162349e-05,
+      "loss": 0.4609,
+      "step": 25668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357335214157476e-05,
+      "loss": 0.3851,
+      "step": 25669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3572710951526034e-05,
+      "loss": 0.3406,
+      "step": 25670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3572069761477305e-05,
+      "loss": 0.4546,
+      "step": 25671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357142857142857e-05,
+      "loss": 0.5464,
+      "step": 25672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357078738137984e-05,
+      "loss": 0.493,
+      "step": 25673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3570146191331114e-05,
+      "loss": 0.4303,
+      "step": 25674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356950500128238e-05,
+      "loss": 0.4443,
+      "step": 25675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356886381123365e-05,
+      "loss": 0.441,
+      "step": 25676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356822262118492e-05,
+      "loss": 0.3819,
+      "step": 25677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3567581431136194e-05,
+      "loss": 0.413,
+      "step": 25678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356694024108746e-05,
+      "loss": 0.5788,
+      "step": 25679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356629905103873e-05,
+      "loss": 0.6073,
+      "step": 25680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3565657860989996e-05,
+      "loss": 0.5397,
+      "step": 25681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356501667094127e-05,
+      "loss": 0.3434,
+      "step": 25682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356437548089254e-05,
+      "loss": 0.5962,
+      "step": 25683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3563734290843804e-05,
+      "loss": 0.5864,
+      "step": 25684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3563093100795076e-05,
+      "loss": 0.5534,
+      "step": 25685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356245191074635e-05,
+      "loss": 0.4344,
+      "step": 25686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356181072069762e-05,
+      "loss": 0.4641,
+      "step": 25687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3561169530648884e-05,
+      "loss": 0.4478,
+      "step": 25688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3560528340600156e-05,
+      "loss": 0.416,
+      "step": 25689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355988715055142e-05,
+      "loss": 0.4255,
+      "step": 25690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355924596050269e-05,
+      "loss": 0.6253,
+      "step": 25691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3558604770453964e-05,
+      "loss": 0.4748,
+      "step": 25692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3557963580405236e-05,
+      "loss": 0.5903,
+      "step": 25693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355732239035651e-05,
+      "loss": 0.4745,
+      "step": 25694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355668120030777e-05,
+      "loss": 0.4686,
+      "step": 25695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3556040010259044e-05,
+      "loss": 0.4724,
+      "step": 25696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355539882021031e-05,
+      "loss": 0.5054,
+      "step": 25697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355475763016158e-05,
+      "loss": 0.5971,
+      "step": 25698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355411644011285e-05,
+      "loss": 0.4881,
+      "step": 25699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355347525006412e-05,
+      "loss": 0.4935,
+      "step": 25700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355283406001539e-05,
+      "loss": 0.4356,
+      "step": 25701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355219286996666e-05,
+      "loss": 0.5889,
+      "step": 25702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355155167991793e-05,
+      "loss": 0.4373,
+      "step": 25703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35509104898692e-05,
+      "loss": 0.4762,
+      "step": 25704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355026929982047e-05,
+      "loss": 0.433,
+      "step": 25705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3549628109771734e-05,
+      "loss": 0.5177,
+      "step": 25706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3548986919723006e-05,
+      "loss": 0.4501,
+      "step": 25707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354834572967428e-05,
+      "loss": 0.3506,
+      "step": 25708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354770453962555e-05,
+      "loss": 0.4489,
+      "step": 25709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354706334957682e-05,
+      "loss": 0.4655,
+      "step": 25710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3546422159528086e-05,
+      "loss": 0.469,
+      "step": 25711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354578096947936e-05,
+      "loss": 0.5538,
+      "step": 25712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354513977943062e-05,
+      "loss": 0.5487,
+      "step": 25713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3544498589381894e-05,
+      "loss": 0.5362,
+      "step": 25714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354385739933316e-05,
+      "loss": 0.4339,
+      "step": 25715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354321620928443e-05,
+      "loss": 0.608,
+      "step": 25716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35425750192357e-05,
+      "loss": 0.4423,
