diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31339 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999856314957613,
+  "global_step": 5219,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2738853503184715e-07,
+      "loss": 1.2808,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.547770700636943e-07,
+      "loss": 1.3333,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8216560509554143e-07,
+      "loss": 1.3078,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.095541401273886e-07,
+      "loss": 1.2553,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.369426751592357e-07,
+      "loss": 1.3083,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.643312101910829e-07,
+      "loss": 1.335,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.9171974522293e-07,
+      "loss": 1.2872,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0191082802547772e-06,
+      "loss": 1.2884,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1464968152866242e-06,
+      "loss": 1.3453,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2738853503184715e-06,
+      "loss": 1.3292,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4012738853503185e-06,
+      "loss": 1.2684,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5286624203821657e-06,
+      "loss": 1.2826,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6560509554140127e-06,
+      "loss": 0.3135,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.78343949044586e-06,
+      "loss": 1.2601,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9108280254777074e-06,
+      "loss": 1.2056,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0382165605095544e-06,
+      "loss": 1.1862,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1656050955414015e-06,
+      "loss": 1.1818,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2929936305732485e-06,
+      "loss": 1.1374,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.420382165605096e-06,
+      "loss": 1.1287,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.547770700636943e-06,
+      "loss": 1.0876,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.67515923566879e-06,
+      "loss": 1.1102,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.802547770700637e-06,
+      "loss": 1.1536,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9299363057324844e-06,
+      "loss": 1.1058,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0573248407643314e-06,
+      "loss": 1.0721,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1847133757961785e-06,
+      "loss": 1.1321,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3121019108280255e-06,
+      "loss": 1.0965,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4394904458598725e-06,
+      "loss": 0.9958,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.56687898089172e-06,
+      "loss": 0.9759,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.694267515923567e-06,
+      "loss": 1.0873,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.821656050955415e-06,
+      "loss": 1.0365,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.949044585987262e-06,
+      "loss": 1.0582,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.076433121019109e-06,
+      "loss": 1.0943,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.203821656050956e-06,
+      "loss": 0.3268,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.331210191082803e-06,
+      "loss": 1.0165,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.45859872611465e-06,
+      "loss": 0.968,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.585987261146497e-06,
+      "loss": 1.0304,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.713375796178344e-06,
+      "loss": 0.2808,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.840764331210192e-06,
+      "loss": 1.0763,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.968152866242039e-06,
+      "loss": 0.9797,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.095541401273886e-06,
+      "loss": 0.9748,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.222929936305733e-06,
+      "loss": 1.0653,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.35031847133758e-06,
+      "loss": 1.0011,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.477707006369427e-06,
+      "loss": 1.0659,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.605095541401274e-06,
+      "loss": 1.0166,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.732484076433121e-06,
+      "loss": 1.0238,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.859872611464969e-06,
+      "loss": 1.059,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.987261146496816e-06,
+      "loss": 0.966,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.114649681528663e-06,
+      "loss": 1.0102,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.24203821656051e-06,
+      "loss": 1.0243,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.369426751592357e-06,
+      "loss": 1.0318,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.496815286624204e-06,
+      "loss": 0.9754,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.624203821656051e-06,
+      "loss": 1.0001,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.751592356687898e-06,
+      "loss": 0.9666,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.878980891719745e-06,
+      "loss": 0.9815,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.006369426751593e-06,
+      "loss": 1.0228,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.13375796178344e-06,
+      "loss": 0.9763,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.261146496815287e-06,
+      "loss": 0.943,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.388535031847134e-06,
+      "loss": 1.0249,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.515923566878982e-06,
+      "loss": 1.0121,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.64331210191083e-06,
+      "loss": 1.0003,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.770700636942676e-06,
+      "loss": 0.932,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.898089171974524e-06,
+      "loss": 0.9542,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.02547770700637e-06,
+      "loss": 0.961,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.152866242038218e-06,
+      "loss": 0.9643,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.280254777070064e-06,
+      "loss": 0.304,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.407643312101912e-06,
+      "loss": 0.958,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.53503184713376e-06,
+      "loss": 0.9869,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.662420382165606e-06,
+      "loss": 1.0115,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.789808917197454e-06,
+      "loss": 0.9746,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.9171974522293e-06,
+      "loss": 0.9381,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.044585987261148e-06,
+      "loss": 0.924,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.171974522292994e-06,
+      "loss": 1.0288,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.299363057324842e-06,
+      "loss": 0.9652,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.426751592356688e-06,
+      "loss": 0.9762,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.554140127388536e-06,
+      "loss": 0.3177,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.681528662420384e-06,
+      "loss": 0.9976,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.80891719745223e-06,
+      "loss": 0.9081,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.936305732484078e-06,
+      "loss": 0.9912,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0063694267515924e-05,
+      "loss": 0.946,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0191082802547772e-05,
+      "loss": 1.0079,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0318471337579618e-05,
+      "loss": 0.3101,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0445859872611466e-05,
+      "loss": 0.9446,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0573248407643314e-05,
+      "loss": 0.9693,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.070063694267516e-05,
+      "loss": 0.9738,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0828025477707008e-05,
+      "loss": 0.9677,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0955414012738854e-05,
+      "loss": 1.0497,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1082802547770702e-05,
+      "loss": 0.9811,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1210191082802548e-05,
+      "loss": 0.8994,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1337579617834396e-05,
+      "loss": 0.9785,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1464968152866242e-05,
+      "loss": 0.9173,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.159235668789809e-05,
+      "loss": 0.9631,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1719745222929938e-05,
+      "loss": 0.9429,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1847133757961784e-05,
+      "loss": 0.9615,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1974522292993632e-05,
+      "loss": 0.9135,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2101910828025478e-05,
+      "loss": 0.9496,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2229299363057326e-05,
+      "loss": 0.9072,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2356687898089172e-05,
+      "loss": 0.981,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.248407643312102e-05,
+      "loss": 0.8827,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2611464968152866e-05,
+      "loss": 0.9132,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2738853503184714e-05,
+      "loss": 0.9653,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2866242038216562e-05,
+      "loss": 0.9279,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2993630573248408e-05,
+      "loss": 0.9583,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3121019108280256e-05,
+      "loss": 0.9264,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3248407643312102e-05,
+      "loss": 0.9723,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.337579617834395e-05,
+      "loss": 0.9345,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3503184713375796e-05,
+      "loss": 0.2675,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3630573248407644e-05,
+      "loss": 0.9975,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.375796178343949e-05,
+      "loss": 0.9542,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3885350318471338e-05,
+      "loss": 0.9253,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4012738853503186e-05,
+      "loss": 0.8882,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4140127388535032e-05,
+      "loss": 0.9109,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.426751592356688e-05,
+      "loss": 0.9529,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4394904458598726e-05,
+      "loss": 0.9331,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4522292993630574e-05,
+      "loss": 0.9854,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.464968152866242e-05,
+      "loss": 1.0297,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4777070063694268e-05,
+      "loss": 0.9979,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4904458598726114e-05,
+      "loss": 0.9575,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5031847133757964e-05,
+      "loss": 0.8792,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5159235668789811e-05,
+      "loss": 0.9674,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.528662420382166e-05,
+      "loss": 0.9271,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5414012738853506e-05,
+      "loss": 0.9641,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5541401273885352e-05,
+      "loss": 1.0081,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.56687898089172e-05,
+      "loss": 0.9904,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5796178343949047e-05,
+      "loss": 0.9221,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5923566878980894e-05,
+      "loss": 0.9261,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.605095541401274e-05,
+      "loss": 0.9137,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.617834394904459e-05,
+      "loss": 0.9133,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6305732484076436e-05,
+      "loss": 0.8348,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.643312101910828e-05,
+      "loss": 0.9276,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6560509554140128e-05,
+      "loss": 0.8997,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6687898089171977e-05,
+      "loss": 0.9424,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6815286624203824e-05,
+      "loss": 0.922,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.694267515923567e-05,
+      "loss": 0.8929,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.707006369426752e-05,
+      "loss": 0.9186,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7197452229299365e-05,
+      "loss": 0.9567,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.732484076433121e-05,
+      "loss": 0.9418,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7452229299363058e-05,
+      "loss": 0.8996,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7579617834394907e-05,
+      "loss": 0.9948,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7707006369426754e-05,
+      "loss": 0.9434,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.78343949044586e-05,
+      "loss": 0.9661,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.796178343949045e-05,
+      "loss": 0.9952,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8089171974522295e-05,
+      "loss": 0.3187,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.821656050955414e-05,
+      "loss": 0.9224,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8343949044585988e-05,
+      "loss": 0.9598,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8471337579617837e-05,
+      "loss": 0.9084,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8598726114649684e-05,
+      "loss": 0.2948,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.872611464968153e-05,
+      "loss": 0.9323,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8853503184713376e-05,
+      "loss": 0.8868,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8980891719745225e-05,
+      "loss": 0.8772,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.910828025477707e-05,
+      "loss": 0.968,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9235668789808918e-05,
+      "loss": 0.9372,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9363057324840767e-05,
+      "loss": 0.9901,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9490445859872614e-05,
+      "loss": 0.9829,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.961783439490446e-05,
+      "loss": 0.9405,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9745222929936306e-05,
+      "loss": 0.9367,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9872611464968155e-05,
+      "loss": 0.9301,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.936,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998074136714e-05,
+      "loss": 0.9486,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992296547597e-05,
+      "loss": 0.9327,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999982667234872e-05,
+      "loss": 0.9293,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999969186202253e-05,
+      "loss": 0.9435,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999951853454925e-05,
+      "loss": 0.8977,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993066899957e-05,
+      "loss": 0.889,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999905632844348e-05,
+      "loss": 0.9431,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99998767449989e-05,
+      "loss": 0.9745,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999844005474355e-05,
+      "loss": 0.9646,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999807414283317e-05,
+      "loss": 0.3291,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999766971439886e-05,
+      "loss": 0.9422,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972267695964e-05,
+      "loss": 0.9421,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999674530859634e-05,
+      "loss": 0.9645,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999962253315842e-05,
+      "loss": 1.0279,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999566683876025e-05,
+      "loss": 1.0195,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999506983033953e-05,
+      "loss": 0.9161,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999443430655208e-05,
+      "loss": 0.9295,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999376026764264e-05,
+      "loss": 0.9754,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999304771387082e-05,
+      "loss": 0.9272,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999229664551113e-05,
+      "loss": 0.9213,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999915070628528e-05,
+      "loss": 0.8696,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999906789662e-05,
+      "loss": 0.9148,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999898123558717e-05,
+      "loss": 0.9496,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998890723220165e-05,
+      "loss": 0.9482,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999879635955385e-05,
+      "loss": 0.9426,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998698144624567e-05,
+      "loss": 0.3011,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998596078470153e-05,
+      "loss": 0.9517,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998490161129923e-05,
+      "loss": 0.9224,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999838039264466e-05,
+      "loss": 0.9284,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998266773056656e-05,
+      "loss": 0.2936,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998149302409673e-05,
+      "loss": 0.9972,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999802798074895e-05,
+      "loss": 0.9663,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997902808121225e-05,
+      "loss": 0.9534,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999777378457471e-05,
+      "loss": 1.0082,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997640910159097e-05,
+      "loss": 0.9425,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997504184925565e-05,
+      "loss": 0.9264,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999736360892678e-05,
+      "loss": 0.9695,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997219182216886e-05,
+      "loss": 0.9044,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997070904851518e-05,
+      "loss": 0.9581,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996918776887785e-05,
+      "loss": 1.0295,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996762798384278e-05,
+      "loss": 0.8775,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999660296940108e-05,
+      "loss": 1.009,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999643928999975e-05,
+      "loss": 0.3335,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999627176024334e-05,
+      "loss": 0.9786,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999610038019637e-05,
+      "loss": 0.932,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995925149924857e-05,
+      "loss": 0.9896,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995746069496287e-05,
+      "loss": 0.8967,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995563138979642e-05,
+      "loss": 0.9259,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995376358445383e-05,
+      "loss": 0.9076,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999518572796545e-05,
+      "loss": 1.0008,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999499124761327e-05,
+      "loss": 0.9507,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994792917463752e-05,
+      "loss": 0.9336,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994590737593288e-05,
+      "loss": 0.9531,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994384708079746e-05,
+      "loss": 0.9628,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999417482900249e-05,
+      "loss": 0.932,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999396110044236e-05,
+      "loss": 0.9372,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999374352248167e-05,
+      "loss": 0.9666,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993522095204238e-05,
+      "loss": 0.9309,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999329681869534e-05,
+      "loss": 0.9211,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993067693041754e-05,
+      "loss": 0.9253,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992834718331728e-05,
+      "loss": 0.9299,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992597894654997e-05,
+      "loss": 0.9154,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992357222102783e-05,
+      "loss": 0.9336,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992112700767787e-05,
+      "loss": 0.8964,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999186433074419e-05,
+      "loss": 0.3319,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991612112127653e-05,
+      "loss": 0.9125,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999135604501533e-05,
+      "loss": 0.9458,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999109612950585e-05,
+      "loss": 0.9302,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990832365699325e-05,
+      "loss": 0.3517,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990564753697346e-05,
+      "loss": 0.919,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990293293602993e-05,
+      "loss": 0.8665,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990017985520825e-05,
+      "loss": 1.01,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989738829556886e-05,
+      "loss": 0.9086,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998945582581869e-05,
+      "loss": 0.9422,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998916897441525e-05,
+      "loss": 0.9649,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988878275457057e-05,
+      "loss": 0.9966,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998858372905607e-05,
+      "loss": 0.8985,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988285335325746e-05,
+      "loss": 0.9567,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987983094381018e-05,
+      "loss": 0.3233,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987677006338302e-05,
+      "loss": 0.9746,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987367071315493e-05,
+      "loss": 0.3156,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987053289431966e-05,
+      "loss": 0.9913,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986735660808585e-05,
+      "loss": 1.0082,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986414185567694e-05,
+      "loss": 1.0044,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986088863833113e-05,
+      "loss": 0.9654,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985759695730148e-05,
+      "loss": 0.8713,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985426681385586e-05,
+      "loss": 0.9318,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985089820927693e-05,
+      "loss": 0.8832,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984749114486223e-05,
+      "loss": 0.9314,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99844045621924e-05,
+      "loss": 0.8988,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984056164178943e-05,
+      "loss": 0.9598,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998370392058004e-05,
+      "loss": 0.9589,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998334783153137e-05,
+      "loss": 0.9116,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982987897170084e-05,
+      "loss": 0.935,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982624117634824e-05,
+      "loss": 0.8893,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982256493065708e-05,
+      "loss": 0.9705,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998188502360433e-05,
+      "loss": 0.9541,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981509709393772e-05,
+      "loss": 0.9288,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981130550578595e-05,
+      "loss": 0.8691,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980747547304842e-05,
+      "loss": 0.9576,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980360699720033e-05,
+      "loss": 0.8873,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997997000797317e-05,
+      "loss": 0.9118,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979575472214745e-05,
+      "loss": 0.9573,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979177092596714e-05,
+      "loss": 0.9549,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997877486927252e-05,
+      "loss": 0.9364,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978368802397097e-05,
+      "loss": 0.9032,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977958892126846e-05,
+      "loss": 0.902,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977545138619655e-05,
+      "loss": 0.8418,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977127542034885e-05,
+      "loss": 0.913,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976706102533394e-05,
+      "loss": 0.9252,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976280820277496e-05,
+      "loss": 0.935,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975851695431006e-05,
+      "loss": 0.9031,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997541872815921e-05,
+      "loss": 0.9456,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974981918628877e-05,
+      "loss": 0.9675,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997454126700825e-05,
+      "loss": 0.882,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974096773467056e-05,
+      "loss": 0.9307,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973648438176502e-05,
+      "loss": 0.8898,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973196261309277e-05,
+      "loss": 0.9409,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997274024303955e-05,
+      "loss": 0.9191,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997228038354296e-05,
+      "loss": 0.9915,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971816682996633e-05,
+      "loss": 0.9111,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997134914157918e-05,
+      "loss": 0.9453,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997087775947068e-05,
+      "loss": 0.9188,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970402536852694e-05,
+      "loss": 0.9532,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996992347390827e-05,
+      "loss": 0.9132,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996944057082193e-05,
+      "loss": 0.9519,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968953827779674e-05,
+      "loss": 0.9854,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996846324496898e-05,
+      "loss": 0.9255,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996796882257881e-05,
+      "loss": 0.9299,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967470560799595e-05,
+      "loss": 0.8973,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966968459823263e-05,
+      "loss": 0.9531,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99664625198432e-05,
+      "loss": 0.3551,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996595274105429e-05,
+      "loss": 0.9993,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965439123652875e-05,
+      "loss": 0.8408,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996492166783679e-05,
+      "loss": 0.333,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996440037380535e-05,
+      "loss": 0.9584,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996387524175934e-05,
+      "loss": 0.9072,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963346271901023e-05,
+      "loss": 0.9146,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996281346443415e-05,
+      "loss": 0.9643,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962276819563934e-05,
+      "loss": 0.2999,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961736337497087e-05,
+      "loss": 0.977,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961192018441784e-05,
+      "loss": 0.937,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960643862607682e-05,
+      "loss": 0.9439,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996009187020591e-05,
+      "loss": 0.9135,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995953604144909e-05,
+      "loss": 0.9486,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958976376551303e-05,
+      "loss": 0.8768,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958412875728124e-05,
+      "loss": 0.9877,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957845539196594e-05,
+      "loss": 1.0084,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957274367175235e-05,
+      "loss": 0.9374,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995669935988405e-05,
+      "loss": 0.9898,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956120517544515e-05,
+      "loss": 0.9469,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955537840379585e-05,
+      "loss": 0.9033,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954951328613686e-05,
+      "loss": 0.9429,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954360982472733e-05,
+      "loss": 0.8879,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953766802184107e-05,
+      "loss": 0.9743,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995316878797667e-05,
+      "loss": 0.8553,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952566940080763e-05,
+      "loss": 0.9462,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951961258728204e-05,
+      "loss": 0.9989,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951351744152277e-05,
+      "loss": 0.9073,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950738396587757e-05,
+      "loss": 0.9228,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950121216270888e-05,
+      "loss": 0.8975,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949500203439387e-05,
+      "loss": 0.9547,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948875358332453e-05,
+      "loss": 0.8945,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948246681190763e-05,
+      "loss": 0.8997,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947614172256462e-05,
+      "loss": 0.9471,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946977831773172e-05,
+      "loss": 0.3598,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946337659986003e-05,
+      "loss": 0.9433,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994569365714153e-05,
+      "loss": 0.3381,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945045823487796e-05,
+      "loss": 0.9287,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944394159274332e-05,
+      "loss": 0.9143,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943738664752152e-05,
+      "loss": 0.9181,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943079340173722e-05,
+      "loss": 0.9479,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942416185793002e-05,
+      "loss": 0.9085,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994174920186542e-05,
+      "loss": 0.9278,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941078388647882e-05,
+      "loss": 0.9058,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994040374639876e-05,
+      "loss": 0.9388,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939725275377912e-05,
+      "loss": 0.9401,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993904297584667e-05,
+      "loss": 0.9268,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938356848067834e-05,
+      "loss": 0.9399,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937666892305677e-05,
+      "loss": 0.9124,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936973108825958e-05,
+      "loss": 0.8563,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99362754978959e-05,
+      "loss": 0.865,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935574059784207e-05,
+      "loss": 0.9139,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993486879476105e-05,
+      "loss": 0.955,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993415970309808e-05,
+      "loss": 0.9104,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993344678506842e-05,
+      "loss": 0.8579,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993273004094666e-05,
+      "loss": 0.9448,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993200947100888e-05,
+      "loss": 0.9091,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931285075532617e-05,
+      "loss": 0.9418,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930556854796893e-05,
+      "loss": 0.9002,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929824809082198e-05,
+      "loss": 0.9212,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992908893867049e-05,
+      "loss": 0.8479,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928349243845215e-05,
+      "loss": 0.9584,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927605724891276e-05,
+      "loss": 0.8921,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926858382095064e-05,
+      "loss": 0.9317,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926107215744423e-05,
+      "loss": 0.3518,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925352226128692e-05,
+      "loss": 0.8981,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992459341353867e-05,
+      "loss": 0.9068,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923830778266628e-05,
+      "loss": 0.9546,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923064320606315e-05,
+      "loss": 0.8937,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922294040852952e-05,
+      "loss": 0.9045,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921519939303225e-05,
+      "loss": 0.9518,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920742016255293e-05,
+      "loss": 0.8674,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99199602720088e-05,
+      "loss": 0.9727,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919174706864854e-05,
+      "loss": 0.9544,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991838532112602e-05,
+      "loss": 0.9708,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917592115096357e-05,
+      "loss": 0.8702,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991679508908139e-05,
+      "loss": 0.9496,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915994243388098e-05,
+      "loss": 0.9131,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915189578324957e-05,
+      "loss": 0.9018,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99143810942019e-05,
+      "loss": 0.3125,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913568791330334e-05,
+      "loss": 0.9647,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912752670023128e-05,
+      "loss": 0.3269,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911932730594637e-05,
+      "loss": 0.9091,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911108973360673e-05,
+      "loss": 0.9582,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910281398638533e-05,
+      "loss": 0.9381,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990945000674697e-05,
+      "loss": 0.9347,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908614798006215e-05,
+      "loss": 0.9514,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990777577273797e-05,
+      "loss": 0.9486,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99069329312654e-05,
+      "loss": 0.904,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990608627391315e-05,
+      "loss": 0.9566,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990523580100732e-05,
+      "loss": 0.9163,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904381512875502e-05,
+      "loss": 0.8795,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903523409846734e-05,
+      "loss": 0.8923,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990266149225154e-05,
+      "loss": 0.9333,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99017957604219e-05,
+      "loss": 0.866,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990092621469128e-05,
+      "loss": 0.8881,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900052855394597e-05,
+      "loss": 0.9208,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899175682868247e-05,
+      "loss": 0.8656,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898294697450096e-05,
+      "loss": 0.9517,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989740989947947e-05,
+      "loss": 0.8891,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896521289297177e-05,
+      "loss": 1.0017,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989562886724548e-05,
+      "loss": 0.8946,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894732633668115e-05,
+      "loss": 0.9068,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989383258891029e-05,
+      "loss": 0.9166,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892928733318673e-05,
+      "loss": 0.9332,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989202106724141e-05,
+      "loss": 0.9213,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891109591028103e-05,
+      "loss": 0.9072,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989019430502983e-05,
+      "loss": 0.3188,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988927520959914e-05,
+      "loss": 0.8988,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888352305090035e-05,
+      "loss": 0.8463,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887425591857998e-05,
+      "loss": 0.9195,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988649507025997e-05,
+      "loss": 0.9281,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885560740654363e-05,
+      "loss": 0.9379,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884622603401057e-05,
+      "loss": 0.9191,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883680658861396e-05,
+      "loss": 0.9622,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882734907398195e-05,
+      "loss": 0.969,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881785349375726e-05,
+      "loss": 0.9307,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880831985159737e-05,
+      "loss": 1.0092,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987987481511743e-05,
+      "loss": 0.9557,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987891383961749e-05,
+      "loss": 0.9071,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877949059030057e-05,
+      "loss": 0.9074,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876980473726735e-05,
+      "loss": 0.9325,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876008084080596e-05,
+      "loss": 0.895,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875031890466178e-05,
+      "loss": 0.9499,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987405189325949e-05,
+      "loss": 0.8957,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873068092837996e-05,
+      "loss": 0.9367,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872080489580626e-05,
+      "loss": 0.8805,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871089083867778e-05,
+      "loss": 0.8866,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987009387608132e-05,
+      "loss": 0.9427,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869094866604578e-05,
+      "loss": 0.8993,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868092055822336e-05,
+      "loss": 0.9356,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867085444120856e-05,
+      "loss": 0.9178,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866075031887857e-05,
+      "loss": 0.3202,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986506081951252e-05,
+      "loss": 0.9024,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986404280738549e-05,
+      "loss": 0.8822,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863020995898883e-05,
+      "loss": 0.9095,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986199538544627e-05,
+      "loss": 0.9039,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860965976422688e-05,
+      "loss": 0.967,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859932769224633e-05,
+      "loss": 0.8819,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858895764250073e-05,
+      "loss": 0.9906,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857854961898437e-05,
+      "loss": 0.9182,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856810362570607e-05,
+      "loss": 0.9358,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985576196666894e-05,
+      "loss": 0.8966,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854709774597244e-05,
+      "loss": 0.918,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853653786760797e-05,
+      "loss": 0.8698,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852594003566338e-05,
+      "loss": 0.2925,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851530425422064e-05,
+      "loss": 0.9207,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850463052737635e-05,
+      "loss": 0.9033,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984939188592418e-05,
+      "loss": 0.9068,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848316925394274e-05,
+      "loss": 0.3187,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847238171561972e-05,
+      "loss": 0.9173,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846155624842778e-05,
+      "loss": 0.9428,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845069285653655e-05,
+      "loss": 0.9499,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843979154413033e-05,
+      "loss": 0.8619,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842885231540805e-05,
+      "loss": 0.9636,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984178751745831e-05,
+      "loss": 0.9691,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984068601258837e-05,
+      "loss": 0.8153,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839580717355246e-05,
+      "loss": 0.9337,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838471632184674e-05,
+      "loss": 0.9382,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837358757503834e-05,
+      "loss": 0.8865,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836242093741383e-05,
+      "loss": 0.9512,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835121641327425e-05,
+      "loss": 0.8537,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833997400693532e-05,
+      "loss": 0.8735,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832869372272728e-05,
+      "loss": 0.9371,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831737556499498e-05,
+      "loss": 0.2988,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830601953809787e-05,
+      "loss": 0.9048,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829462564640995e-05,
+      "loss": 0.7987,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828319389431993e-05,
+      "loss": 0.9212,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827172428623092e-05,
+      "loss": 0.952,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982602168265607e-05,
+      "loss": 0.8868,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982486715197417e-05,
+      "loss": 0.3167,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823708837022078e-05,
+      "loss": 0.9299,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822546738245952e-05,
+      "loss": 0.9639,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821380856093393e-05,
+      "loss": 0.8911,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820211191013475e-05,
+      "loss": 0.9218,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819037743456714e-05,
+      "loss": 0.937,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817860513875095e-05,
+      "loss": 0.8579,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816679502722052e-05,
+      "loss": 0.9147,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815494710452476e-05,
+      "loss": 0.8411,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814306137522722e-05,
+      "loss": 0.8997,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813113784390596e-05,
+      "loss": 0.3109,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811917651515357e-05,
+      "loss": 0.885,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810717739357716e-05,
+      "loss": 0.9692,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809514048379858e-05,
+      "loss": 0.9431,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808306579045405e-05,
+      "loss": 0.325,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807095331819447e-05,
+      "loss": 0.8917,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805880307168514e-05,
+      "loss": 0.8557,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980466150556061e-05,
+      "loss": 0.8967,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803438927465178e-05,
+      "loss": 0.9197,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802212573353127e-05,
+      "loss": 0.8994,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800982443696806e-05,
+      "loss": 0.8577,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799748538970033e-05,
+      "loss": 0.926,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798510859648078e-05,
+      "loss": 0.8629,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797269406207653e-05,
+      "loss": 0.8456,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979602417912694e-05,
+      "loss": 0.9036,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794775178885563e-05,
+      "loss": 0.9132,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97935224059646e-05,
+      "loss": 0.9051,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979226586084659e-05,
+      "loss": 0.9347,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791005544015514e-05,
+      "loss": 0.8434,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789741455956816e-05,
+      "loss": 0.9368,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788473597157386e-05,
+      "loss": 0.93,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787201968105572e-05,
+      "loss": 0.3468,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785926569291167e-05,
+      "loss": 0.8969,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978464740120542e-05,
+      "loss": 0.9004,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978336446434103e-05,
+      "loss": 0.8207,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782077759192156e-05,
+      "loss": 0.9423,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780787286254398e-05,
+      "loss": 0.8552,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977949304602481e-05,
+      "loss": 0.9354,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778195039001896e-05,
+      "loss": 0.9379,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776893265685614e-05,
+      "loss": 0.901,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775587726577376e-05,
+      "loss": 0.8791,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774278422180036e-05,
+      "loss": 0.8753,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97729653529979e-05,
+      "loss": 0.917,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771648519536732e-05,
+      "loss": 0.9289,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770327922303734e-05,
+      "loss": 0.9079,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976900356180757e-05,
+      "loss": 0.9245,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976767543855834e-05,
+      "loss": 0.895,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766343553067613e-05,
+      "loss": 0.9837,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765007905848385e-05,
+      "loss": 0.9168,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976366849741511e-05,
+      "loss": 0.8101,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762325328283696e-05,
+      "loss": 0.9557,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760978398971494e-05,
+      "loss": 0.8708,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759627709997304e-05,
+      "loss": 0.9222,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758273261881374e-05,
+      "loss": 0.9144,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97569150551454e-05,
+      "loss": 0.9212,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975555309031253e-05,
+      "loss": 0.9178,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975418736790735e-05,
+      "loss": 0.9443,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752817888455902e-05,
+      "loss": 0.9116,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975144465248567e-05,
+      "loss": 0.8079,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750067660525595e-05,
+      "loss": 0.8682,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748686913106045e-05,
+      "loss": 0.9114,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747302410758848e-05,
+      "loss": 0.9191,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745914154017283e-05,
+      "loss": 0.3243,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974452214341607e-05,
+      "loss": 0.8777,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743126379491362e-05,
+      "loss": 0.8667,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974172686278078e-05,
+      "loss": 0.8679,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740323593823372e-05,
+      "loss": 0.9189,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738916573159642e-05,
+      "loss": 0.8975,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973750580133154e-05,
+      "loss": 0.9036,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973609127888245e-05,
+      "loss": 0.8903,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973467300635721e-05,
+      "loss": 0.9119,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97332509843021e-05,
+      "loss": 0.9015,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731825213264846e-05,
+      "loss": 0.9309,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730395693794614e-05,
+      "loss": 0.8498,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728962426442013e-05,
+      "loss": 0.8641,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727525411759103e-05,
+      "loss": 0.9108,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972608465029938e-05,
+      "loss": 0.898,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724640142617787e-05,
+      "loss": 0.8834,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972319188927071e-05,
+      "loss": 0.8886,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721739890815973e-05,
+      "loss": 0.8224,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972028414781285e-05,
+      "loss": 0.8429,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718824660822055e-05,
+      "loss": 0.8977,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717361430405733e-05,
+      "loss": 0.8886,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715894457127492e-05,
+      "loss": 0.8928,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714423741552363e-05,
+      "loss": 0.905,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712949284246828e-05,
+      "loss": 0.9225,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711471085778803e-05,
+      "loss": 0.8363,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970998914671766e-05,
+      "loss": 0.942,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970850346763419e-05,
+      "loss": 0.9257,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707014049100638e-05,
+      "loss": 0.9417,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705520891690697e-05,
+      "loss": 0.9491,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970402399597948e-05,
+      "loss": 0.9377,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702523362543552e-05,
+      "loss": 0.8922,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970101899196092e-05,
+      "loss": 0.9315,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969951088481102e-05,
+      "loss": 0.9599,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697999041674743e-05,
+      "loss": 0.8443,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696483463134404e-05,
+      "loss": 0.8992,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969496414977376e-05,
+      "loss": 0.8729,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693441102178014e-05,
+      "loss": 0.935,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96919143209338e-05,
+      "loss": 0.8892,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690383806629195e-05,
+      "loss": 0.9115,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688849559853702e-05,
+      "loss": 0.3094,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687311581198284e-05,
+      "loss": 0.8451,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685769871255317e-05,
+      "loss": 0.9365,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684224430618634e-05,
+      "loss": 0.9442,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968267525988349e-05,
+      "loss": 0.9021,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968112235964659e-05,
+      "loss": 0.8988,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679565730506064e-05,
+      "loss": 0.8953,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967800537306148e-05,
+      "loss": 0.9058,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676441287913853e-05,
+      "loss": 0.9303,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967487347566562e-05,
+      "loss": 0.8024,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673301936920663e-05,
+      "loss": 0.8637,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671726672284288e-05,
+      "loss": 0.916,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670147682363256e-05,
+      "loss": 0.935,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668564967765742e-05,
+      "loss": 0.8979,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666978529101366e-05,
+      "loss": 0.8825,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665388366981185e-05,
+      "loss": 0.9397,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663794482017678e-05,
+      "loss": 0.8938,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662196874824774e-05,
+      "loss": 0.8918,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660595546017823e-05,
+      "loss": 0.9367,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658990496213614e-05,
+      "loss": 0.894,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657381726030368e-05,
+      "loss": 0.9356,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965576923608774e-05,
+      "loss": 0.8951,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965415302700682e-05,
+      "loss": 0.8767,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652533099410123e-05,
+      "loss": 0.8713,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96509094539216e-05,
+      "loss": 0.8856,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649282091166636e-05,
+      "loss": 0.8921,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647651011772055e-05,
+      "loss": 0.9273,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646016216366088e-05,
+      "loss": 0.8438,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644377705578426e-05,
+      "loss": 0.3687,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642735480040175e-05,
+      "loss": 0.8798,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641089540383877e-05,
+      "loss": 0.9165,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96394398872435e-05,
+      "loss": 0.8559,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637786521254445e-05,
+      "loss": 0.3438,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636129443053545e-05,
+      "loss": 0.9098,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634468653279062e-05,
+      "loss": 0.8783,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632804152570685e-05,
+      "loss": 0.9063,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963113594156954e-05,
+      "loss": 0.9107,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629464020918164e-05,
+      "loss": 0.8861,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627788391260547e-05,
+      "loss": 0.8373,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626109053242092e-05,
+      "loss": 0.9228,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962442600750963e-05,
+      "loss": 0.8539,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962273925471143e-05,
+      "loss": 0.8463,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962104879549718e-05,
+      "loss": 0.8747,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619354630518e-05,
+      "loss": 0.8932,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961765676042643e-05,
+      "loss": 0.8929,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615955185876454e-05,
+      "loss": 0.8738,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614249907523466e-05,
+      "loss": 0.9454,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961254092602429e-05,
+      "loss": 0.8918,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610828242037187e-05,
+      "loss": 0.8847,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609111856221827e-05,
+      "loss": 0.882,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607391769239317e-05,
+      "loss": 0.8603,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605667981752193e-05,
+      "loss": 0.9063,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603940494424404e-05,
+      "loss": 0.9393,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602209307921336e-05,
+      "loss": 0.8417,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600474422909793e-05,
+      "loss": 0.9007,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598735840058002e-05,
+      "loss": 0.8706,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596993560035623e-05,
+      "loss": 0.9105,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595247583513733e-05,
+      "loss": 0.9187,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593497911164834e-05,
+      "loss": 0.92,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959174454366285e-05,
+      "loss": 0.9307,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589987481683132e-05,
+      "loss": 0.8687,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588226725902453e-05,
+      "loss": 0.8595,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586462276999004e-05,
+      "loss": 0.9091,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958469413565241e-05,
+      "loss": 0.927,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95829223025437e-05,
+      "loss": 0.9223,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581146778355347e-05,
+      "loss": 0.902,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957936756377123e-05,
+      "loss": 0.8921,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577584659476647e-05,
+      "loss": 0.3509,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575798066158336e-05,
+      "loss": 0.8688,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574007784504435e-05,
+      "loss": 0.897,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572213815204516e-05,
+      "loss": 0.9038,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570416158949563e-05,
+      "loss": 0.9068,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568614816431987e-05,
+      "loss": 0.8645,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566809788345614e-05,
+      "loss": 0.9558,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565001075385698e-05,
+      "loss": 0.9507,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563188678248893e-05,
+      "loss": 0.8758,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95613725976333e-05,
+      "loss": 0.8902,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955955283423841e-05,
+      "loss": 0.8704,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557729388765155e-05,
+      "loss": 0.9019,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555902261915875e-05,
+      "loss": 0.8704,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554071454394325e-05,
+      "loss": 0.8651,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955223696690568e-05,
+      "loss": 0.8962,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550398800156548e-05,
+      "loss": 0.8604,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548556954854926e-05,
+      "loss": 0.858,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546711431710254e-05,
+      "loss": 0.8857,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544862231433365e-05,
+      "loss": 0.9059,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543009354736528e-05,
+      "loss": 0.9234,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954115280233342e-05,
+      "loss": 0.9714,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953929257493913e-05,
+      "loss": 0.8651,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537428673270176e-05,
+      "loss": 0.8977,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953556109804447e-05,
+      "loss": 0.8978,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953368984998136e-05,
+      "loss": 0.9374,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531814929801595e-05,
+      "loss": 0.8724,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529936338227346e-05,
+      "loss": 0.8712,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528054075982194e-05,
+      "loss": 0.8435,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526168143791133e-05,
+      "loss": 0.9022,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524278542380575e-05,
+      "loss": 0.9211,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952238527247834e-05,
+      "loss": 0.9207,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520488334813667e-05,
+      "loss": 0.8502,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95185877301172e-05,
+      "loss": 0.9078,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516683459121008e-05,
+      "loss": 0.8572,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514775522558555e-05,
+      "loss": 0.8908,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951286392116473e-05,
+      "loss": 0.8749,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510948655675836e-05,
+      "loss": 0.8596,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950902972682957e-05,
+      "loss": 0.9243,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950710713536505e-05,
+      "loss": 0.8833,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505180882022817e-05,