+      "step": 25717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3541933829186975e-05,
+      "loss": 0.5302,
+      "step": 25718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3541292639138246e-05,
+      "loss": 0.362,
+      "step": 25719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354065144908951e-05,
+      "loss": 0.5096,
+      "step": 25720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354001025904078e-05,
+      "loss": 0.5915,
+      "step": 25721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353936906899205e-05,
+      "loss": 0.5002,
+      "step": 25722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353872787894332e-05,
+      "loss": 0.3756,
+      "step": 25723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3538086688894584e-05,
+      "loss": 0.4433,
+      "step": 25724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353744549884586e-05,
+      "loss": 0.5534,
+      "step": 25725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3536804308797135e-05,
+      "loss": 0.4215,
+      "step": 25726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35361631187484e-05,
+      "loss": 0.4579,
+      "step": 25727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353552192869967e-05,
+      "loss": 0.6385,
+      "step": 25728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3534880738650936e-05,
+      "loss": 0.5129,
+      "step": 25729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353423954860221e-05,
+      "loss": 0.4702,
+      "step": 25730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353359835855347e-05,
+      "loss": 0.4543,
+      "step": 25731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3532957168504745e-05,
+      "loss": 0.4816,
+      "step": 25732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3532315978456016e-05,
+      "loss": 0.4858,
+      "step": 25733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353167478840729e-05,
+      "loss": 0.3748,
+      "step": 25734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353103359835856e-05,
+      "loss": 0.5017,
+      "step": 25735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3530392408309825e-05,
+      "loss": 0.5205,
+      "step": 25736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3529751218261096e-05,
+      "loss": 0.5383,
+      "step": 25737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352911002821236e-05,
+      "loss": 0.4951,
+      "step": 25738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352846883816363e-05,
+      "loss": 0.4744,
+      "step": 25739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35278276481149e-05,
+      "loss": 0.4925,
+      "step": 25740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3527186458066177e-05,
+      "loss": 0.4506,
+      "step": 25741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352654526801744e-05,
+      "loss": 0.4568,
+      "step": 25742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352590407796871e-05,
+      "loss": 0.5376,
+      "step": 25743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3525262887919985e-05,
+      "loss": 0.5058,
+      "step": 25744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352462169787125e-05,
+      "loss": 0.4325,
+      "step": 25745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352398050782252e-05,
+      "loss": 0.4437,
+      "step": 25746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3523339317773787e-05,
+      "loss": 0.4284,
+      "step": 25747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352269812772506e-05,
+      "loss": 0.4806,
+      "step": 25748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352205693767633e-05,
+      "loss": 0.471,
+      "step": 25749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35214157476276e-05,
+      "loss": 0.4651,
+      "step": 25750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352077455757887e-05,
+      "loss": 0.4657,
+      "step": 25751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352013336753014e-05,
+      "loss": 0.4779,
+      "step": 25752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351949217748141e-05,
+      "loss": 0.5089,
+      "step": 25753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3518850987432675e-05,
+      "loss": 0.5289,
+      "step": 25754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351820979738395e-05,
+      "loss": 0.5819,
+      "step": 25755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351756860733521e-05,
+      "loss": 0.433,
+      "step": 25756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351692741728648e-05,
+      "loss": 0.4857,
+      "step": 25757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3516286227237755e-05,
+      "loss": 0.3904,
+      "step": 25758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351564503718903e-05,
+      "loss": 0.59,
+      "step": 25759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351500384714029e-05,
+      "loss": 0.483,
+      "step": 25760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3514362657091563e-05,
+      "loss": 0.478,
+      "step": 25761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3513721467042835e-05,
+      "loss": 0.5479,
+      "step": 25762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35130802769941e-05,
+      "loss": 0.4812,
+      "step": 25763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351243908694537e-05,
+      "loss": 0.4724,
+      "step": 25764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3511797896896644e-05,
+      "loss": 0.4387,
+      "step": 25765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3511156706847915e-05,
+      "loss": 0.5653,
+      "step": 25766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351051551679918e-05,