+      "loss": 0.887,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950325096754481e-05,
+      "loss": 0.8433,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501317392674366e-05,
+      "loss": 0.9222,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499380158156258e-05,
+      "loss": 0.9427,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497439264736652e-05,
+      "loss": 0.9299,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495494713163127e-05,
+      "loss": 0.838,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493546504184668e-05,
+      "loss": 0.8633,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491594638551676e-05,
+      "loss": 0.8829,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489639117015956e-05,
+      "loss": 0.853,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948767994033072e-05,
+      "loss": 0.9269,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948571710925059e-05,
+      "loss": 0.8509,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483750624531593e-05,
+      "loss": 0.918,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481780486931163e-05,
+      "loss": 0.9035,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479806697208153e-05,
+      "loss": 0.8899,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477829256122805e-05,
+      "loss": 0.8972,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947584816443677e-05,
+      "loss": 0.8977,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473863422913127e-05,
+      "loss": 0.8562,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947187503231633e-05,
+      "loss": 0.8455,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469882993412257e-05,
+      "loss": 0.889,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467887306968187e-05,
+      "loss": 0.8959,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465887973752804e-05,
+      "loss": 0.9442,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463884994536198e-05,
+      "loss": 0.8603,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461878370089854e-05,
+      "loss": 0.8946,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459868101186684e-05,
+      "loss": 0.8697,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457854188600972e-05,
+      "loss": 0.8943,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455836633108433e-05,
+      "loss": 0.8439,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453815435486168e-05,
+      "loss": 0.8546,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451790596512696e-05,
+      "loss": 0.8958,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449762116967915e-05,
+      "loss": 0.8726,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447729997633155e-05,
+      "loss": 0.8878,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445694239291124e-05,
+      "loss": 0.8513,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944365484272594e-05,
+      "loss": 0.8219,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944161180872313e-05,
+      "loss": 0.8115,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439565138069606e-05,
+      "loss": 0.8122,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437514831553698e-05,
+      "loss": 0.9237,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943546088996512e-05,
+      "loss": 0.891,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433403314095e-05,
+      "loss": 0.902,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943134210473586e-05,
+      "loss": 0.8657,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429277262681615e-05,
+      "loss": 0.8747,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942720878872759e-05,
+      "loss": 0.8083,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942513668367051e-05,
+      "loss": 0.9538,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423060948308487e-05,
+      "loss": 0.8964,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420981583441035e-05,
+      "loss": 0.9404,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941889858986907e-05,
+      "loss": 0.9023,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941681196839491e-05,
+      "loss": 0.8842,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941472171982226e-05,
+      "loss": 0.7933,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412627844956227e-05,
+      "loss": 0.9078,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410530344603312e-05,
+      "loss": 0.8435,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408429219571423e-05,
+      "loss": 0.8393,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406324470669846e-05,
+      "loss": 0.8812,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940421609870928e-05,
+      "loss": 0.3442,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402104104501805e-05,
+      "loss": 0.8467,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939998848886091e-05,
+      "loss": 0.9159,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397869252601472e-05,
+      "loss": 0.9648,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395746396539755e-05,
+      "loss": 0.9126,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393619921493437e-05,
+      "loss": 0.8767,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939148982828157e-05,
+      "loss": 0.9391,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938935611772461e-05,
+      "loss": 0.8197,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93872187906444e-05,
+      "loss": 0.8985,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938507784786419e-05,
+      "loss": 0.9148,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382933290208603e-05,
+      "loss": 0.8571,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380785118503668e-05,
+      "loss": 0.9604,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93786333335768e-05,
+      "loss": 0.8707,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376477936256806e-05,
+      "loss": 0.8364,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937431892737389e-05,
+      "loss": 0.3591,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372156307759647e-05,
+      "loss": 0.9173,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936999007824705e-05,
+      "loss": 0.8693,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367820239670475e-05,
+      "loss": 0.871,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365646792865687e-05,
+      "loss": 0.9147,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363469738669834e-05,
+      "loss": 0.9281,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361289077921465e-05,
+      "loss": 0.9058,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93591048114605e-05,
+      "loss": 0.3596,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356916940128266e-05,
+      "loss": 0.868,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354725464767474e-05,
+      "loss": 0.9623,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935253038622221e-05,
+      "loss": 0.8398,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350331705337967e-05,
+      "loss": 0.9022,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348129422961618e-05,
+      "loss": 0.9387,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345923539941417e-05,
+      "loss": 0.9143,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343714057127006e-05,
+      "loss": 0.8683,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341500975369424e-05,
+      "loss": 0.8156,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933928429552109e-05,
+      "loss": 0.9377,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337064018435807e-05,
+      "loss": 0.8444,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334840144968763e-05,
+      "loss": 0.8927,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332612675976536e-05,
+      "loss": 0.9055,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933038161231708e-05,
+      "loss": 0.8373,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932814695484975e-05,
+      "loss": 0.9296,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932590870443527e-05,
+      "loss": 0.8299,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932366686193575e-05,
+      "loss": 0.9071,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321421428214686e-05,
+      "loss": 0.9168,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319172404136963e-05,
+      "loss": 0.8579,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316919790568843e-05,
+      "loss": 0.9296,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314663588377968e-05,
+      "loss": 0.905,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312403798433364e-05,
+      "loss": 0.8699,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310140421605445e-05,
+      "loss": 0.8675,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307873458766e-05,
+      "loss": 0.8975,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93056029107882e-05,
+      "loss": 0.8292,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303328778546602e-05,
+      "loss": 0.8075,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930105106291713e-05,
+      "loss": 0.8894,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929876976477711e-05,
+      "loss": 0.932,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296484885005228e-05,
+      "loss": 0.8368,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929419642448156e-05,
+      "loss": 0.878,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291904384087556e-05,
+      "loss": 0.9171,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289608764706052e-05,
+      "loss": 0.9109,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928730956722125e-05,
+      "loss": 0.9037,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285006792518746e-05,
+      "loss": 0.9206,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92827004414855e-05,
+      "loss": 0.8324,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928039051500986e-05,
+      "loss": 0.9214,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927807701398154e-05,
+      "loss": 0.8884,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275759939291648e-05,
+      "loss": 0.3489,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273439291832646e-05,
+      "loss": 0.9288,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927111507249839e-05,
+      "loss": 0.8751,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926878728218411e-05,
+      "loss": 0.8884,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92664559217864e-05,
+      "loss": 0.8728,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264120992203236e-05,
+      "loss": 0.8624,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261782494333974e-05,
+      "loss": 0.9506,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259440429079335e-05,
+      "loss": 0.8998,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257094797341427e-05,
+      "loss": 0.8745,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254745600023713e-05,
+      "loss": 0.8582,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925239283803104e-05,
+      "loss": 0.8915,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250036512269635e-05,
+      "loss": 0.8933,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247676623647086e-05,
+      "loss": 0.9122,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245313173072354e-05,
+      "loss": 0.8891,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242946161455783e-05,
+      "loss": 0.8412,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924057558970908e-05,
+      "loss": 0.9168,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238201458745313e-05,
+      "loss": 0.9668,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923582376947895e-05,
+      "loss": 0.7905,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92334425228258e-05,
+      "loss": 0.9207,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231057719703057e-05,
+      "loss": 0.8761,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922866936102928e-05,
+      "loss": 0.8969,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226277447724407e-05,
+      "loss": 0.9204,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223881980709727e-05,
+      "loss": 0.8907,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221482960907914e-05,
+      "loss": 0.8937,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921908038924301e-05,
+      "loss": 0.8902,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921667426664041e-05,
+      "loss": 0.8863,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921426459402689e-05,
+      "loss": 0.8699,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211851372330595e-05,
+      "loss": 0.3846,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209434602481026e-05,
+      "loss": 0.9152,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920701428540906e-05,
+      "loss": 0.8869,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204590422046934e-05,
+      "loss": 0.8779,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202163013328256e-05,
+      "loss": 0.9113,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199732060187996e-05,
+      "loss": 0.8402,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197297563562494e-05,
+      "loss": 0.8239,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194859524389447e-05,
+      "loss": 0.8852,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192417943607923e-05,
+      "loss": 0.8323,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189972822158355e-05,
+      "loss": 0.8614,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187524160982533e-05,
+      "loss": 0.9441,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185071961023617e-05,
+      "loss": 0.9202,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182616223226126e-05,
+      "loss": 0.9348,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918015694853594e-05,
+      "loss": 0.9632,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917769413790031e-05,
+      "loss": 0.8535,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175227792267842e-05,
+      "loss": 0.8741,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91727579125885e-05,
+      "loss": 0.8579,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170284499813618e-05,
+      "loss": 0.8562,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916780755489589e-05,
+      "loss": 0.8696,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165327078789363e-05,
+      "loss": 0.8856,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916284307244945e-05,
+      "loss": 0.8638,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916035553683292e-05,
+      "loss": 0.934,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157864472897908e-05,
+      "loss": 0.8846,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155369881603904e-05,
+      "loss": 0.9355,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915287176391175e-05,
+      "loss": 0.9314,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915037012078366e-05,
+      "loss": 0.8797,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914786495318319e-05,
+      "loss": 0.842,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914535626207527e-05,
+      "loss": 0.8951,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142844048426177e-05,
+      "loss": 0.8974,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140328313203548e-05,
+      "loss": 0.8831,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913780905737637e-05,
+      "loss": 0.839,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135286281915e-05,
+      "loss": 0.8347,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913275998779113e-05,
+      "loss": 0.8531,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130230175977834e-05,
+      "loss": 0.8285,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127696847449514e-05,
+      "loss": 0.9763,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125160003181945e-05,
+      "loss": 0.8365,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912261964415225e-05,
+      "loss": 0.3607,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120075771338907e-05,
+      "loss": 0.8032,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117528385721743e-05,
+      "loss": 0.917,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911497748828194e-05,
+      "loss": 0.8582,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112423080002038e-05,
+      "loss": 0.3755,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109865161865925e-05,
+      "loss": 0.9324,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910730373485884e-05,
+      "loss": 0.8842,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104738799967373e-05,
+      "loss": 0.8945,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910217035817947e-05,
+      "loss": 0.8526,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099598410484418e-05,
+      "loss": 0.8512,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909702295787287e-05,
+      "loss": 0.8906,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094444001336813e-05,
+      "loss": 0.8598,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091861541869595e-05,
+      "loss": 0.8802,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089275580465907e-05,
+      "loss": 0.8232,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908668611812179e-05,
+      "loss": 0.8873,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084093155834635e-05,
+      "loss": 0.8135,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081496694603178e-05,
+      "loss": 0.8323,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907889673542751e-05,
+      "loss": 0.8687,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076293279309057e-05,
+      "loss": 0.8486,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073686327250605e-05,
+      "loss": 0.8516,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907107588025628e-05,
+      "loss": 0.8589,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906846193933155e-05,
+      "loss": 0.8518,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906584450548324e-05,
+      "loss": 0.8624,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063223579719507e-05,
+      "loss": 0.8543,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906059916304987e-05,
+      "loss": 0.8394,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905797125648517e-05,
+      "loss": 0.8745,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055339861037618e-05,
+      "loss": 0.8195,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905270497772074e-05,
+      "loss": 0.8981,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050066607549438e-05,
+      "loss": 0.8556,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904742475153993e-05,
+      "loss": 0.8883,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044779410709786e-05,
+      "loss": 0.3586,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042130586077923e-05,
+      "loss": 0.9391,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039478278664593e-05,
+      "loss": 0.89,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036822489491396e-05,
+      "loss": 0.8317,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034163219581264e-05,
+      "loss": 0.9237,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903150046995848e-05,
+      "loss": 0.9401,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902883424164866e-05,
+      "loss": 0.9086,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026164535678763e-05,
+      "loss": 0.8851,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023491353077085e-05,
+      "loss": 0.843,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020814694873267e-05,
+      "loss": 0.9327,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901813456209828e-05,
+      "loss": 0.7914,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901545095578444e-05,
+      "loss": 0.8728,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012763876965395e-05,
+      "loss": 0.9161,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901007332667614e-05,
+      "loss": 0.4126,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007379305953003e-05,
+      "loss": 0.969,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900468181583364e-05,
+      "loss": 0.8231,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001980857357053e-05,
+      "loss": 0.8936,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899927643156358e-05,
+      "loss": 0.8606,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996568539494887e-05,
+      "loss": 0.8276,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993857182193985e-05,
+      "loss": 0.8779,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899114236070521e-05,
+      "loss": 0.8573,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898842407607424e-05,
+      "loss": 0.8757,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898570232934809e-05,
+      "loss": 0.8749,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982977121575088e-05,
+      "loss": 0.8899,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980248453804918e-05,
+      "loss": 0.8872,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977516327088592e-05,
+      "loss": 0.3744,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974780742478445e-05,
+      "loss": 0.826,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897204170102815e-05,
+      "loss": 0.9123,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969299203792707e-05,
+      "loss": 0.8717,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966553251828462e-05,
+      "loss": 0.9436,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896380384619307e-05,
+      "loss": 0.7949,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896105098794553e-05,
+      "loss": 0.8444,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958294678146173e-05,
+      "loss": 0.8769,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895553491785665e-05,
+      "loss": 0.3471,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895277170813994e-05,
+      "loss": 0.8685,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950005050060362e-05,
+      "loss": 0.8559,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947234944683557e-05,
+      "loss": 0.8581,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944461393076493e-05,
+      "loss": 0.8504,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941684396307466e-05,
+      "loss": 0.9042,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89389039554461e-05,
+      "loss": 0.8446,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893612007156334e-05,
+      "loss": 0.321,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893333274573147e-05,
+      "loss": 0.8741,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930541979024086e-05,
+      "loss": 0.8885,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892774777251612e-05,
+      "loss": 0.8968,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924950127283813e-05,
+      "loss": 0.9258,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922149044404757e-05,
+      "loss": 0.8615,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919344524957838e-05,
+      "loss": 0.8419,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891653657002329e-05,
+      "loss": 0.8629,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913725180682658e-05,
+      "loss": 0.8616,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891091035801881e-05,
+      "loss": 0.842,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908092103115942e-05,
+      "loss": 0.3355,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905270417059563e-05,
+      "loss": 0.3299,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902445300936515e-05,
+      "loss": 0.8894,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899616755834954e-05,
+      "loss": 0.9292,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896784782844358e-05,
+      "loss": 0.9024,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889394938305553e-05,
+      "loss": 0.8613,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889111055756058e-05,
+      "loss": 0.7948,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888826830745295e-05,
+      "loss": 0.8333,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885422633827396e-05,
+      "loss": 0.9777,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882573537779996e-05,
+      "loss": 0.9098,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879721020408143e-05,
+      "loss": 0.3215,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887686508281055e-05,
+      "loss": 0.8015,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874005726087237e-05,
+      "loss": 0.8201,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871142951339564e-05,
+      "loss": 0.9115,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868276759670184e-05,
+      "loss": 0.9242,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865407152183082e-05,
+      "loss": 0.8665,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862534129983546e-05,
+      "loss": 0.8551,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859657694178192e-05,
+      "loss": 0.9014,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856777845874937e-05,
+      "loss": 0.8188,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853894586183025e-05,
+      "loss": 0.9154,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851007916213007e-05,
+      "loss": 0.87,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848117837076752e-05,
+      "loss": 0.3442,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845224349887433e-05,
+      "loss": 0.8265,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842327455759553e-05,
+      "loss": 0.3713,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839427155808904e-05,
+      "loss": 0.8434,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836523451152612e-05,
+      "loss": 0.8709,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88336163429091e-05,
+      "loss": 0.9221,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830705832198107e-05,
+      "loss": 0.3455,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882779192014068e-05,
+      "loss": 0.9341,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824874607859187e-05,
+      "loss": 0.8375,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821953896477288e-05,
+      "loss": 0.917,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819029787119962e-05,
+      "loss": 0.8343,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816102280913496e-05,
+      "loss": 0.8784,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813171378985494e-05,
+      "loss": 0.8484,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810237082464847e-05,
+      "loss": 0.8798,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807299392481772e-05,
+      "loss": 0.8641,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880435831016779e-05,
+      "loss": 0.9236,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801413836655718e-05,
+      "loss": 0.8744,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879846597307969e-05,
+      "loss": 0.8639,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795514720575145e-05,
+      "loss": 0.8586,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792560080278822e-05,
+      "loss": 0.8156,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789602053328767e-05,
+      "loss": 0.9123,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786640640864332e-05,
+      "loss": 0.8835,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783675844026176e-05,
+      "loss": 0.8085,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878070766395625e-05,
+      "loss": 0.9003,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777736101797822e-05,
+      "loss": 0.8396,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774761158695453e-05,
+      "loss": 0.8631,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771782835795015e-05,
+      "loss": 0.8831,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876880113424367e-05,
+      "loss": 0.9079,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765816055189888e-05,
+      "loss": 0.8738,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762827599783447e-05,
+      "loss": 0.3445,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759835769175412e-05,
+      "loss": 0.9434,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756840564518155e-05,
+      "loss": 0.8939,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875384198696535e-05,
+      "loss": 0.9569,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750840037671963e-05,
+      "loss": 0.9374,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747834717794263e-05,
+      "loss": 0.8574,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744826028489823e-05,
+      "loss": 0.9074,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741813970917502e-05,
+      "loss": 0.8879,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738798546237465e-05,
+      "loss": 0.8896,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735779755611167e-05,
+      "loss": 0.8582,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732757600201374e-05,
+      "loss": 0.9164,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872973208117212e-05,
+      "loss": 0.9406,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726703199688773e-05,
+      "loss": 0.858,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723670956917955e-05,
+      "loss": 0.823,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872063535402762e-05,
+      "loss": 0.8245,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871759639218699e-05,
+      "loss": 0.9159,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714554072566593e-05,
+      "loss": 0.3366,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711508396338246e-05,
+      "loss": 0.9103,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870845936467506e-05,
+      "loss": 0.851,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870540697875144e-05,
+      "loss": 0.8871,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870235123974308e-05,
+      "loss": 0.8928,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699292148826972e-05,
+      "loss": 0.3364,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696229707181385e-05,
+      "loss": 0.8614,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693163915985896e-05,
+      "loss": 0.8552,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869009477642136e-05,
+      "loss": 0.8374,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687022289669926e-05,
+      "loss": 0.8568,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683946456915032e-05,
+      "loss": 0.3513,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680867279341406e-05,
+      "loss": 0.9031,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867778475813506e-05,
+      "loss": 0.832,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674698894483303e-05,
+      "loss": 0.8896,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867160968957472e-05,
+      "loss": 0.8961,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866851714459919e-05,
+      "loss": 0.8727,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665421260747875e-05,
+      "loss": 0.9242,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662322039213227e-05,