+      "loss": 0.3517,
+      "step": 25767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350987432675045e-05,
+      "loss": 0.4906,
+      "step": 25768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350923313670172e-05,
+      "loss": 0.597,
+      "step": 25769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350859194665299e-05,
+      "loss": 0.4415,
+      "step": 25770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350795075660426e-05,
+      "loss": 0.5182,
+      "step": 25771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3507309566555525e-05,
+      "loss": 0.4243,
+      "step": 25772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35066683765068e-05,
+      "loss": 0.7354,
+      "step": 25773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350602718645807e-05,
+      "loss": 0.4355,
+      "step": 25774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350538599640934e-05,
+      "loss": 0.5678,
+      "step": 25775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3504744806360605e-05,
+      "loss": 0.5036,
+      "step": 25776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350410361631188e-05,
+      "loss": 0.5559,
+      "step": 25777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350346242626314e-05,
+      "loss": 0.5225,
+      "step": 25778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3502821236214414e-05,
+      "loss": 0.5115,
+      "step": 25779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3502180046165685e-05,
+      "loss": 0.4368,
+      "step": 25780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350153885611696e-05,
+      "loss": 0.5701,
+      "step": 25781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350089766606823e-05,
+      "loss": 0.5063,
+      "step": 25782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3500256476019494e-05,
+      "loss": 0.4685,
+      "step": 25783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3499615285970766e-05,
+      "loss": 0.4487,
+      "step": 25784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349897409592203e-05,
+      "loss": 0.6072,
+      "step": 25785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.34983329058733e-05,
+      "loss": 0.4761,
+      "step": 25786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349769171582457e-05,
+      "loss": 0.4039,
+      "step": 25787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349705052577584e-05,
+      "loss": 0.5086,
+      "step": 25788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349640933572711e-05,
+      "loss": 0.5269,
+      "step": 25789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349576814567838e-05,
+      "loss": 0.4409,
+      "step": 25790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3495126955629654e-05,
+      "loss": 0.5412,
+      "step": 25791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349448576558092e-05,
+      "loss": 0.4749,
+      "step": 25792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349384457553219e-05,
+      "loss": 0.4837,
+      "step": 25793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3493203385483456e-05,
+      "loss": 0.6548,
+      "step": 25794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349256219543473e-05,
+      "loss": 0.4855,
+      "step": 25795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349192100538599e-05,
+      "loss": 0.4757,
+      "step": 25796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349127981533727e-05,
+      "loss": 0.336,
+      "step": 25797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349063862528854e-05,
+      "loss": 0.4695,
+      "step": 25798
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348999743523981e-05,
+      "loss": 0.4756,
+      "step": 25799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348935624519108e-05,
+      "loss": 0.5018,
+      "step": 25800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3488715055142344e-05,
+      "loss": 0.546,
+      "step": 25801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3488073865093616e-05,
+      "loss": 0.589,
+      "step": 25802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348743267504488e-05,
+      "loss": 0.4939,
+      "step": 25803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348679148499615e-05,
+      "loss": 0.5122,
+      "step": 25804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3486150294947424e-05,
+      "loss": 0.4831,
+      "step": 25805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3485509104898696e-05,
+      "loss": 0.4299,
+      "step": 25806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348486791484997e-05,
+      "loss": 0.5469,
+      "step": 25807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348422672480123e-05,
+      "loss": 0.4938,
+      "step": 25808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3483585534752504e-05,
+      "loss": 0.5737,
+      "step": 25809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348294434470377e-05,
+      "loss": 0.4388,
+      "step": 25810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348230315465504e-05,
+      "loss": 0.5324,
+      "step": 25811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3481661964606306e-05,
+      "loss": 0.4455,
+      "step": 25812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3481020774557584e-05,
+      "loss": 0.4086,
+      "step": 25813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.348037958450885e-05,
+      "loss": 0.3942,
+      "step": 25814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347973839446012e-05,
+      "loss": 0.4733,
+      "step": 25815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347909720441139e-05,
+      "loss": 0.5041,