+      "loss": 0.9225,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865921948118898e-05,
+      "loss": 0.8925,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656113587870154e-05,
+      "loss": 0.892,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653004360453058e-05,
+      "loss": 0.8708,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864989180013528e-05,
+      "loss": 0.8524,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646775908115685e-05,
+      "loss": 0.8544,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643656685594443e-05,
+      "loss": 0.8432,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640534133772982e-05,
+      "loss": 0.8534,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863740825385403e-05,
+      "loss": 0.8482,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863427904704159e-05,
+      "loss": 0.9062,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631146514540943e-05,
+      "loss": 0.9354,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628010657558656e-05,
+      "loss": 0.8729,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862487147730258e-05,
+      "loss": 0.901,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621728974981836e-05,
+      "loss": 0.8826,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618583151806832e-05,
+      "loss": 0.8917,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615434008989253e-05,
+      "loss": 0.8328,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612281547742068e-05,
+      "loss": 0.8554,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609125769279507e-05,
+      "loss": 0.8887,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605966674817097e-05,
+      "loss": 0.8941,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602804265571634e-05,
+      "loss": 0.8601,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599638542761194e-05,
+      "loss": 0.8898,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596469507605123e-05,
+      "loss": 0.7884,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593297161324047e-05,
+      "loss": 0.8524,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859012150513987e-05,
+      "loss": 0.8209,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858694254027577e-05,
+      "loss": 0.8756,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583760267956187e-05,
+      "loss": 0.9076,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580574689406853e-05,
+      "loss": 0.8597,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577385805854767e-05,
+      "loss": 0.8672,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574193618528195e-05,
+      "loss": 0.8388,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570998128656685e-05,
+      "loss": 0.832,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567799337471048e-05,
+      "loss": 0.8474,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564597246203374e-05,
+      "loss": 0.7966,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561391856087018e-05,
+      "loss": 0.8335,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558183168356612e-05,
+      "loss": 0.8545,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554971184248052e-05,
+      "loss": 0.8385,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551755904998506e-05,
+      "loss": 0.8808,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548537331846416e-05,
+      "loss": 0.8873,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545315466031484e-05,
+      "loss": 0.8871,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854209030879469e-05,
+      "loss": 0.8759,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853886186137827e-05,
+      "loss": 0.8256,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535630125025737e-05,
+      "loss": 0.8468,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532395100981867e-05,
+      "loss": 0.873,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85291567904927e-05,
+      "loss": 0.8164,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852591519480555e-05,
+      "loss": 0.8708,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522670315168985e-05,
+      "loss": 0.8663,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519422152832853e-05,
+      "loss": 0.3927,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516170709048246e-05,
+      "loss": 0.8974,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851291598506754e-05,
+      "loss": 0.9396,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850965798214436e-05,
+      "loss": 0.8102,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506396701533607e-05,
+      "loss": 0.9215,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503132144491427e-05,
+      "loss": 0.8421,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499864312275244e-05,
+      "loss": 0.8954,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849659320614374e-05,
+      "loss": 0.9396,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849331882735685e-05,
+      "loss": 0.9292,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849004117717578e-05,
+      "loss": 0.9029,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486760256862986e-05,
+      "loss": 0.3413,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483476067682193e-05,
+      "loss": 0.3804,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848018861089838e-05,
+      "loss": 0.8667,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476897887777788e-05,
+      "loss": 0.9178,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473603899587906e-05,
+      "loss": 0.8475,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470306647597493e-05,
+      "loss": 0.8405,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467006133076565e-05,
+      "loss": 0.8033,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846370235729638e-05,
+      "loss": 0.9325,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460395321529468e-05,
+      "loss": 0.8809,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845708502704961e-05,
+      "loss": 0.8996,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845377147513184e-05,
+      "loss": 0.9685,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450454667052446e-05,
+      "loss": 0.8806,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844713460408897e-05,
+      "loss": 0.9172,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443811287520216e-05,
+      "loss": 0.8848,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440484718626222e-05,
+      "loss": 0.8765,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843715489868831e-05,
+      "loss": 0.8717,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433821828989018e-05,
+      "loss": 0.3451,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430485510812165e-05,
+      "loss": 0.3644,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427145945442802e-05,
+      "loss": 0.8313,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423803134167244e-05,
+      "loss": 0.8479,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420457078273047e-05,
+      "loss": 0.8407,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417107779049025e-05,
+      "loss": 0.3508,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413755237785228e-05,
+      "loss": 0.8512,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841039945577297e-05,
+      "loss": 0.8758,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407040434304805e-05,
+      "loss": 0.9087,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403678174674532e-05,
+      "loss": 0.8203,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400312678177207e-05,
+      "loss": 0.9127,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396943946109128e-05,
+      "loss": 0.819,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839357197976783e-05,
+      "loss": 0.8641,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390196780452113e-05,
+      "loss": 0.8708,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386818349462007e-05,
+      "loss": 0.9033,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838343668809879e-05,
+      "loss": 0.9135,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380051797664987e-05,
+      "loss": 0.8081,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837666367946436e-05,
+      "loss": 0.8379,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837327233480193e-05,
+      "loss": 0.8349,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369877764983944e-05,
+      "loss": 0.8113,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366479971317895e-05,
+      "loss": 0.9129,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363078955112522e-05,
+      "loss": 0.877,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359674717677808e-05,
+      "loss": 0.8574,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356267260324963e-05,
+      "loss": 0.9255,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835285658436646e-05,
+      "loss": 0.8844,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349442691115987e-05,
+      "loss": 0.8323,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346025581888482e-05,
+      "loss": 0.8322,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342605258000127e-05,
+      "loss": 0.7554,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833918172076834e-05,
+      "loss": 0.9079,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335754971511766e-05,
+      "loss": 0.8989,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83323250115503e-05,
+      "loss": 0.8296,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328891842205065e-05,
+      "loss": 0.8487,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832545546479843e-05,
+      "loss": 0.8883,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322015880653988e-05,
+      "loss": 0.8151,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318573091096578e-05,
+      "loss": 0.8649,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831512709745226e-05,
+      "loss": 0.8912,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311677901048347e-05,
+      "loss": 0.8992,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308225503213367e-05,
+      "loss": 0.8659,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304769905277094e-05,
+      "loss": 0.8467,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830131110857053e-05,
+      "loss": 0.7672,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82978491144259e-05,
+      "loss": 0.3626,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294383924176685e-05,
+      "loss": 0.3705,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290915539157568e-05,
+      "loss": 0.8672,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287443960704482e-05,
+      "loss": 0.8399,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283969190154582e-05,
+      "loss": 0.9148,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280491228846257e-05,
+      "loss": 0.873,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277010078119123e-05,
+      "loss": 0.8672,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273525739314025e-05,
+      "loss": 0.8161,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270038213773027e-05,
+      "loss": 0.8499,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266547502839436e-05,
+      "loss": 0.9155,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263053607857775e-05,
+      "loss": 0.8638,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259556530173802e-05,
+      "loss": 0.8708,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256056271134487e-05,
+      "loss": 0.8778,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252552832088044e-05,
+      "loss": 0.8513,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249046214383893e-05,
+      "loss": 0.3781,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245536419372692e-05,
+      "loss": 0.8347,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242023448406322e-05,
+      "loss": 0.8246,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238507302837874e-05,
+      "loss": 0.7802,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234987984021673e-05,
+      "loss": 0.3468,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231465493313273e-05,
+      "loss": 0.8923,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227939832069435e-05,
+      "loss": 0.8737,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8224411001648144e-05,
+      "loss": 0.8884,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220879003408616e-05,
+      "loss": 0.8497,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217343838711276e-05,
+      "loss": 0.8586,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213805508917773e-05,
+      "loss": 0.894,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210264015390974e-05,
+      "loss": 0.8903,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206719359494967e-05,
+      "loss": 0.9138,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820317154259506e-05,
+      "loss": 0.8328,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199620566057766e-05,
+      "loss": 0.9152,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196066431250834e-05,
+      "loss": 0.8335,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819250913954321e-05,
+      "loss": 0.8493,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188948692305075e-05,
+      "loss": 0.8975,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818538509090781e-05,
+      "loss": 0.8606,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181818336724018e-05,
+      "loss": 0.8271,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178248431127518e-05,
+      "loss": 0.7976,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174675375493337e-05,
+      "loss": 0.9182,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171099171197715e-05,
+      "loss": 0.8964,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167519819618116e-05,
+      "loss": 0.865,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163937322133205e-05,
+      "loss": 0.8848,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160351680122858e-05,
+      "loss": 0.8678,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156762894968173e-05,
+      "loss": 0.8801,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153170968051447e-05,
+      "loss": 0.8493,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149575900756193e-05,
+      "loss": 0.3686,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814597769446713e-05,
+      "loss": 0.8505,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142376350570198e-05,
+      "loss": 0.8349,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138771870452528e-05,
+      "loss": 0.7867,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135164255502468e-05,
+      "loss": 0.8463,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131553507109575e-05,
+      "loss": 0.8758,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812793962666461e-05,
+      "loss": 0.8561,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812432261555954e-05,
+      "loss": 0.928,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812070247518754e-05,
+      "loss": 0.8939,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117079206942986e-05,
+      "loss": 0.8524,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811345281222147e-05,
+      "loss": 0.908,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109823292419765e-05,
+      "loss": 0.8129,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106190648935877e-05,
+      "loss": 0.8326,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102554883169e-05,
+      "loss": 0.8959,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098915996519522e-05,
+      "loss": 0.9113,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8095273990389047e-05,
+      "loss": 0.8328,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809162886618038e-05,
+      "loss": 0.7993,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087980625297522e-05,
+      "loss": 0.8545,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8084329269145673e-05,
+      "loss": 0.8519,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080674799131235e-05,
+      "loss": 0.8637,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077017216661815e-05,
+      "loss": 0.8348,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807335652314621e-05,
+      "loss": 0.9158,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069692719994413e-05,
+      "loss": 0.8796,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066025808617636e-05,
+      "loss": 0.8422,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062355790428258e-05,
+      "loss": 0.9159,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805868266683988e-05,
+      "loss": 0.8262,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805500643926728e-05,
+      "loss": 0.8311,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805132710912645e-05,
+      "loss": 0.8598,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047644677834562e-05,
+      "loss": 0.8954,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043959146809986e-05,
+      "loss": 0.9036,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804027051747229e-05,
+      "loss": 0.8696,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036578791242237e-05,
+      "loss": 0.8875,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803288396954177e-05,
+      "loss": 0.8786,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802918605379404e-05,
+      "loss": 0.8545,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025485045423386e-05,
+      "loss": 0.8188,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021780945855325e-05,
+      "loss": 0.9014,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801807375651658e-05,
+      "loss": 0.9482,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014363478835064e-05,
+      "loss": 0.8584,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801065011423987e-05,
+      "loss": 0.8621,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800693366416128e-05,
+      "loss": 0.9181,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003214130030778e-05,
+      "loss": 0.3301,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799949151328102e-05,
+      "loss": 0.8475,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799576581534586e-05,
+      "loss": 0.867,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992037037660336e-05,
+      "loss": 0.7964,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988305181660665e-05,
+      "loss": 0.8258,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798457024878426e-05,
+      "loss": 0.8977,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980832240469716e-05,
+      "loss": 0.9012,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797709115815681e-05,
+      "loss": 0.9138,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973347003286503e-05,
+      "loss": 0.8931,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969599777300945e-05,
+      "loss": 0.894,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796584948164346e-05,
+      "loss": 0.9267,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962096117758563e-05,
+      "loss": 0.8439,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795833968709195e-05,
+      "loss": 0.884,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954580191090492e-05,
+      "loss": 0.8923,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795081763120224e-05,
+      "loss": 0.835,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947052008876436e-05,
+      "loss": 0.7916,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794328332556349e-05,
+      "loss": 0.8107,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939511582714998e-05,
+      "loss": 0.8014,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935736781783736e-05,
+      "loss": 0.8447,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931958924223645e-05,
+      "loss": 0.8148,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792817801148986e-05,
+      "loss": 0.8133,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792439404503868e-05,
+      "loss": 0.8836,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792060702632759e-05,
+      "loss": 0.8301,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916816956815242e-05,
+      "loss": 0.3867,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791302383796147e-05,
+      "loss": 0.8776,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909227671227283e-05,
+      "loss": 0.9058,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790542845807485e-05,
+      "loss": 0.8919,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901626199967534e-05,
+      "loss": 0.9234,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897820898369857e-05,
+      "loss": 0.3365,
+      "step": 1221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789401255474752e-05,
+      "loss": 0.9205,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890201170567383e-05,
+      "loss": 0.8651,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78863867472975e-05,
+      "loss": 0.8797,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788256928640708e-05,
+      "loss": 0.774,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878748789366497e-05,
+      "loss": 0.8592,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874925257647303e-05,
+      "loss": 0.8576,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787109869272223e-05,
+      "loss": 0.898,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786726909606515e-05,
+      "loss": 0.8655,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786343646915113e-05,
+      "loss": 0.8581,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859600813456386e-05,
+      "loss": 0.8007,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855762130458316e-05,
+      "loss": 0.8469,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785192042163547e-05,
+      "loss": 0.8941,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848075688467567e-05,
+      "loss": 0.8769,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78442279324355e-05,
+      "loss": 0.8601,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784037715502131e-05,
+      "loss": 0.8566,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783652335770822e-05,
+      "loss": 0.9034,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832666541980604e-05,
+      "loss": 0.9008,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828806709324005e-05,
+      "loss": 0.8788,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782494386122512e-05,
+      "loss": 0.8504,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821077999171813e-05,
+      "loss": 0.8624,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817209124653113e-05,
+      "loss": 0.8668,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78133372391592e-05,
+      "loss": 0.8896,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809462344181422e-05,
+      "loss": 0.9138,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805584441212277e-05,
+      "loss": 0.8664,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801703531745432e-05,
+      "loss": 0.8594,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797819617275704e-05,
+      "loss": 0.8881,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793932699299078e-05,
+      "loss": 0.3296,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790042779312676e-05,
+      "loss": 0.8465,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786149858814803e-05,
+      "loss": 0.9074,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782253939304897e-05,
+      "loss": 0.8785,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777835502228356e-05,
+      "loss": 0.3673,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774453109252548e-05,
+      "loss": 0.8529,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770548201714774e-05,
+      "loss": 0.9204,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77666403011743e-05,
+      "loss": 0.9102,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762729409136342e-05,
+      "loss": 0.9406,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758815527107273e-05,
+      "loss": 0.8251,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754898656594607e-05,
+      "loss": 0.8838,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775097879910702e-05,
+      "loss": 0.821,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774705595615433e-05,
+      "loss": 0.3658,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774313012924751e-05,
+      "loss": 0.8721,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739201319898683e-05,
+      "loss": 0.9285,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773526952962112e-05,
+      "loss": 0.8711,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731334759929236e-05,
+      "loss": 0.8188,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727397012338598e-05,
+      "loss": 0.9008,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772345628836592e-05,
+      "loss": 0.9297,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7719512589529054e-05,
+      "loss": 0.9056,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715565917347017e-05,
+      "loss": 0.9071,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771161627333995e-05,
+      "loss": 0.9017,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707663659029152e-05,
+      "loss": 0.888,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770370807593706e-05,
+      "loss": 0.9086,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699749525587257e-05,
+      "loss": 0.819,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769578800950447e-05,
+      "loss": 0.8604,
+      "step": 1273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691823529214566e-05,
+      "loss": 0.845,
+      "step": 1274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768785608624455e-05,
+      "loss": 0.849,
+      "step": 1275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683885682122576e-05,
+      "loss": 0.35,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767991231837794e-05,
+      "loss": 0.838,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675935996541063e-05,
+      "loss": 0.8291,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671956718143524e-05,
+      "loss": 0.8891,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667974484718026e-05,
+      "loss": 0.8768,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663989297798423e-05,
+      "loss": 0.9203,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660001158919694e-05,
+      "loss": 0.8989,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656010069617962e-05,
+      "loss": 0.8697,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652016031430493e-05,
+      "loss": 0.3617,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648019045895674e-05,
+      "loss": 0.8811,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764401911455303e-05,
+      "loss": 0.3486,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640016238943235e-05,
+      "loss": 0.8221,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636010420608084e-05,
+      "loss": 0.849,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632001661090505e-05,
+      "loss": 0.8542,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627989961934567e-05,
+      "loss": 0.8804,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623975324685467e-05,
+      "loss": 0.8211,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761995775088953e-05,
+      "loss": 0.8642,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761593724209422e-05,
+      "loss": 0.8666,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761191379984812e-05,
+      "loss": 0.8389,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607887425700957e-05,
+      "loss": 0.8486,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760385812120358e-05,
+      "loss": 0.8355,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759982588790796e-05,
+      "loss": 0.8759,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759579072736721e-05,
+      "loss": 0.8393,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759175264113556e-05,
+      "loss": 0.83,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587711630768374e-05,
+      "loss": 0.8642,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583667697822133e-05,
+      "loss": 0.8182,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579620843854454e-05,
+      "loss": 0.858,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575571070424073e-05,
+      "loss": 0.898,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757151837909085e-05,
+      "loss": 0.8544,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567462771415775e-05,
+      "loss": 0.8153,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563404248960957e-05,
+      "loss": 0.8209,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559342813289623e-05,
+      "loss": 0.7612,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555278465966134e-05,
+      "loss": 0.8305,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551211208555957e-05,
+      "loss": 0.8615,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547141042625695e-05,
+      "loss": 0.8439,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754306796974306e-05,
+      "loss": 0.8495,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753899199147689e-05,
+      "loss": 0.8886,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534913109397147e-05,
+      "loss": 0.9005,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530831325074894e-05,
+      "loss": 0.8469,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526746640082332e-05,
+      "loss": 0.8777,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522659055992763e-05,
+      "loss": 0.8644,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518568574380615e-05,
+      "loss": 0.8878,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751447519682143e-05,
+      "loss": 0.9309,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510378924891864e-05,
+      "loss": 0.8222,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750627976016969e-05,
+      "loss": 0.8481,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502177704233798e-05,
+      "loss": 0.8668,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749807275866418e-05,
+      "loss": 0.8838,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493964925041955e-05,
+      "loss": 0.8744,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489854204949343e-05,
+      "loss": 0.873,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485740599969686e-05,
+      "loss": 0.8211,
+      "step": 1325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748162411168743e-05,
+      "loss": 0.817,
+      "step": 1326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477504741688135e-05,
+      "loss": 0.9246,
+      "step": 1327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747338249155847e-05,
+      "loss": 0.3873,
+      "step": 1328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469257362886206e-05,
+      "loss": 0.8707,
+      "step": 1329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746512935726024e-05,
+      "loss": 0.8721,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746099847627056e-05,
+      "loss": 0.8326,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456864721508272e-05,