+      "step": 25816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347845601436266e-05,
+      "loss": 0.6219,
+      "step": 25817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347781482431393e-05,
+      "loss": 0.336,
+      "step": 25818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3477173634265194e-05,
+      "loss": 0.4554,
+      "step": 25819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3476532444216466e-05,
+      "loss": 0.5,
+      "step": 25820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347589125416774e-05,
+      "loss": 0.595,
+      "step": 25821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347525006411901e-05,
+      "loss": 0.4788,
+      "step": 25822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3474608874070274e-05,
+      "loss": 0.5409,
+      "step": 25823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3473967684021546e-05,
+      "loss": 0.4085,
+      "step": 25824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347332649397282e-05,
+      "loss": 0.5516,
+      "step": 25825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347268530392408e-05,
+      "loss": 0.4061,
+      "step": 25826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3472044113875354e-05,
+      "loss": 0.5435,
+      "step": 25827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347140292382662e-05,
+      "loss": 0.499,
+      "step": 25828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347076173377789e-05,
+      "loss": 0.3684,
+      "step": 25829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.347012054372916e-05,
+      "loss": 0.5068,
+      "step": 25830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3469479353680435e-05,
+      "loss": 0.4673,
+      "step": 25831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.34688381636317e-05,
+      "loss": 0.4204,
+      "step": 25832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346819697358297e-05,
+      "loss": 0.4926,
+      "step": 25833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346755578353424e-05,
+      "loss": 0.517,
+      "step": 25834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346691459348551e-05,
+      "loss": 0.4076,
+      "step": 25835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346627340343678e-05,
+      "loss": 0.3819,
+      "step": 25836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346563221338805e-05,
+      "loss": 0.4544,
+      "step": 25837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346499102333932e-05,
+      "loss": 0.3871,
+      "step": 25838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346434983329059e-05,
+      "loss": 0.5594,
+      "step": 25839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346370864324186e-05,
+      "loss": 0.281,
+      "step": 25840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3463067453193125e-05,
+      "loss": 0.4091,
+      "step": 25841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3462426263144396e-05,
+      "loss": 0.4626,
+      "step": 25842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346178507309567e-05,
+      "loss": 0.4323,
+      "step": 25843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346114388304693e-05,
+      "loss": 0.5678,
+      "step": 25844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3460502692998205e-05,
+      "loss": 0.6113,
+      "step": 25845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3459861502949476e-05,
+      "loss": 0.4142,
+      "step": 25846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345922031290075e-05,
+      "loss": 0.4645,
+      "step": 25847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345857912285201e-05,
+      "loss": 0.4012,
+      "step": 25848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3457937932803285e-05,
+      "loss": 0.4823,
+      "step": 25849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345729674275455e-05,
+      "loss": 0.6323,
+      "step": 25850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345665555270582e-05,
+      "loss": 0.4429,
+      "step": 25851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345601436265709e-05,
+      "loss": 0.4456,
+      "step": 25852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3455373172608365e-05,
+      "loss": 0.4809,
+      "step": 25853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3454731982559637e-05,
+      "loss": 0.3457,
+      "step": 25854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.34540907925109e-05,
+      "loss": 0.5741,
+      "step": 25855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345344960246217e-05,
+      "loss": 0.3799,
+      "step": 25856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345280841241344e-05,
+      "loss": 0.5749,
+      "step": 25857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345216722236471e-05,
+      "loss": 0.4459,
+      "step": 25858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3451526032315975e-05,
+      "loss": 0.4078,
+      "step": 25859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3450884842267247e-05,
+      "loss": 0.5159,
+      "step": 25860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.345024365221852e-05,
+      "loss": 0.4133,
+      "step": 25861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344960246216979e-05,
+      "loss": 0.4483,
+      "step": 25862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344896127212106e-05,
+      "loss": 0.5526,
+      "step": 25863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344832008207233e-05,
+      "loss": 0.3414,
+      "step": 25864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.34476788920236e-05,
+      "loss": 0.4045,
+      "step": 25865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344703770197486e-05,