+      "loss": 0.8723,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745272809456558e-05,
+      "loss": 0.9244,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448588597035806e-05,
+      "loss": 0.8443,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744444623051337e-05,
+      "loss": 0.845,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440300996593795e-05,
+      "loss": 0.8391,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436152896873717e-05,
+      "loss": 0.9207,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432001932950865e-05,
+      "loss": 0.8663,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742784810642408e-05,
+      "loss": 0.8472,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423691418893303e-05,
+      "loss": 0.8744,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741953187195957e-05,
+      "loss": 0.8928,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415369467225035e-05,
+      "loss": 0.8387,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411204206292938e-05,
+      "loss": 0.8533,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740703609076762e-05,
+      "loss": 0.897,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402865122254525e-05,
+      "loss": 0.8695,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7398691302360203e-05,
+      "loss": 0.8385,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394514632692296e-05,
+      "loss": 0.3404,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390335114859532e-05,
+      "loss": 0.8275,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386152750471755e-05,
+      "loss": 0.7325,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381967541139894e-05,
+      "loss": 0.9041,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737777948847598e-05,
+      "loss": 0.9198,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373588594093137e-05,
+      "loss": 0.8871,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369394859605576e-05,
+      "loss": 0.8696,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736519828662862e-05,
+      "loss": 0.8684,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360998876778667e-05,
+      "loss": 0.8285,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356796631673213e-05,
+      "loss": 0.8598,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352591552930852e-05,
+      "loss": 0.8503,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348383642171268e-05,
+      "loss": 0.8552,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344172901015227e-05,
+      "loss": 0.8339,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339959331084594e-05,
+      "loss": 0.8122,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335742934002322e-05,
+      "loss": 0.9142,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733152371139245e-05,
+      "loss": 0.8626,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7327301664880106e-05,
+      "loss": 0.912,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732307679609151e-05,
+      "loss": 0.8961,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318849106653962e-05,
+      "loss": 0.9046,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731461859819586e-05,
+      "loss": 0.9117,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731038527234667e-05,
+      "loss": 0.8602,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306149130736965e-05,
+      "loss": 0.9249,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730191017499838e-05,
+      "loss": 0.927,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729766840676365e-05,
+      "loss": 0.3535,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7293423827666588e-05,
+      "loss": 0.9117,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728917643934209e-05,
+      "loss": 0.8614,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728492624342613e-05,
+      "loss": 0.8857,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280673241555773e-05,
+      "loss": 0.9049,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727641743536916e-05,
+      "loss": 0.3778,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272158826505502e-05,
+      "loss": 0.9129,
+      "step": 1376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267897416605104e-05,
+      "loss": 0.8417,
+      "step": 1377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726363320730935e-05,
+      "loss": 0.8325,
+      "step": 1378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725936620026069e-05,
+      "loss": 0.9388,
+      "step": 1379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725509639710266e-05,
+      "loss": 0.9118,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7250823799479868e-05,
+      "loss": 0.8702,
+      "step": 1381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246548409038007e-05,
+      "loss": 0.8293,
+      "step": 1382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724227022742384e-05,
+      "loss": 0.8944,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237989256285207e-05,
+      "loss": 0.8156,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233705497271015e-05,
+      "loss": 0.8386,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229418952031254e-05,
+      "loss": 0.8824,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225129622216984e-05,
+      "loss": 0.3416,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722083750948034e-05,
+      "loss": 0.8265,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216542615474524e-05,
+      "loss": 0.8683,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721224494185381e-05,
+      "loss": 0.8052,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720794449027355e-05,
+      "loss": 0.892,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203641262390158e-05,
+      "loss": 0.8896,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199335259861115e-05,
+      "loss": 0.8492,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195026484344978e-05,
+      "loss": 0.7561,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190714937501375e-05,
+      "loss": 0.854,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186400620990986e-05,
+      "loss": 0.3577,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182083536475577e-05,
+      "loss": 0.8162,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177763685617967e-05,
+      "loss": 0.8361,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173441070082042e-05,
+      "loss": 0.3431,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716911569153276e-05,
+      "loss": 0.8499,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164787551636136e-05,
+      "loss": 0.8643,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716045665205925e-05,
+      "loss": 0.3918,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156122994470253e-05,
+      "loss": 0.895,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151786580538343e-05,
+      "loss": 0.9065,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714744741193379e-05,
+      "loss": 0.8678,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7143105490327924e-05,
+      "loss": 0.8717,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138760817393133e-05,
+      "loss": 0.8545,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134413394802872e-05,
+      "loss": 0.7864,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130063224231643e-05,
+      "loss": 0.8955,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125710307355017e-05,
+      "loss": 0.3371,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121354645849618e-05,
+      "loss": 0.9478,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116996241393122e-05,
+      "loss": 0.9386,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112635095664272e-05,
+      "loss": 0.8365,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108271210342863e-05,
+      "loss": 0.8889,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103904587109745e-05,
+      "loss": 0.9125,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099535227646817e-05,
+      "loss": 0.8192,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709516313363704e-05,
+      "loss": 0.8356,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090788306764427e-05,
+      "loss": 0.8567,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086410748714035e-05,
+      "loss": 0.7805,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082030461171986e-05,
+      "loss": 0.8672,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077647445825447e-05,
+      "loss": 0.8631,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073261704362632e-05,
+      "loss": 0.8643,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706887323847281e-05,
+      "loss": 0.8971,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70644820498463e-05,
+      "loss": 0.8139,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060088140174466e-05,
+      "loss": 0.8649,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055691511149722e-05,
+      "loss": 0.8789,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051292164465526e-05,
+      "loss": 0.8732,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046890101816393e-05,
+      "loss": 0.8327,
+      "step": 1428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704248532489787e-05,
+      "loss": 0.8265,
+      "step": 1429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7038077835406565e-05,
+      "loss": 0.8576,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033667635040115e-05,
+      "loss": 0.8569,
+      "step": 1431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702925472549721e-05,
+      "loss": 0.8564,
+      "step": 1432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024839108477592e-05,
+      "loss": 0.8127,
+      "step": 1433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702042078568202e-05,
+      "loss": 0.8551,
+      "step": 1434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701599975881232e-05,
+      "loss": 0.8876,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701157602957135e-05,
+      "loss": 0.8283,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007149599663005e-05,
+      "loss": 0.8408,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700272047079223e-05,
+      "loss": 0.8357,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998288644665006e-05,
+      "loss": 0.8283,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699385412298834e-05,
+      "loss": 0.8831,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989416907470304e-05,
+      "loss": 0.8783,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984976999819983e-05,
+      "loss": 0.8611,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980534401747507e-05,
+      "loss": 0.3401,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697608911496405e-05,
+      "loss": 0.9522,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697164114118181e-05,
+      "loss": 0.8138,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967190482114025e-05,
+      "loss": 0.8766,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962737139474966e-05,
+      "loss": 0.347,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958281114979943e-05,
+      "loss": 0.8375,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953822410345295e-05,
+      "loss": 0.8459,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949361027288383e-05,
+      "loss": 0.8571,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944896967527623e-05,
+      "loss": 0.9045,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694043023278244e-05,
+      "loss": 0.8471,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935960824773307e-05,
+      "loss": 0.8659,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931488745221707e-05,
+      "loss": 0.7915,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692701399585017e-05,
+      "loss": 0.8427,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922536578382246e-05,
+      "loss": 0.8863,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691805649454251e-05,
+      "loss": 0.8638,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913573746056574e-05,
+      "loss": 0.9035,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690908833465106e-05,
+      "loss": 0.9076,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904600262053638e-05,
+      "loss": 0.8501,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900109529992983e-05,
+      "loss": 0.8893,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895616140198808e-05,
+      "loss": 0.8283,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6891120094401835e-05,
+      "loss": 0.8641,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886621394333826e-05,
+      "loss": 0.8645,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882120041727557e-05,
+      "loss": 0.8511,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687761603831682e-05,
+      "loss": 0.8393,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687310938583644e-05,
+      "loss": 0.8375,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686860008602225e-05,
+      "loss": 0.8213,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864088140611115e-05,
+      "loss": 0.8628,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685957355134091e-05,
+      "loss": 0.9274,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855056319950535e-05,
+      "loss": 0.8752,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68505364481799e-05,
+      "loss": 0.8339,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846013937769935e-05,
+      "loss": 0.8192,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6841488790462593e-05,
+      "loss": 0.8636,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683696100800083e-05,
+      "loss": 0.8131,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832430592128628e-05,
+      "loss": 0.8836,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682789754459098e-05,
+      "loss": 0.8687,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823361867133892e-05,
+      "loss": 0.923,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681882356150438e-05,
+      "loss": 0.8859,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814282629450478e-05,
+      "loss": 0.8139,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809739072721227e-05,
+      "loss": 0.8576,
+      "step": 1481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805192893066682e-05,
+      "loss": 0.7979,
+      "step": 1482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800644092237907e-05,
+      "loss": 0.8423,
+      "step": 1483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796092671986977e-05,
+      "loss": 0.8713,
+      "step": 1484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679153863406697e-05,
+      "loss": 0.8523,
+      "step": 1485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786981980231984e-05,
+      "loss": 0.8628,
+      "step": 1486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782422712237107e-05,
+      "loss": 0.856,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777860831838455e-05,
+      "loss": 0.8872,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773296340793135e-05,
+      "loss": 0.9099,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768729240859264e-05,
+      "loss": 0.853,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764159533795968e-05,
+      "loss": 0.9235,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759587221363366e-05,
+      "loss": 0.8338,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755012305322595e-05,
+      "loss": 0.9153,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750434787435778e-05,
+      "loss": 0.882,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745854669466065e-05,
+      "loss": 0.8717,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741271953177576e-05,
+      "loss": 0.8542,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736686640335455e-05,
+      "loss": 0.8466,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732098732705844e-05,
+      "loss": 0.847,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727508232055867e-05,
+      "loss": 0.8448,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722915140153676e-05,
+      "loss": 0.8312,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671831945876839e-05,
+      "loss": 0.9385,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713721189670148e-05,
+      "loss": 0.8279,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670912033463007e-05,
+      "loss": 0.936,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670451689542029e-05,
+      "loss": 0.8933,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669991087381392e-05,
+      "loss": 0.3793,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695302271585073e-05,
+      "loss": 0.8942,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669069109050886e-05,
+      "loss": 0.8158,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668607733236138e-05,
+      "loss": 0.8619,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681460998919727e-05,
+      "loss": 0.8232,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6676842091961985e-05,
+      "loss": 0.8528,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672220613267233e-05,
+      "loss": 0.8617,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666759656461554e-05,
+      "loss": 0.8228,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6662969947787955e-05,
+      "loss": 0.9115,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658340764566533e-05,
+      "loss": 0.3792,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653709016734306e-05,
+      "loss": 0.857,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649074706075296e-05,
+      "loss": 0.8395,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644437834374508e-05,
+      "loss": 0.3622,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639798403417944e-05,
+      "loss": 0.8476,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663515641499259e-05,
+      "loss": 0.8545,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630511870886404e-05,
+      "loss": 0.8724,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662586477288834e-05,
+      "loss": 0.7855,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621215122788336e-05,
+      "loss": 0.889,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661656292237731e-05,
+      "loss": 0.8878,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661190817344716e-05,
+      "loss": 0.8865,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660725087779077e-05,
+      "loss": 0.844,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602591037202e-05,
+      "loss": 0.8003,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597928653475692e-05,
+      "loss": 0.8519,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659326372840768e-05,
+      "loss": 0.8938,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588596263794752e-05,
+      "loss": 0.8819,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658392626143469e-05,
+      "loss": 0.3537,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579253723126258e-05,
+      "loss": 0.8863,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574578650669187e-05,
+      "loss": 0.3578,
+      "step": 1532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569901045864186e-05,
+      "loss": 0.7495,
+      "step": 1533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565220910512937e-05,
+      "loss": 0.866,
+      "step": 1534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656053824641811e-05,
+      "loss": 0.8338,
+      "step": 1535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555853055383327e-05,
+      "loss": 0.8764,
+      "step": 1536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551165339213208e-05,
+      "loss": 0.8207,
+      "step": 1537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654647509971332e-05,
+      "loss": 0.8792,
+      "step": 1538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654178233869023e-05,
+      "loss": 0.8397,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537087057951445e-05,
+      "loss": 0.8646,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653238925930547e-05,
+      "loss": 0.7911,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527688944561766e-05,
+      "loss": 0.7968,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522986115530767e-05,
+      "loss": 0.7915,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651828077402387e-05,
+      "loss": 0.9083,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513572921853445e-05,
+      "loss": 0.7999,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508862560832827e-05,
+      "loss": 0.9311,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504149692776326e-05,
+      "loss": 0.8929,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499434319499198e-05,
+      "loss": 0.8242,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494716442817685e-05,
+      "loss": 0.8958,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489996064548985e-05,
+      "loss": 0.895,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485273186511247e-05,
+      "loss": 0.8883,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648054781052361e-05,
+      "loss": 0.8646,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475819938406146e-05,
+      "loss": 0.8664,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647108957197991e-05,
+      "loss": 0.8933,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466356713066913e-05,
+      "loss": 0.7867,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646162136349011e-05,
+      "loss": 0.8143,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645688352507344e-05,
+      "loss": 0.8317,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452143199641782e-05,
+      "loss": 0.8891,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447400389020983e-05,
+      "loss": 0.8496,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442655095037843e-05,
+      "loss": 0.8769,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437907319520123e-05,
+      "loss": 0.7693,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6433157064296532e-05,
+      "loss": 0.8883,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428404331196738e-05,
+      "loss": 0.8851,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6423649122051366e-05,
+      "loss": 0.878,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418891438691992e-05,
+      "loss": 0.8705,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414131282951147e-05,
+      "loss": 0.887,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409368656662308e-05,
+      "loss": 0.8463,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404603561659914e-05,
+      "loss": 0.8506,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399835999779345e-05,
+      "loss": 0.8558,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639506597285694e-05,
+      "loss": 0.8248,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639029348272998e-05,
+      "loss": 0.8572,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385518531236696e-05,
+      "loss": 0.8524,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6380741120216275e-05,
+      "loss": 0.8424,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637596125150884e-05,
+      "loss": 0.9056,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371178926955466e-05,
+      "loss": 0.7612,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366394148398173e-05,
+      "loss": 0.3276,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361606917679926e-05,
+      "loss": 0.3897,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356817236644643e-05,
+      "loss": 0.8865,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6352025107137173e-05,
+      "loss": 0.8208,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634723053100331e-05,
+      "loss": 0.8407,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63424335100898e-05,
+      "loss": 0.8859,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337634046244317e-05,
+      "loss": 0.7901,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6332832141315485e-05,
+      "loss": 0.8699,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328027797152875e-05,
+      "loss": 0.8724,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632322101560698e-05,
+      "loss": 0.9055,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318411798529245e-05,
+      "loss": 0.8678,
+      "step": 1586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313600147772046e-05,
+      "loss": 0.8083,
+      "step": 1587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6308786065188703e-05,
+      "loss": 0.8198,
+      "step": 1588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303969552633464e-05,
+      "loss": 0.8338,
+      "step": 1589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299150611961522e-05,
+      "loss": 0.7576,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6294329245029e-05,
+      "loss": 0.9061,
+      "step": 1591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.628950545369296e-05,
+      "loss": 0.866,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284679239811385e-05,
+      "loss": 0.8608,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279850605243204e-05,
+      "loss": 0.8262,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627501955184829e-05,
+      "loss": 0.9076,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627018608148741e-05,
+      "loss": 0.8218,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265350196022298e-05,
+      "loss": 0.8838,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6260511897315606e-05,
+      "loss": 0.8662,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6255671187230905e-05,
+      "loss": 0.4004,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250828067632715e-05,
+      "loss": 0.8452,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624598254038646e-05,
+      "loss": 0.3948,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241134607358516e-05,
+      "loss": 0.9011,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623628427041617e-05,
+      "loss": 0.8051,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231431531427644e-05,
+      "loss": 0.8886,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622657639226207e-05,
+      "loss": 0.7861,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221718854789526e-05,
+      "loss": 0.8809,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6216858920881e-05,
+      "loss": 0.7969,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621199659240839e-05,
+      "loss": 0.8395,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207131871244558e-05,
+      "loss": 0.8564,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202264759263245e-05,
+      "loss": 0.9406,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197395258339135e-05,
+      "loss": 0.8532,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619252337034782e-05,
+      "loss": 0.3777,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187649097165827e-05,
+      "loss": 0.8531,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182772440670585e-05,
+      "loss": 0.8186,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177893402740457e-05,
+      "loss": 0.8634,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6173011985254707e-05,
+      "loss": 0.7988,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616812819009353e-05,
+      "loss": 0.8665,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163242019138025e-05,
+      "loss": 0.9063,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6158353474270216e-05,
+      "loss": 0.8918,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153462557373032e-05,
+      "loss": 0.8294,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148569270330323e-05,
+      "loss": 0.8955,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614367361502685e-05,
+      "loss": 0.8113,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6138775593348288e-05,
+      "loss": 0.874,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133875207181216e-05,
+      "loss": 0.8614,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128972458413128e-05,
+      "loss": 0.8665,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124067348932434e-05,
+      "loss": 0.8728,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119159880628443e-05,
+      "loss": 0.8693,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114250055391387e-05,
+      "loss": 0.8388,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109337875112384e-05,
+      "loss": 0.8756,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104423341683476e-05,
+      "loss": 0.8425,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099506456997613e-05,
+      "loss": 0.846,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6094587222948634e-05,
+      "loss": 0.7866,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089665641431302e-05,
+      "loss": 0.8529,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084741714341273e-05,
+      "loss": 0.8693,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079815443575107e-05,
+      "loss": 0.8362,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607488683103027e-05,
+      "loss": 0.8271,
+      "step": 1636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069955878605135e-05,
+      "loss": 0.875,
+      "step": 1637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065022588198964e-05,
+      "loss": 0.86,
+      "step": 1638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606008696171192e-05,
+      "loss": 0.8108,
+      "step": 1639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055149001045078e-05,
+      "loss": 0.3902,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050208708100407e-05,
+      "loss": 0.7143,
+      "step": 1641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045266084780772e-05,
+      "loss": 0.8672,
+      "step": 1642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040321132989933e-05,
+      "loss": 0.8866,
+      "step": 1643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035373854632552e-05,
+      "loss": 0.8295,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6030424251614187e-05,
+      "loss": 0.8178,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025472325841284e-05,
+      "loss": 0.8067,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60205180792212e-05,
+      "loss": 0.8616,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015561513662165e-05,
+      "loss": 0.8097,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6010602631073316e-05,
+      "loss": 0.8169,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005641433364682e-05,
+      "loss": 0.9054,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000677922447175e-05,
+      "loss": 0.8489,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995712100232613e-05,
+      "loss": 0.8977,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990743968633682e-05,
+      "loss": 0.8455,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598577352956398e-05,
+      "loss": 0.8614,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598080078493798e-05,
+      "loss": 0.8876,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597582573667105e-05,
+      "loss": 0.8291,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970848386679442e-05,
+      "loss": 0.3514,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965868736880295e-05,
+      "loss": 0.8971,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960886789191632e-05,
+      "loss": 0.8121,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955902545532366e-05,
+      "loss": 0.8185,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950916007822285e-05,
+      "loss": 0.9049,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945927177982076e-05,
+      "loss": 0.8658,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940936057933293e-05,
+      "loss": 0.8461,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935942649598386e-05,
+      "loss": 0.7922,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593094695490067e-05,
+      "loss": 0.8612,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592594897576436e-05,
+      "loss": 0.7863,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592094871411453e-05,
+      "loss": 0.8347,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915946171877155e-05,
+      "loss": 0.8384,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591094135097907e-05,
+      "loss": 0.8264,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905934253347998e-05,
+      "loss": 0.3889,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900924880912535e-05,
+      "loss": 0.406,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895913235602157e-05,
+      "loss": 0.8618,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589089931934721e-05,
+      "loss": 0.8717,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885883134078915e-05,
+      "loss": 0.862,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880864681729376e-05,
+      "loss": 0.8392,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587584396423156e-05,
+      "loss": 0.8021,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587082098351931e-05,
+      "loss": 0.8738,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586579574152734e-05,
+      "loss": 0.8797,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860768240191237e-05,
+      "loss": 0.8599,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855738481447458e-05,
+      "loss": 0.8615,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585070646723333e-05,
+      "loss": 0.8885,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845672199487037e-05,
+      "loss": 0.7862,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5840635680147654e-05,
+      "loss": 0.9293,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5835596911155107e-05,
+      "loss": 0.7562,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830555894450188e-05,
+      "loss": 0.8365,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582551263197456e-05,
+      "loss": 0.8852,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5820467125670757e-05,
+      "loss": 0.8552,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581541937748216e-05,
+      "loss": 0.8395,
+      "step": 1688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581036938935303e-05,
+      "loss": 0.8683,
+      "step": 1689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805317163228483e-05,
+      "loss": 0.8056,
+      "step": 1690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800262701054494e-05,
+      "loss": 0.8555,
+      "step": 1691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579520600477791e-05,
+      "loss": 0.3529,
+      "step": 1692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579014707634643e-05,
+      "loss": 0.8633,
+      "step": 1693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785085917708612e-05,
+      "loss": 0.3765,
+      "step": 1694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780022530813883e-05,
+      "loss": 0.8284,
+      "step": 1695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577495691761251e-05,
+      "loss": 0.7907,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769889080055638e-05,
+      "loss": 0.8543,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764819020095258e-05,
+      "loss": 0.8428,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759746739684213e-05,
+      "loss": 0.8172,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754672240776214e-05,
+      "loss": 0.7899,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749595525325814e-05,
+      "loss": 0.8394,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574451659528843e-05,
+      "loss": 0.8073,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573943545262032e-05,
+      "loss": 0.8199,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573435209927861e-05,
+      "loss": 0.8199,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572926653722126e-05,
+      "loss": 0.7988,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724178768407094e-05,
+      "loss": 0.8406,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719088794795782e-05,
+      "loss": 0.8257,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571399661834784e-05,
+      "loss": 0.897,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708902241024638e-05,
+      "loss": 0.8402,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570380566478839e-05,
+      "loss": 0.8837,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698706891602156e-05,
+      "loss": 0.8279,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693605923429846e-05,
+      "loss": 0.8835,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688502762236207e-05,
+      "loss": 0.8886,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5683397409986852e-05,
+      "loss": 0.8216,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567828986864821e-05,
+      "loss": 0.8508,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567318014018757e-05,
+      "loss": 0.7873,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566806822657306e-05,
+      "loss": 0.8295,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566295412977365e-05,
+      "loss": 0.9164,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565783785175915e-05,
+      "loss": 0.9055,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652719394500207e-05,
+      "loss": 0.8628,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647598759968318e-05,
+      "loss": 0.8097,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642475950135803e-05,
+      "loss": 0.8227,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637350966975832e-05,
+      "loss": 0.8085,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563222381246241e-05,
+      "loss": 0.8761,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627094488570378e-05,
+      "loss": 0.8022,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562196299727541e-05,
+      "loss": 0.906,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561682934055401e-05,
+      "loss": 0.8721,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561169352038353e-05,
+      "loss": 0.8384,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5606555538742144e-05,
+      "loss": 0.8671,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560141539760886e-05,
+      "loss": 0.8367,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596273098963526e-05,
+      "loss": 0.8833,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591128644786813e-05,
+      "loss": 0.8496,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558598203706022e-05,
+      "loss": 0.8826,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558083327776608e-05,
+      "loss": 0.8231,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575682368887563e-05,
+      "loss": 0.3741,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5570529312408643e-05,
+      "loss": 0.374,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556537411031415e-05,
+      "loss": 0.8798,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556021676458972e-05,
+      "loss": 0.8367,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555057277221827e-05,
+      "loss": 0.8753,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554989565019776e-05,
+      "loss": 0.8571,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544731885505636e-05,
+      "loss": 0.8555,
+      "step": 1741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55395659851344e-05,
+      "loss": 0.8692,
+      "step": 1742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534397951073804e-05,
+      "loss": 0.8006,
+      "step": 1743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529227785314455e-05,
+      "loss": 0.7901,
+      "step": 1744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524055489847742e-05,
+      "loss": 0.9017,
+      "step": 1745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518881066665894e-05,
+      "loss": 0.7955,
+      "step": 1746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551370451776196e-05,
+      "loss": 0.815,
+      "step": 1747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508525845129807e-05,
+      "loss": 0.4006,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550334505076412e-05,
+      "loss": 0.8876,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549816213666039e-05,
+      "loss": 0.846,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492977104814943e-05,
+      "loss": 0.8852,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487789957224904e-05,
+      "loss": 0.8875,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548260069588823e-05,
+      "loss": 0.8395,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547740932280367e-05,
+      "loss": 0.8003,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547221583997081e-05,
+      "loss": 0.8591,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467020249390038e-05,
+      "loss": 0.8085,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546182255306254e-05,
+      "loss": 0.8561,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545662275299034e-05,
+      "loss": 0.7987,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451420851176257e-05,
+      "loss": 0.9062,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544621684962391e-05,
+      "loss": 0.8496,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441010750337754e-05,
+      "loss": 0.8613,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435802555323024e-05,
+      "loss": 0.7911,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430592266585787e-05,
+      "loss": 0.8647,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425379886132888e-05,
+      "loss": 0.8132,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420165415972005e-05,
+      "loss": 0.799,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414948858111608e-05,
+      "loss": 0.7383,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5409730214560962e-05,
+      "loss": 0.7989,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404509487330163e-05,
+      "loss": 0.812,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399286678430075e-05,
+      "loss": 0.8653,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394061789872395e-05,
+      "loss": 0.8677,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53888348236696e-05,
+      "loss": 0.3696,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383605781834975e-05,
+      "loss": 0.9026,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378374666382606e-05,
+      "loss": 0.8616,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373141479327375e-05,
+      "loss": 0.7865,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536790622268496e-05,
+      "loss": 0.9614,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362668898471846e-05,
+      "loss": 0.9102,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357429508705297e-05,
+      "loss": 0.793,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535218805540339e-05,
+      "loss": 0.797,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346944540584986e-05,
+      "loss": 0.8759,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341698966269745e-05,
+      "loss": 0.8467,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533645133447812e-05,
+      "loss": 0.8407,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331201647231352e-05,
+      "loss": 0.8507,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532594990655148e-05,
+      "loss": 0.8638,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320696114461327e-05,
+      "loss": 0.9058,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315440272984518e-05,
+      "loss": 0.8752,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310182384145453e-05,
+      "loss": 0.84,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304922449969326e-05,
+      "loss": 0.8128,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529966047248212e-05,
+      "loss": 0.8329,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294396453710613e-05,
+      "loss": 0.8355,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5289130395682352e-05,
+      "loss": 0.8492,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528386230042568e-05,
+      "loss": 0.8288,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278592169969725e-05,
+      "loss": 0.8499,
+      "step": 1792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52733200063444e-05,
+      "loss": 0.8602,
+      "step": 1793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268045811580393e-05,
+      "loss": 0.8054,
+      "step": 1794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526276958770918e-05,
+      "loss": 0.8179,
+      "step": 1795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257491336763023e-05,
+      "loss": 0.3407,
+      "step": 1796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252211060774956e-05,
+      "loss": 0.8699,
+      "step": 1797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52469287617788e-05,
+      "loss": 0.8378,
+      "step": 1798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241644441809146e-05,
+      "loss": 0.8647,
+      "step": 1799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236358102901375e-05,
+      "loss": 0.8785,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523106974709164e-05,
+      "loss": 0.8717,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225779376416871e-05,
+      "loss": 0.8637,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522048699291477e-05,
+      "loss": 0.8589,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215192598623825e-05,
+      "loss": 0.3659,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209896195583284e-05,
+      "loss": 0.856,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5204597785833182e-05,
+      "loss": 0.8001,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199297371414324e-05,
+      "loss": 0.8191,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193994954368277e-05,
+      "loss": 0.9114,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188690536737394e-05,
+      "loss": 0.8073,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183384120564788e-05,
+      "loss": 0.8694,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178075707894346e-05,
+      "loss": 0.8244,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5172765300770724e-05,
+      "loss": 0.9008,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167452901239346e-05,
+      "loss": 0.8237,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51621385113464e-05,
+      "loss": 0.8312,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156822133138852e-05,
+      "loss": 0.8742,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151503768664416e-05,
+      "loss": 0.891,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514618341997158e-05,
+      "loss": 0.8361,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140861089109609e-05,
+      "loss": 0.8767,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135536778128504e-05,
+      "loss": 0.3561,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5130210489079055e-05,
+      "loss": 0.8534,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124882224012802e-05,
+      "loss": 0.8001,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119551984982042e-05,
+      "loss": 0.8341,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114219774039836e-05,
+      "loss": 0.8366,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108885593240015e-05,
+      "loss": 0.8291,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103549444637151e-05,
+      "loss": 0.7878,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098211330286585e-05,
+      "loss": 0.797,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092871252244416e-05,
+      "loss": 0.8596,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508752921256749e-05,
+      "loss": 0.8318,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082185213313423e-05,
+      "loss": 0.9167,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076839256540568e-05,
+      "loss": 0.89,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071491344308048e-05,
+      "loss": 0.7831,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066141478675729e-05,
+      "loss": 0.8637,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060789661704234e-05,
+      "loss": 0.8397,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055435895454936e-05,
+      "loss": 0.9033,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050080181989966e-05,
+      "loss": 0.8327,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044722523372187e-05,
+      "loss": 0.7576,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503936292166523e-05,
+      "loss": 0.8899,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034001378933463e-05,
+      "loss": 0.85,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028637897242011e-05,
+      "loss": 0.3882,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023272478656737e-05,
+      "loss": 0.8115,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017905125244254e-05,
+      "loss": 0.916,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012535839071918e-05,
+      "loss": 0.8923,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007164622207832e-05,
+      "loss": 0.8923,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001791476720845e-05,
+      "loss": 0.8188,
+      "step": 1844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996416404680543e-05,
+      "loss": 0.8191,
+      "step": 1845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499103940815726e-05,
+      "loss": 0.8074,
+      "step": 1846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985660489222063e-05,
+      "loss": 0.8548,
+      "step": 1847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980279649946765e-05,
+      "loss": 0.8313,
+      "step": 1848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974896892403923e-05,
+      "loss": 0.8421,
+      "step": 1849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969512218666821e-05,
+      "loss": 0.8358,
+      "step": 1850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964125630809496e-05,
+      "loss": 0.9068,
+      "step": 1851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495873713090671e-05,
+      "loss": 0.81,
+      "step": 1852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4953346721033966e-05,
+      "loss": 0.818,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947954403267501e-05,
+      "loss": 0.8293,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942560179684293e-05,
+      "loss": 0.8943,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937164052362047e-05,
+      "loss": 0.8838,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931766023379202e-05,
+      "loss": 0.8726,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492636609481493e-05,
+      "loss": 0.8057,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920964268749141e-05,
+      "loss": 0.8705,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915560547262469e-05,
+      "loss": 0.8331,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910154932436278e-05,
+      "loss": 0.8537,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904747426352662e-05,
+      "loss": 0.8061,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899338031094448e-05,
+      "loss": 0.8769,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893926748745183e-05,
+      "loss": 0.8167,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488851358138915e-05,
+      "loss": 0.836,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883098531111352e-05,
+      "loss": 0.8357,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877681599997514e-05,
+      "loss": 0.9003,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872262790134091e-05,
+      "loss": 0.8598,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4866842103608265e-05,
+      "loss": 0.7058,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861419542507931e-05,
+      "loss": 0.8518,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855995108921714e-05,
+      "loss": 0.8626,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850568804938958e-05,
+      "loss": 0.8308,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845140632649729e-05,
+      "loss": 0.8651,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839710594144803e-05,
+      "loss": 0.8167,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834278691515687e-05,
+      "loss": 0.8683,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48288449268546e-05,
+      "loss": 0.8896,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823409302254485e-05,
+      "loss": 0.8885,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817971819808987e-05,
+      "loss": 0.8905,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812532481612483e-05,
+      "loss": 0.8673,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807091289760056e-05,
+      "loss": 0.8098,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48016482463475e-05,
+      "loss": 0.8839,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479620335347133e-05,
+      "loss": 0.852,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479075661322877e-05,
+      "loss": 0.7881,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785308027717755e-05,
+      "loss": 0.9067,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779857599036931e-05,
+      "loss": 0.8319,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774405329285655e-05,
+      "loss": 0.8564,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768951220563988e-05,
+      "loss": 0.888,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763495274972709e-05,
+      "loss": 0.7799,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758037494613295e-05,
+      "loss": 0.8649,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752577881587935e-05,
+      "loss": 0.3911,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747116437999526e-05,
+      "loss": 0.7875,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741653165951664e-05,
+      "loss": 0.821,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736188067548648e-05,
+      "loss": 0.8373,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730721144895494e-05,
+      "loss": 0.8331,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725252400097903e-05,
+      "loss": 0.851,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719781835262288e-05,
+      "loss": 0.8706,
+      "step": 1896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714309452495761e-05,
+      "loss": 0.8998,
+      "step": 1897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708835253906135e-05,
+      "loss": 0.8181,
+      "step": 1898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703359241601921e-05,
+      "loss": 0.773,
+      "step": 1899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697881417692333e-05,
+      "loss": 0.848,
+      "step": 1900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692401784287274e-05,
+      "loss": 0.8827,
+      "step": 1901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686920343497349e-05,
+      "loss": 0.8166,
+      "step": 1902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681437097433862e-05,
+      "loss": 0.8429,
+      "step": 1903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675952048208806e-05,
+      "loss": 0.8624,
+      "step": 1904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670465197934875e-05,
+      "loss": 0.9061,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664976548725453e-05,
+      "loss": 0.8475,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659486102694618e-05,
+      "loss": 0.8653,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653993861957139e-05,
+      "loss": 0.9031,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648499828628476e-05,
+      "loss": 0.8854,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464300400482478e-05,
+      "loss": 0.8628,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637506392662899e-05,
+      "loss": 0.7718,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632006994260353e-05,
+      "loss": 0.8431,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626505811735365e-05,
+      "loss": 0.8674,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462100284720684e-05,
+      "loss": 0.3723,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615498102794367e-05,
+      "loss": 0.8293,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4609991580618228e-05,
+      "loss": 0.8555,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460448328279938e-05,
+      "loss": 0.7862,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598973211459471e-05,
+      "loss": 0.8779,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459346136872083e-05,
+      "loss": 0.8062,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587947756706465e-05,
+      "loss": 0.8466,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458243237754007e-05,
+      "loss": 0.8529,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576915233346022e-05,
+      "loss": 0.3743,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571396326249371e-05,
+      "loss": 0.8615,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565875658375847e-05,
+      "loss": 0.8466,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560353231851868e-05,
+      "loss": 0.8574,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554829048804508e-05,
+      "loss": 0.7948,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549303111361544e-05,
+      "loss": 0.8291,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543775421651413e-05,
+      "loss": 0.8427,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538245981803227e-05,
+      "loss": 0.8689,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4532714793946779e-05,
+      "loss": 0.854,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4527181860212524e-05,
+      "loss": 0.8158,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4521647182731606e-05,
+      "loss": 0.8296,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516110763635827e-05,
+      "loss": 0.8445,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510572605057662e-05,
+      "loss": 0.8547,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450503270913026e-05,
+      "loss": 0.7934,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499491077987443e-05,
+      "loss": 0.8286,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493947713763687e-05,
+      "loss": 0.8282,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4488402618594149e-05,
+      "loss": 0.8565,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482855794614647e-05,
+      "loss": 0.8843,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477307243961667e-05,
+      "loss": 0.8487,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471756968772357e-05,
+      "loss": 0.8379,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466204971184531e-05,
+      "loss": 0.861,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460651253336673e-05,
+      "loss": 0.8367,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455095817367915e-05,
+      "loss": 0.8805,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4449538665418064e-05,
+      "loss": 0.3786,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444397979962758e-05,
+      "loss": 0.8196,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438419222137588e-05,
+      "loss": 0.8407,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432856935089869e-05,
+      "loss": 0.7773,
+      "step": 1948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427292940626867e-05,
+      "loss": 0.8371,
+      "step": 1949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4421727240891676e-05,
+      "loss": 0.8887,
+      "step": 1950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4416159838028054e-05,
+      "loss": 0.8198,
+      "step": 1951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4410590734180411e-05,
+      "loss": 0.7981,
+      "step": 1952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405019931493818e-05,
+      "loss": 0.8497,
+      "step": 1953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399447432113988e-05,
+      "loss": 0.8703,
+      "step": 1954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393873238187304e-05,
+      "loss": 0.8479,
+      "step": 1955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388297351860789e-05,
+      "loss": 0.8348,
+      "step": 1956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382719775282119e-05,
+      "loss": 0.8804,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377140510599627e-05,
+      "loss": 0.8192,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4371559559962296e-05,
+      "loss": 0.7981,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365976925519749e-05,
+      "loss": 0.8414,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360392609422271e-05,
+      "loss": 0.8366,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354806613820782e-05,
+      "loss": 0.8558,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349218940866858e-05,
+      "loss": 0.7914,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434362959271272e-05,
+      "loss": 0.8807,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338038571511225e-05,
+      "loss": 0.8361,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332445879415888e-05,
+      "loss": 0.3516,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432685151858086e-05,
+      "loss": 0.8906,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321255491160935e-05,
+      "loss": 0.8408,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315657799311548e-05,
+      "loss": 0.811,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431005844518878e-05,
+      "loss": 0.8152,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304457430949343e-05,
+      "loss": 0.8468,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298854758750605e-05,
+      "loss": 0.9077,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293250430750549e-05,
+      "loss": 0.8821,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287644449107823e-05,
+      "loss": 0.8603,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4282036815981688e-05,
+      "loss": 0.8702,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276427533532054e-05,
+      "loss": 0.8316,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4270816603919461e-05,
+      "loss": 0.7948,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265204029305091e-05,
+      "loss": 0.8168,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259589811850749e-05,
+      "loss": 0.8219,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425397395371888e-05,
+      "loss": 0.8813,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424835645707256e-05,
+      "loss": 0.8264,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242737324075495e-05,