+      "loss": 0.498,
+      "step": 25866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3446396511926135e-05,
+      "loss": 0.4104,
+      "step": 25867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.34457553218774e-05,
+      "loss": 0.5295,
+      "step": 25868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344511413182868e-05,
+      "loss": 0.3848,
+      "step": 25869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344447294177995e-05,
+      "loss": 0.4527,
+      "step": 25870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3443831751731215e-05,
+      "loss": 0.3998,
+      "step": 25871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344319056168249e-05,
+      "loss": 0.5074,
+      "step": 25872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344254937163375e-05,
+      "loss": 0.5617,
+      "step": 25873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3441908181585024e-05,
+      "loss": 0.5039,
+      "step": 25874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344126699153629e-05,
+      "loss": 0.4015,
+      "step": 25875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.344062580148756e-05,
+      "loss": 0.4781,
+      "step": 25876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.343998461143883e-05,
+      "loss": 0.6236,
+      "step": 25877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3439343421390104e-05,
+      "loss": 0.4664,
+      "step": 25878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3438702231341375e-05,
+      "loss": 0.5582,
+      "step": 25879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343806104129264e-05,
+      "loss": 0.4997,
+      "step": 25880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343741985124391e-05,
+      "loss": 0.4522,
+      "step": 25881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343677866119518e-05,
+      "loss": 0.4954,
+      "step": 25882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343613747114645e-05,
+      "loss": 0.3097,
+      "step": 25883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3435496281097714e-05,
+      "loss": 0.5275,
+      "step": 25884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343485509104899e-05,
+      "loss": 0.4886,
+      "step": 25885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3434213901000264e-05,
+      "loss": 0.4949,
+      "step": 25886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343357271095153e-05,
+      "loss": 0.4406,
+      "step": 25887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.34329315209028e-05,
+      "loss": 0.5978,
+      "step": 25888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3432290330854065e-05,
+      "loss": 0.5106,
+      "step": 25889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343164914080534e-05,
+      "loss": 0.4331,
+      "step": 25890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.34310079507566e-05,
+      "loss": 0.4236,
+      "step": 25891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3430366760707874e-05,
+      "loss": 0.4545,
+      "step": 25892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3429725570659145e-05,
+      "loss": 0.4572,
+      "step": 25893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342908438061042e-05,
+      "loss": 0.5276,
+      "step": 25894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342844319056169e-05,
+      "loss": 0.4715,
+      "step": 25895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3427802000512954e-05,
+      "loss": 0.4739,
+      "step": 25896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3427160810464226e-05,
+      "loss": 0.4971,
+      "step": 25897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342651962041549e-05,
+      "loss": 0.4111,
+      "step": 25898
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342587843036676e-05,
+      "loss": 0.5114,
+      "step": 25899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342523724031803e-05,
+      "loss": 0.5077,
+      "step": 25900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3424596050269306e-05,
+      "loss": 0.353,
+      "step": 25901
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342395486022057e-05,
+      "loss": 0.5426,
+      "step": 25902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342331367017184e-05,
+      "loss": 0.4349,
+      "step": 25903
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3422672480123114e-05,
+      "loss": 0.5108,
+      "step": 25904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342203129007438e-05,
+      "loss": 0.4587,
+      "step": 25905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342139010002565e-05,
+      "loss": 0.5464,
+      "step": 25906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3420748909976916e-05,
+      "loss": 0.5012,
+      "step": 25907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342010771992819e-05,
+      "loss": 0.4069,
+      "step": 25908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341946652987946e-05,
+      "loss": 0.4376,
+      "step": 25909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341882533983073e-05,
+      "loss": 0.4301,
+      "step": 25910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3418184149781996e-05,
+      "loss": 0.4309,
+      "step": 25911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341754295973327e-05,
+      "loss": 0.4733,
+      "step": 25912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341690176968454e-05,
+      "loss": 0.4759,
+      "step": 25913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3416260579635804e-05,
+      "loss": 0.4754,
+      "step": 25914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3415619389587076e-05,
+      "loss": 0.4511,
+      "step": 25915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341497819953834e-05,
+      "loss": 0.5945,
+      "step": 25916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341433700948961e-05,
+      "loss": 0.4678,
+      "step": 25917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3413695819440884e-05,
+      "loss": 0.4646,
+      "step": 25918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3413054629392156e-05,
+      "loss": 0.5248,
+      "step": 25919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341241343934342e-05,
+      "loss": 0.5231,
+      "step": 25920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341177224929469e-05,
+      "loss": 0.436,
+      "step": 25921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3411131059245964e-05,
+      "loss": 0.4001,
+      "step": 25922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341048986919723e-05,
+      "loss": 0.4447,
+      "step": 25923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.34098486791485e-05,
+      "loss": 0.5235,
+      "step": 25924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340920748909977e-05,
+      "loss": 0.3627,
+      "step": 25925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3408566299051044e-05,
+      "loss": 0.4043,
+      "step": 25926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340792510900231e-05,
+      "loss": 0.4972,
+      "step": 25927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340728391895358e-05,
+      "loss": 0.5997,
+      "step": 25928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3406642728904846e-05,
+      "loss": 0.5966,
+      "step": 25929
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340600153885612e-05,
+      "loss": 0.5955,
+      "step": 25930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340536034880739e-05,
+      "loss": 0.5346,
+      "step": 25931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3404719158758654e-05,
+      "loss": 0.4382,
+      "step": 25932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3404077968709926e-05,
+      "loss": 0.4973,
+      "step": 25933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.34034367786612e-05,
+      "loss": 0.5414,
+      "step": 25934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340279558861247e-05,
+      "loss": 0.584,
+      "step": 25935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3402154398563734e-05,
+      "loss": 0.5359,
+      "step": 25936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3401513208515006e-05,
+      "loss": 0.414,
+      "step": 25937
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340087201846627e-05,
+      "loss": 0.5887,
+      "step": 25938
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340023082841754e-05,
+      "loss": 0.4098,
+      "step": 25939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3399589638368814e-05,
+      "loss": 0.5431,
+      "step": 25940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3398948448320086e-05,
+      "loss": 0.4309,
+      "step": 25941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339830725827136e-05,
+      "loss": 0.3694,
+      "step": 25942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339766606822262e-05,
+      "loss": 0.5393,
+      "step": 25943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3397024878173895e-05,
+      "loss": 0.541,
+      "step": 25944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339638368812516e-05,
+      "loss": 0.4894,
+      "step": 25945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339574249807643e-05,
+      "loss": 0.4596,
+      "step": 25946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3395101308027696e-05,
+      "loss": 0.4026,
+      "step": 25947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339446011797897e-05,
+      "loss": 0.4658,
+      "step": 25948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339381892793024e-05,
+      "loss": 0.4899,
+      "step": 25949
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339317773788151e-05,
+      "loss": 0.5362,
+      "step": 25950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339253654783278e-05,
+      "loss": 0.5493,
+      "step": 25951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339189535778405e-05,
+      "loss": 0.5043,
+      "step": 25952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339125416773532e-05,
+      "loss": 0.3974,
+      "step": 25953
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3390612977686585e-05,
+      "loss": 0.4951,
+      "step": 25954
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3389971787637856e-05,
+      "loss": 0.4489,
+      "step": 25955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338933059758912e-05,
+      "loss": 0.4151,
+      "step": 25956
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.33886894075404e-05,
+      "loss": 0.6462,
+      "step": 25957
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338804821749167e-05,
+      "loss": 0.4915,
+      "step": 25958
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3387407027442936e-05,
+      "loss": 0.5255,
+      "step": 25959
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338676583739421e-05,
+      "loss": 0.3522,
+      "step": 25960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338612464734547e-05,
+      "loss": 0.4539,
+      "step": 25961
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3385483457296745e-05,
+      "loss": 0.4514,
+      "step": 25962
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338484226724801e-05,
+      "loss": 0.4128,
+      "step": 25963
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338420107719928e-05,
+      "loss": 0.3999,
+      "step": 25964