+      "loss": 0.8999,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423711655689202e-05,
+      "loss": 0.9051,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42314941576871e-05,
+      "loss": 0.8327,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422587012862633e-05,
+      "loss": 0.8531,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220244471875936e-05,
+      "loss": 0.8181,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214617189602765e-05,
+      "loss": 0.8625,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208988283974288e-05,
+      "loss": 0.8158,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420335775715861e-05,
+      "loss": 0.8244,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197725611324455e-05,
+      "loss": 0.8316,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192091848641174e-05,
+      "loss": 0.9076,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186456471278734e-05,
+      "loss": 0.8707,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418081948140773e-05,
+      "loss": 0.842,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175180881199378e-05,
+      "loss": 0.8424,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169540672825509e-05,
+      "loss": 0.8429,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416389885845858e-05,
+      "loss": 0.8518,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158255440271664e-05,
+      "loss": 0.8329,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152610420438449e-05,
+      "loss": 0.815,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146963801133242e-05,
+      "loss": 0.821,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141315584530973e-05,
+      "loss": 0.8258,
+      "step": 2000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413566577280717e-05,
+      "loss": 0.8393,
+      "step": 2001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130014368137993e-05,
+      "loss": 0.8571,
+      "step": 2002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412436137270021e-05,
+      "loss": 0.8952,
+      "step": 2003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118706788671192e-05,
+      "loss": 0.8556,
+      "step": 2004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411305061822894e-05,
+      "loss": 0.3827,
+      "step": 2005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107392863552051e-05,
+      "loss": 0.8425,
+      "step": 2006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101733526819737e-05,
+      "loss": 0.8675,
+      "step": 2007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096072610211821e-05,
+      "loss": 0.8558,
+      "step": 2008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090410115908734e-05,
+      "loss": 0.8505,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084746046091513e-05,
+      "loss": 0.9349,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40790804029418e-05,
+      "loss": 0.8414,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073413188641854e-05,
+      "loss": 0.841,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067744405374522e-05,
+      "loss": 0.8871,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406207405532327e-05,
+      "loss": 0.3542,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4056402140672159e-05,
+      "loss": 0.8337,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050728663605852e-05,
+      "loss": 0.8034,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045053626309631e-05,
+      "loss": 0.8175,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039377030969346e-05,
+      "loss": 0.8094,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403369887977148e-05,
+      "loss": 0.9249,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4028019174903102e-05,
+      "loss": 0.8945,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402233791855187e-05,
+      "loss": 0.8594,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016655112906052e-05,
+      "loss": 0.8476,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010970760154514e-05,
+      "loss": 0.8631,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005284862486708e-05,
+      "loss": 0.852,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999597422092688e-05,
+      "loss": 0.8633,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993908441163103e-05,
+      "loss": 0.8503,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988217921889188e-05,
+      "loss": 0.8349,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982525866462779e-05,
+      "loss": 0.8533,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39768322770763e-05,
+      "loss": 0.8635,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971137155922762e-05,
+      "loss": 0.91,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965440505195779e-05,
+      "loss": 0.8897,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959742327089536e-05,
+      "loss": 0.8372,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954042623798815e-05,
+      "loss": 0.8939,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948341397518993e-05,
+      "loss": 0.8882,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942638650446023e-05,
+      "loss": 0.7642,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936934384776446e-05,
+      "loss": 0.7926,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931228602707391e-05,
+      "loss": 0.8464,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392552130643657e-05,
+      "loss": 0.801,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919812498162273e-05,
+      "loss": 0.8624,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914102180083382e-05,
+      "loss": 0.822,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908390354399352e-05,
+      "loss": 0.871,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902677023310225e-05,
+      "loss": 0.8809,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3896962189016619e-05,
+      "loss": 0.8841,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891245853719727e-05,
+      "loss": 0.8065,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885528019621333e-05,
+      "loss": 0.8594,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3879808688923786e-05,
+      "loss": 0.8318,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874087863830013e-05,
+      "loss": 0.801,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868365546543527e-05,
+      "loss": 0.8135,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38626417392684e-05,
+      "loss": 0.8536,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856916444209294e-05,
+      "loss": 0.8761,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851189663571432e-05,
+      "loss": 0.3618,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384546139956061e-05,
+      "loss": 0.8384,
+      "step": 2052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839731654383203e-05,
+      "loss": 0.8171,
+      "step": 2053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834000430246155e-05,
+      "loss": 0.8727,
+      "step": 2054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382826772935697e-05,
+      "loss": 0.8381,
+      "step": 2055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822533553923732e-05,
+      "loss": 0.7593,
+      "step": 2056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816797906155086e-05,
+      "loss": 0.8746,
+      "step": 2057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381106078826025e-05,
+      "loss": 0.7824,
+      "step": 2058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805322202448999e-05,
+      "loss": 0.3515,
+      "step": 2059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799582150931684e-05,
+      "loss": 0.8554,
+      "step": 2060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793840635919217e-05,
+      "loss": 0.7949,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3788097659623073e-05,
+      "loss": 0.7822,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782353224255283e-05,
+      "loss": 0.8597,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776607332028454e-05,
+      "loss": 0.7989,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770859985155744e-05,
+      "loss": 0.3556,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765111185850868e-05,
+      "loss": 0.7718,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759360936328117e-05,
+      "loss": 0.8068,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753609238802321e-05,
+      "loss": 0.7722,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747856095488881e-05,
+      "loss": 0.8526,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742101508603748e-05,
+      "loss": 0.8715,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736345480363435e-05,
+      "loss": 0.3505,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730588012985004e-05,
+      "loss": 0.8536,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724829108686076e-05,
+      "loss": 0.8246,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371906876968482e-05,
+      "loss": 0.8364,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713306998199962e-05,
+      "loss": 0.7967,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707543796450784e-05,
+      "loss": 0.8353,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701779166657108e-05,
+      "loss": 0.8355,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369601311103931e-05,
+      "loss": 0.8586,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690245631818326e-05,
+      "loss": 0.8631,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684476731215622e-05,
+      "loss": 0.7983,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678706411453227e-05,
+      "loss": 0.8138,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672934674753705e-05,
+      "loss": 0.8195,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667161523340177e-05,
+      "loss": 0.7739,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36613869594363e-05,
+      "loss": 0.8144,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655610985266279e-05,
+      "loss": 0.8179,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364983360305486e-05,
+      "loss": 0.8099,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644054815027333e-05,
+      "loss": 0.8758,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638274623409529e-05,
+      "loss": 0.8157,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363249303042782e-05,
+      "loss": 0.777,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626710038309118e-05,
+      "loss": 0.8052,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620925649280873e-05,
+      "loss": 0.9001,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3615139865571074e-05,
+      "loss": 0.8967,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609352689408248e-05,
+      "loss": 0.7621,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603564123021454e-05,
+      "loss": 0.812,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359777416864029e-05,
+      "loss": 0.831,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591982828494891e-05,
+      "loss": 0.7856,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586190104815919e-05,
+      "loss": 0.7996,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580395999834575e-05,
+      "loss": 0.8185,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357460051578259e-05,
+      "loss": 0.8511,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3568803654892227e-05,
+      "loss": 0.8576,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563005419396276e-05,
+      "loss": 0.8555,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557205811528059e-05,
+      "loss": 0.8844,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551404833521425e-05,
+      "loss": 0.8598,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545602487610755e-05,
+      "loss": 0.8649,
+      "step": 2104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539798776030954e-05,
+      "loss": 0.8023,
+      "step": 2105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353399370101745e-05,
+      "loss": 0.8915,
+      "step": 2106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528187264806208e-05,
+      "loss": 0.8258,
+      "step": 2107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522379469633694e-05,
+      "loss": 0.3468,
+      "step": 2108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351657031773692e-05,
+      "loss": 0.8095,
+      "step": 2109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510759811353418e-05,
+      "loss": 0.8494,
+      "step": 2110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504947952721226e-05,
+      "loss": 0.7548,
+      "step": 2111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499134744078917e-05,
+      "loss": 0.8822,
+      "step": 2112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493320187665582e-05,
+      "loss": 0.8587,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3487504285720827e-05,
+      "loss": 0.8359,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348168704048478e-05,
+      "loss": 0.8199,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347586845419808e-05,
+      "loss": 0.762,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470048529101893e-05,
+      "loss": 0.9259,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464227267437893e-05,
+      "loss": 0.797,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458404671448273e-05,
+      "loss": 0.8314,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345258074337573e-05,
+      "loss": 0.8633,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446755485463495e-05,
+      "loss": 0.8507,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440928899955285e-05,
+      "loss": 0.7885,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435100989095346e-05,
+      "loss": 0.8142,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3429271755128437e-05,
+      "loss": 0.7953,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3423441200299807e-05,
+      "loss": 0.893,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417609326855237e-05,
+      "loss": 0.9076,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411776137041002e-05,
+      "loss": 0.8384,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340594163310388e-05,
+      "loss": 0.8354,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400105817291174e-05,
+      "loss": 0.8366,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339426869185067e-05,
+      "loss": 0.8495,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338843025903068e-05,
+      "loss": 0.8838,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382590521079999e-05,
+      "loss": 0.8135,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376749480247941e-05,
+      "loss": 0.8596,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337090713878431e-05,
+      "loss": 0.7941,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365063498939422e-05,
+      "loss": 0.8312,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335921856296408e-05,
+      "loss": 0.7828,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33533723331096e-05,
+      "loss": 0.8742,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347524811627788e-05,
+      "loss": 0.8829,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341676000770946e-05,
+      "loss": 0.3498,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335825902791881e-05,
+      "loss": 0.7706,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329974519943892e-05,
+      "loss": 0.8935,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324121854480763e-05,
+      "loss": 0.8417,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331826790865679e-05,
+      "loss": 0.8282,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312412684726743e-05,
+      "loss": 0.8006,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306556184945904e-05,
+      "loss": 0.8118,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300698411570035e-05,
+      "loss": 0.8742,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294839366855386e-05,
+      "loss": 0.8559,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288979053058701e-05,
+      "loss": 0.7807,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283117472437215e-05,
+      "loss": 0.8659,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3277254627248648e-05,
+      "loss": 0.8576,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271390519751206e-05,
+      "loss": 0.8718,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326552515220359e-05,
+      "loss": 0.8293,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259658526864967e-05,
+      "loss": 0.8128,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253790645995011e-05,
+      "loss": 0.8106,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247921511853865e-05,
+      "loss": 0.8339,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242051126702155e-05,
+      "loss": 0.7964,
+      "step": 2156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236179492801005e-05,
+      "loss": 0.7963,
+      "step": 2157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230306612411994e-05,
+      "loss": 0.863,
+      "step": 2158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3224432487797203e-05,
+      "loss": 0.8384,
+      "step": 2159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218557121219186e-05,
+      "loss": 0.8545,
+      "step": 2160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212680514940965e-05,
+      "loss": 0.8136,
+      "step": 2161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3206802671226054e-05,
+      "loss": 0.9162,
+      "step": 2162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200923592338436e-05,
+      "loss": 0.8617,
+      "step": 2163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3195043280542575e-05,
+      "loss": 0.8441,
+      "step": 2164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3189161738103401e-05,
+      "loss": 0.8258,
+      "step": 2165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3183278967286328e-05,
+      "loss": 0.9148,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3177394970357233e-05,
+      "loss": 0.8618,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171509749582476e-05,
+      "loss": 0.8282,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165623307228882e-05,
+      "loss": 0.88,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159735645563743e-05,
+      "loss": 0.8149,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153846766854831e-05,
+      "loss": 0.8103,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147956673370381e-05,
+      "loss": 0.8844,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3142065367379089e-05,
+      "loss": 0.8341,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136172851150137e-05,
+      "loss": 0.802,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130279126953147e-05,
+      "loss": 0.3889,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3124384197058231e-05,
+      "loss": 0.8268,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311848806373595e-05,
+      "loss": 0.8029,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3112590729257337e-05,
+      "loss": 0.8402,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310669219589388e-05,
+      "loss": 0.8232,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100792465917535e-05,
+      "loss": 0.8467,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094891541600716e-05,
+      "loss": 0.8193,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3088989425216299e-05,
+      "loss": 0.8462,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083086119037613e-05,
+      "loss": 0.8136,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077181625338455e-05,
+      "loss": 0.8712,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071275946393076e-05,
+      "loss": 0.877,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065369084476179e-05,
+      "loss": 0.9001,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059461041862922e-05,
+      "loss": 0.8615,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053551820828932e-05,
+      "loss": 0.3568,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047641423650267e-05,
+      "loss": 0.3797,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3041729852603458e-05,
+      "loss": 0.3727,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035817109965483e-05,
+      "loss": 0.7658,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3029903198013759e-05,
+      "loss": 0.8284,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023988119026169e-05,
+      "loss": 0.8715,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018071875281043e-05,
+      "loss": 0.8024,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301215446905715e-05,
+      "loss": 0.34,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006235902633716e-05,
+      "loss": 0.8807,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000316178290414e-05,
+      "loss": 0.935,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2994395298307354e-05,
+      "loss": 0.8721,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29884732649651e-05,
+      "loss": 0.8135,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982550080544658e-05,
+      "loss": 0.3701,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2976625747327473e-05,
+      "loss": 0.8761,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970700267595444e-05,
+      "loss": 0.8278,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964773643630895e-05,
+      "loss": 0.8887,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958845877716604e-05,
+      "loss": 0.8708,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952916972135786e-05,
+      "loss": 0.8517,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946986929172087e-05,
+      "loss": 0.8217,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941055751109604e-05,
+      "loss": 0.7972,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935123440232862e-05,
+      "loss": 0.8132,
+      "step": 2208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929189998826826e-05,
+      "loss": 0.3757,
+      "step": 2209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923255429176892e-05,
+      "loss": 0.8532,
+      "step": 2210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917319733568899e-05,
+      "loss": 0.8239,
+      "step": 2211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291138291428911e-05,
+      "loss": 0.7981,
+      "step": 2212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905444973624227e-05,
+      "loss": 0.8994,
+      "step": 2213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2899505913861385e-05,
+      "loss": 0.8732,
+      "step": 2214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289356573728814e-05,
+      "loss": 0.8175,
+      "step": 2215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887624446192497e-05,
+      "loss": 0.8562,
+      "step": 2216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288168204286287e-05,
+      "loss": 0.8396,
+      "step": 2217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2875738529588112e-05,
+      "loss": 0.8234,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869793908657507e-05,
+      "loss": 0.7796,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2863848182360752e-05,
+      "loss": 0.805,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857901352987983e-05,
+      "loss": 0.8174,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851953422829757e-05,
+      "loss": 0.8445,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846004394177051e-05,
+      "loss": 0.8587,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2840054269321271e-05,
+      "loss": 0.7916,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834103050554241e-05,
+      "loss": 0.8441,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828150740168208e-05,
+      "loss": 0.8029,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282219734045584e-05,
+      "loss": 0.8261,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281624285371022e-05,
+      "loss": 0.8087,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810287282224859e-05,
+      "loss": 0.8733,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804330628293676e-05,
+      "loss": 0.8251,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798372894211015e-05,
+      "loss": 0.8138,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792414082271627e-05,
+      "loss": 0.85,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278645419477069e-05,
+      "loss": 0.8217,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780493234003782e-05,
+      "loss": 0.8619,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774531202266908e-05,
+      "loss": 0.8217,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768568101856482e-05,
+      "loss": 0.801,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762603935069317e-05,
+      "loss": 0.8466,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756638704202657e-05,
+      "loss": 0.8581,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750672411554144e-05,
+      "loss": 0.7659,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2744705059421825e-05,
+      "loss": 0.8217,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738736650104166e-05,
+      "loss": 0.8679,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732767185900033e-05,
+      "loss": 0.8273,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2726796669108703e-05,
+      "loss": 0.8152,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720825102029852e-05,
+      "loss": 0.8069,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714852486963568e-05,
+      "loss": 0.824,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708878826210334e-05,
+      "loss": 0.8523,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702904122071049e-05,
+      "loss": 0.7759,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696928376846998e-05,
+      "loss": 0.7669,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690951592839878e-05,
+      "loss": 0.7415,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684973772351783e-05,
+      "loss": 0.8227,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678994917685204e-05,
+      "loss": 0.8722,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673015031143035e-05,
+      "loss": 0.8684,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266703411502856e-05,
+      "loss": 0.8322,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661052171645472e-05,
+      "loss": 0.8566,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655069203297845e-05,
+      "loss": 0.8766,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264908521229016e-05,
+      "loss": 0.822,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643100200927283e-05,
+      "loss": 0.8892,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637114171514478e-05,
+      "loss": 0.7939,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26311271263574e-05,
+      "loss": 0.8451,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262513906776209e-05,
+      "loss": 0.7983,
+      "step": 2260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619149998034998e-05,
+      "loss": 0.8614,
+      "step": 2261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613159919482937e-05,
+      "loss": 0.8689,
+      "step": 2262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607168834413126e-05,
+      "loss": 0.808,
+      "step": 2263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601176745133171e-05,
+      "loss": 0.7733,
+      "step": 2264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595183653951048e-05,
+      "loss": 0.8106,
+      "step": 2265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589189563175149e-05,
+      "loss": 0.7977,
+      "step": 2266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583194475114223e-05,
+      "loss": 0.799,
+      "step": 2267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2577198392077414e-05,
+      "loss": 0.8767,
+      "step": 2268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571201316374255e-05,
+      "loss": 0.7096,
+      "step": 2269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565203250314653e-05,
+      "loss": 0.8255,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559204196208896e-05,
+      "loss": 0.828,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255320415636766e-05,
+      "loss": 0.8544,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547203133101996e-05,
+      "loss": 0.8584,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541201128723328e-05,
+      "loss": 0.862,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253519814554347e-05,
+      "loss": 0.8387,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529194185874606e-05,
+      "loss": 0.7358,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523189252029296e-05,
+      "loss": 0.8517,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2517183346320479e-05,
+      "loss": 0.8399,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2511176471061462e-05,
+      "loss": 0.3863,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250516862856593e-05,
+      "loss": 0.8896,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499159821147942e-05,
+      "loss": 0.8383,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493150051121922e-05,
+      "loss": 0.8803,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2487139320802673e-05,
+      "loss": 0.8284,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481127632505364e-05,
+      "loss": 0.3564,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247511498854553e-05,
+      "loss": 0.8882,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246910139123908e-05,
+      "loss": 0.8076,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463086842902285e-05,
+      "loss": 0.8183,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457071345851786e-05,
+      "loss": 0.8567,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451054902404586e-05,
+      "loss": 0.8088,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445037514878057e-05,
+      "loss": 0.8444,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243901918558993e-05,
+      "loss": 0.8167,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432999916858306e-05,
+      "loss": 0.8378,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426979711001637e-05,
+      "loss": 0.7824,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420958570338739e-05,
+      "loss": 0.8397,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414936497188803e-05,
+      "loss": 0.8437,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408913493871353e-05,
+      "loss": 0.7989,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402889562706294e-05,
+      "loss": 0.3727,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396864706013874e-05,
+      "loss": 0.8677,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390838926114709e-05,
+      "loss": 0.8325,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384812225329761e-05,
+      "loss": 0.8404,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378784605980349e-05,
+      "loss": 0.8335,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2372756070388152e-05,
+      "loss": 0.3442,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366726620875194e-05,