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338355988715055e-05,
+      "loss": 0.4969,
+      "step": 25965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3382918697101825e-05,
+      "loss": 0.6494,
+      "step": 25966
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.33822775070531e-05,
+      "loss": 0.5017,
+      "step": 25967
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338163631700436e-05,
+      "loss": 0.5236,
+      "step": 25968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338099512695563e-05,
+      "loss": 0.5284,
+      "step": 25969
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.33803539369069e-05,
+      "loss": 0.4898,
+      "step": 25970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337971274685817e-05,
+      "loss": 0.4122,
+      "step": 25971
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3379071556809435e-05,
+      "loss": 0.4746,
+      "step": 25972
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337843036676071e-05,
+      "loss": 0.3761,
+      "step": 25973
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337778917671198e-05,
+      "loss": 0.4609,
+      "step": 25974
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337714798666325e-05,
+      "loss": 0.4074,
+      "step": 25975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337650679661452e-05,
+      "loss": 0.4322,
+      "step": 25976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337586560656579e-05,
+      "loss": 0.4587,
+      "step": 25977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337522441651706e-05,
+      "loss": 0.445,
+      "step": 25978
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337458322646832e-05,
+      "loss": 0.4328,
+      "step": 25979
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3373942036419595e-05,
+      "loss": 0.4853,
+      "step": 25980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337330084637087e-05,
+      "loss": 0.4386,
+      "step": 25981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337265965632214e-05,
+      "loss": 0.4172,
+      "step": 25982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3372018466273403e-05,
+      "loss": 0.4833,
+      "step": 25983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3371377276224675e-05,
+      "loss": 0.4601,
+      "step": 25984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337073608617595e-05,
+      "loss": 0.4269,
+      "step": 25985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337009489612721e-05,
+      "loss": 0.4914,
+      "step": 25986
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3369453706078484e-05,
+      "loss": 0.473,
+      "step": 25987
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336881251602975e-05,
+      "loss": 0.4426,
+      "step": 25988
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336817132598102e-05,
+      "loss": 0.5226,
+      "step": 25989
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336753013593229e-05,
+      "loss": 0.5545,
+      "step": 25990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3366888945883564e-05,
+      "loss": 0.5614,
+      "step": 25991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336624775583483e-05,
+      "loss": 0.3646,
+      "step": 25992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.33656065657861e-05,
+      "loss": 0.432,
+      "step": 25993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336496537573737e-05,
+      "loss": 0.4132,
+      "step": 25994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336432418568864e-05,
+      "loss": 0.4139,
+      "step": 25995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336368299563991e-05,
+      "loss": 0.5048,
+      "step": 25996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336304180559118e-05,
+      "loss": 0.3742,
+      "step": 25997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336240061554245e-05,
+      "loss": 0.4237,
+      "step": 25998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336175942549372e-05,
+      "loss": 0.44,
+      "step": 25999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336111823544499e-05,
+      "loss": 0.4601,
+      "step": 26000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3360477045396254e-05,
+      "loss": 0.4061,
+      "step": 26001
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3359835855347525e-05,
+      "loss": 0.4697,
+      "step": 26002
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.33591946652988e-05,
+      "loss": 0.605,
+      "step": 26003
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.335855347525006e-05,
+      "loss": 0.4035,
+      "step": 26004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3357912285201334e-05,
+      "loss": 0.4186,
+      "step": 26005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3357271095152605e-05,
+      "loss": 0.415,
+      "step": 26006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.335662990510388e-05,
+      "loss": 0.4975,
+      "step": 26007
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.335598871505514e-05,
+      "loss": 0.4009,
+      "step": 26008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3355347525006414e-05,
+      "loss": 0.3853,
+      "step": 26009
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.335470633495768e-05,
+      "loss": 0.4749,
+      "step": 26010
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7866152119742694,
+      "eval_loss": 0.45247501134872437,
+      "eval_runtime": 262.4569,
+      "eval_samples_per_second": 1585.628,
+      "eval_steps_per_second": 24.777,
+      "step": 26010
+    }
+  ],
+  "max_steps": 78030,
+  "num_train_epochs": 3,
+  "total_flos": 1.422421861136034e+16,
+  "trial_name": null,
+  "trial_params": null
+}