+      "loss": 0.8045,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360696259763852e-05,
+      "loss": 0.8216,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354664989376856e-05,
+      "loss": 0.8146,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348632812037296e-05,
+      "loss": 0.8374,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342599730068592e-05,
+      "loss": 0.7937,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336565745794521e-05,
+      "loss": 0.7837,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2330530861539219e-05,
+      "loss": 0.8228,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324495079627148e-05,
+      "loss": 0.3676,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318458402383131e-05,
+      "loss": 0.8312,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312420832132332e-05,
+      "loss": 0.7439,
+      "step": 2312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2306382371200253e-05,
+      "loss": 0.7874,
+      "step": 2313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300343021912748e-05,
+      "loss": 0.8276,
+      "step": 2314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229430278659601e-05,
+      "loss": 0.3788,
+      "step": 2315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288261667576572e-05,
+      "loss": 0.8212,
+      "step": 2316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2282219667181306e-05,
+      "loss": 0.8628,
+      "step": 2317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276176787737424e-05,
+      "loss": 0.7967,
+      "step": 2318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270133031572482e-05,
+      "loss": 0.8299,
+      "step": 2319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2264088401014366e-05,
+      "loss": 0.8388,
+      "step": 2320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258042898391306e-05,
+      "loss": 0.3708,
+      "step": 2321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251996526031861e-05,
+      "loss": 0.7975,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245949286264933e-05,
+      "loss": 0.7821,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239901181419747e-05,
+      "loss": 0.8178,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233852213825872e-05,
+      "loss": 0.8359,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227802385813206e-05,
+      "loss": 0.8349,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222175169971197e-05,
+      "loss": 0.844,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215700157852732e-05,
+      "loss": 0.8031,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209647762566374e-05,
+      "loss": 0.7977,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203594516184116e-05,
+      "loss": 0.8474,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197540421037505e-05,
+      "loss": 0.8382,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191485479458406e-05,
+      "loss": 0.7868,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185429693779025e-05,
+      "loss": 0.8294,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217937306633188e-05,
+      "loss": 0.3209,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173315599449821e-05,
+      "loss": 0.8393,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167257295466016e-05,
+      "loss": 0.8328,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2161198156713961e-05,
+      "loss": 0.3585,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155138185527467e-05,
+      "loss": 0.7154,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149077384240673e-05,
+      "loss": 0.8912,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214301575518803e-05,
+      "loss": 0.8075,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136953300704315e-05,
+      "loss": 0.7998,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213089002312462e-05,
+      "loss": 0.8261,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124825924784349e-05,
+      "loss": 0.8742,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211876100801923e-05,
+      "loss": 0.819,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112695275165309e-05,
+      "loss": 0.8481,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106628728558929e-05,
+      "loss": 0.843,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100561370536767e-05,
+      "loss": 0.7419,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20944932034358e-05,
+      "loss": 0.7885,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088424229593318e-05,
+      "loss": 0.8194,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082354451346928e-05,
+      "loss": 0.7888,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2076283871034537e-05,
+      "loss": 0.8737,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070212490994374e-05,
+      "loss": 0.8588,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064140313564963e-05,
+      "loss": 0.8592,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058067341085142e-05,
+      "loss": 0.8131,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051993575894052e-05,
+      "loss": 0.864,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204591902033115e-05,
+      "loss": 0.8707,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039843676736177e-05,
+      "loss": 0.8613,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033767547449196e-05,
+      "loss": 0.8769,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2027690634810565e-05,
+      "loss": 0.764,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202161294116094e-05,
+      "loss": 0.8423,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015534468841293e-05,
+      "loss": 0.7995,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009455220192873e-05,
+      "loss": 0.8676,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200337519755725e-05,
+      "loss": 0.8345,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997294403276275e-05,
+      "loss": 0.7561,
+      "step": 2364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199121283969211e-05,
+      "loss": 0.8096,
+      "step": 2365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985130509147205e-05,
+      "loss": 0.9007,
+      "step": 2366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979047413984306e-05,
+      "loss": 0.8012,
+      "step": 2367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972963556546453e-05,
+      "loss": 0.8788,
+      "step": 2368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966878939176985e-05,
+      "loss": 0.8176,
+      "step": 2369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960793564219532e-05,
+      "loss": 0.7955,
+      "step": 2370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954707434018008e-05,
+      "loss": 0.3809,
+      "step": 2371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194862055091663e-05,
+      "loss": 0.8031,
+      "step": 2372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942532917259897e-05,
+      "loss": 0.7823,
+      "step": 2373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936444535392592e-05,
+      "loss": 0.8636,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1930355407659805e-05,
+      "loss": 0.8843,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924265536406898e-05,
+      "loss": 0.8184,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918174923979517e-05,
+      "loss": 0.8004,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1912083572723603e-05,
+      "loss": 0.7906,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905991484985381e-05,
+      "loss": 0.7688,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899898663111353e-05,
+      "loss": 0.8281,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893805109448308e-05,
+      "loss": 0.7767,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887710826343316e-05,
+      "loss": 0.8101,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188161581614373e-05,
+      "loss": 0.8283,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1875520081197178e-05,
+      "loss": 0.8772,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869423623851575e-05,
+      "loss": 0.8809,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863326446455103e-05,
+      "loss": 0.8511,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857228551356236e-05,
+      "loss": 0.8461,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851129940903712e-05,
+      "loss": 0.8365,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845030617446549e-05,
+      "loss": 0.8183,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838930583334043e-05,
+      "loss": 0.8835,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832829840915752e-05,
+      "loss": 0.815,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826728392541524e-05,
+      "loss": 0.8524,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1820626240561468e-05,
+      "loss": 0.8998,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1814523387325962e-05,
+      "loss": 0.8591,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808419835185662e-05,
+      "loss": 0.8476,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802315586491491e-05,
+      "loss": 0.7941,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1796210643594634e-05,
+      "loss": 0.8445,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179010500884655e-05,
+      "loss": 0.8153,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783998684598966e-05,
+      "loss": 0.8406,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777891673203863e-05,
+      "loss": 0.8439,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1771783977013504e-05,
+      "loss": 0.8355,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765675598380401e-05,
+      "loss": 0.8275,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759566539657334e-05,
+      "loss": 0.7858,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753456803197352e-05,
+      "loss": 0.8751,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747346391353746e-05,
+      "loss": 0.8223,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741235306480093e-05,
+      "loss": 0.8467,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173512355093021e-05,
+      "loss": 0.8345,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729011127058177e-05,
+      "loss": 0.8822,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722898037218332e-05,
+      "loss": 0.8376,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716784283765272e-05,
+      "loss": 0.8476,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171066986905385e-05,
+      "loss": 0.8291,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1704554795439166e-05,
+      "loss": 0.8492,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1698439065276584e-05,
+      "loss": 0.8466,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692322680921712e-05,
+      "loss": 0.7927,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1686205644730417e-05,
+      "loss": 0.8044,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680087959058814e-05,
+      "loss": 0.8046,
+      "step": 2416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673969626263265e-05,
+      "loss": 0.7966,
+      "step": 2417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166785064870039e-05,
+      "loss": 0.8436,
+      "step": 2418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661731028727043e-05,
+      "loss": 0.8232,
+      "step": 2419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655610768700343e-05,
+      "loss": 0.8452,
+      "step": 2420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649489870977648e-05,
+      "loss": 0.836,
+      "step": 2421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643368337916549e-05,
+      "loss": 0.8282,
+      "step": 2422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637246171874905e-05,
+      "loss": 0.7665,
+      "step": 2423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16311233752108e-05,
+      "loss": 0.8131,
+      "step": 2424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624999950282572e-05,
+      "loss": 0.841,
+      "step": 2425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1618875899448794e-05,
+      "loss": 0.8336,
+      "step": 2426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1612751225068283e-05,
+      "loss": 0.3309,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606625929500099e-05,
+      "loss": 0.7979,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1600500015103538e-05,
+      "loss": 0.3919,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159437348423813e-05,
+      "loss": 0.8635,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588246339263649e-05,
+      "loss": 0.7817,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582118582540107e-05,
+      "loss": 0.8176,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575990216427748e-05,
+      "loss": 0.7956,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569861243287047e-05,
+      "loss": 0.3634,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1563731665478722e-05,
+      "loss": 0.8756,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557601485363711e-05,
+      "loss": 0.8111,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15514707053032e-05,
+      "loss": 0.8201,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545339327658596e-05,
+      "loss": 0.7612,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1539207354791534e-05,
+      "loss": 0.8557,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533074789063888e-05,
+      "loss": 0.7742,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152694163283775e-05,
+      "loss": 0.8349,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1520807888475444e-05,
+      "loss": 0.3668,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514673558339524e-05,
+      "loss": 0.8031,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508538644792766e-05,
+      "loss": 0.8413,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502403150198164e-05,
+      "loss": 0.8177,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496267076918957e-05,
+      "loss": 0.8037,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490130427318578e-05,
+      "loss": 0.3507,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483993203760702e-05,
+      "loss": 0.8942,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477855408609225e-05,
+      "loss": 0.8966,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471717044228248e-05,
+      "loss": 0.84,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465578112982107e-05,
+      "loss": 0.8432,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459438617235353e-05,
+      "loss": 0.8351,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1453298559352746e-05,
+      "loss": 0.3431,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447157941699272e-05,
+      "loss": 0.8584,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441016766640127e-05,
+      "loss": 0.8218,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434875036540726e-05,
+      "loss": 0.748,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428732753766694e-05,
+      "loss": 0.8723,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422589920683871e-05,
+      "loss": 0.8483,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416446539658308e-05,
+      "loss": 0.7825,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410302613056266e-05,
+      "loss": 0.8183,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140415814324422e-05,
+      "loss": 0.7978,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1398013132588848e-05,
+      "loss": 0.7936,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391867583457048e-05,
+      "loss": 0.7999,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385721498215907e-05,
+      "loss": 0.8438,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379574879232736e-05,
+      "loss": 0.3327,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373427728875047e-05,
+      "loss": 0.8653,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367280049510544e-05,
+      "loss": 0.7659,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136113184350715e-05,
+      "loss": 0.8222,
+      "step": 2468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354983113232993e-05,
+      "loss": 0.8105,
+      "step": 2469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348833861056388e-05,
+      "loss": 0.82,
+      "step": 2470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342684089345859e-05,
+      "loss": 0.8164,
+      "step": 2471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133653380047013e-05,
+      "loss": 0.6813,
+      "step": 2472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330382996798128e-05,
+      "loss": 0.9097,
+      "step": 2473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324231680698972e-05,
+      "loss": 0.7641,
+      "step": 2474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131807985454198e-05,
+      "loss": 0.3493,
+      "step": 2475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311927520696667e-05,
+      "loss": 0.76,
+      "step": 2476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305774681532747e-05,
+      "loss": 0.8968,
+      "step": 2477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1299621339420119e-05,
+      "loss": 0.8406,
+      "step": 2478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293467496728886e-05,
+      "loss": 0.7934,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287313155829348e-05,
+      "loss": 0.8066,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281158319091971e-05,
+      "loss": 0.8054,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275002988887444e-05,
+      "loss": 0.8852,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268847167586629e-05,
+      "loss": 0.7844,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1262690857560577e-05,
+      "loss": 0.8219,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256534061180534e-05,
+      "loss": 0.7494,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250376780817926e-05,
+      "loss": 0.788,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.124421901884437e-05,
+      "loss": 0.7831,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123806077763167e-05,
+      "loss": 0.8223,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231902059551809e-05,
+      "loss": 0.9238,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225742866976958e-05,
+      "loss": 0.8192,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219583202279471e-05,
+      "loss": 0.8707,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121342306783188e-05,
+      "loss": 0.8473,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207262466006904e-05,
+      "loss": 0.8351,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201101399177434e-05,
+      "loss": 0.8034,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194939869716546e-05,
+      "loss": 0.8376,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188777879997492e-05,
+      "loss": 0.894,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182615432393707e-05,
+      "loss": 0.8233,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176452529278789e-05,
+      "loss": 0.8264,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170289173026527e-05,
+      "loss": 0.7682,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1164125366010872e-05,
+      "loss": 0.8565,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157961110605958e-05,
+      "loss": 0.8151,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151796409186084e-05,
+      "loss": 0.8399,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145631264125729e-05,
+      "loss": 0.8155,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139465677799535e-05,
+      "loss": 0.7874,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133299652582317e-05,
+      "loss": 0.8278,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127133190849062e-05,
+      "loss": 0.8436,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120966294974917e-05,
+      "loss": 0.8211,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114798967335209e-05,
+      "loss": 0.8531,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108631210305418e-05,
+      "loss": 0.8357,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11024630262612e-05,
+      "loss": 0.8295,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096294417578369e-05,
+      "loss": 0.7938,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090125386632899e-05,
+      "loss": 0.8203,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083955935800946e-05,
+      "loss": 0.7388,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077786067458796e-05,
+      "loss": 0.7971,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071615783982925e-05,
+      "loss": 0.7896,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065445087749958e-05,
+      "loss": 0.8132,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059273981136671e-05,
+      "loss": 0.8247,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105310246652001e-05,
+      "loss": 0.8363,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046930546277077e-05,
+      "loss": 0.8051,
+      "step": 2519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040758222785121e-05,
+      "loss": 0.7844,
+      "step": 2520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034585498421557e-05,
+      "loss": 0.8311,
+      "step": 2521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028412375563947e-05,
+      "loss": 0.7825,
+      "step": 2522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022238856590006e-05,
+      "loss": 0.7942,
+      "step": 2523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1016064943877612e-05,
+      "loss": 0.7877,
+      "step": 2524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009890639804782e-05,
+      "loss": 0.825,
+      "step": 2525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100371594674969e-05,
+      "loss": 0.8247,
+      "step": 2526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099754086709066e-05,
+      "loss": 0.7991,
+      "step": 2527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0991365403206162e-05,
+      "loss": 0.8151,
+      "step": 2528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098518955747482e-05,
+      "loss": 0.8158,
+      "step": 2529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979013332275399e-05,
+      "loss": 0.8144,
+      "step": 2530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972836729986807e-05,
+      "loss": 0.7769,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096665975298811e-05,
+      "loss": 0.8182,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0960482403658505e-05,
+      "loss": 0.8055,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954304684377341e-05,
+      "loss": 0.8131,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0948126597524108e-05,
+      "loss": 0.8043,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0941948145478431e-05,
+      "loss": 0.7736,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935769330620085e-05,
+      "loss": 0.8287,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092959015532898e-05,
+      "loss": 0.8152,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923410621985165e-05,
+      "loss": 0.8772,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917230732968824e-05,
+      "loss": 0.7958,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0911050490660289e-05,
+      "loss": 0.3382,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0904869897440012e-05,
+      "loss": 0.8297,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089868895568859e-05,
+      "loss": 0.7826,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892507667786761e-05,
+      "loss": 0.8193,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886326036115376e-05,
+      "loss": 0.8349,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880144063055438e-05,
+      "loss": 0.3446,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873961750988074e-05,
+      "loss": 0.8225,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867779102294538e-05,
+      "loss": 0.862,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0861596119356221e-05,
+      "loss": 0.7573,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855412804554637e-05,
+      "loss": 0.8216,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0849229160271431e-05,
+      "loss": 0.797,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084304518888837e-05,
+      "loss": 0.8501,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836860892787355e-05,
+      "loss": 0.8338,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830676274350403e-05,
+      "loss": 0.8219,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824491335959669e-05,
+      "loss": 0.3742,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818306079997411e-05,
+      "loss": 0.7908,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812120508846026e-05,
+      "loss": 0.8231,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080593462488803e-05,
+      "loss": 0.8346,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079974843050605e-05,
+      "loss": 0.8267,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793561928082843e-05,
+      "loss": 0.8458,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787375120001281e-05,
+      "loss": 0.8329,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0781188008644347e-05,
+      "loss": 0.7182,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775000596395157e-05,
+      "loss": 0.843,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768812885636925e-05,
+      "loss": 0.7711,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762624878752993e-05,
+      "loss": 0.8688,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756436578126808e-05,
+      "loss": 0.8245,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750247986141935e-05,
+      "loss": 0.7464,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744059105182055e-05,
+      "loss": 0.7777,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737869937630949e-05,
+      "loss": 0.7898,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073168048587252e-05,
+      "loss": 0.8307,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725490752290772e-05,
+      "loss": 0.7384,
+      "step": 2571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719300739269827e-05,
+      "loss": 0.8619,
+      "step": 2572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07131104491939e-05,
+      "loss": 0.8317,
+      "step": 2573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0706919884447329e-05,
+      "loss": 0.8169,
+      "step": 2574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700729047414551e-05,
+      "loss": 0.7859,
+      "step": 2575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0694537940480099e-05,
+      "loss": 0.8461,
+      "step": 2576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688346566028622e-05,
+      "loss": 0.6659,
+      "step": 2577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682154926444873e-05,
+      "loss": 0.8658,
+      "step": 2578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0675963024113696e-05,
+      "loss": 0.8324,
+      "step": 2579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669770861420046e-05,
+      "loss": 0.8045,
+      "step": 2580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663578440748972e-05,
+      "loss": 0.8194,
+      "step": 2581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657385764485625e-05,
+      "loss": 0.7644,
+      "step": 2582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651192835015261e-05,
+      "loss": 0.7313,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0644999654723218e-05,
+      "loss": 0.8652,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638806225994945e-05,
+      "loss": 0.3511,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632612551215984e-05,
+      "loss": 0.8016,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626418632771962e-05,
+      "loss": 0.3393,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062022447304861e-05,
+      "loss": 0.8162,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614030074431754e-05,
+      "loss": 0.7698,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607835439307297e-05,
+      "loss": 0.8383,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060164057006125e-05,
+      "loss": 0.8387,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595445469079707e-05,
+      "loss": 0.7946,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589250138748852e-05,
+      "loss": 0.7862,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583054581454953e-05,
+      "loss": 0.7965,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0576858799584373e-05,
+      "loss": 0.7752,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570662795523554e-05,
+      "loss": 0.3865,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0564466571659032e-05,
+      "loss": 0.8107,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055827013037742e-05,
+      "loss": 0.8854,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055207347406542e-05,
+      "loss": 0.8812,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545876605109808e-05,
+      "loss": 0.8475,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0539679525897457e-05,
+      "loss": 0.7859,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533482238815308e-05,
+      "loss": 0.7996,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052728474625039e-05,
+      "loss": 0.8101,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05210870505898e-05,
+      "loss": 0.3427,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514889154220728e-05,
+      "loss": 0.8479,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508691059530436e-05,
+      "loss": 0.853,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0502492768906251e-05,
+      "loss": 0.863,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0496294284735595e-05,
+      "loss": 0.8089,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490095609405951e-05,
+      "loss": 0.8527,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0483896745304875e-05,
+      "loss": 0.8063,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0477697694820008e-05,
+      "loss": 0.8025,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047149846033905e-05,
+      "loss": 0.8554,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465299044249776e-05,
+      "loss": 0.87,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045909944894003e-05,
+      "loss": 0.7876,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452899676797733e-05,
+      "loss": 0.789,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446699730210858e-05,
+      "loss": 0.8052,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044049961156747e-05,
+      "loss": 0.8671,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043429932325567e-05,
+      "loss": 0.8683,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428098867663645e-05,
+      "loss": 0.8473,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421898247179647e-05,
+      "loss": 0.8167,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0415697464191974e-05,
+      "loss": 0.8727,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0409496521089006e-05,
+      "loss": 0.861,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403295420259177e-05,
+      "loss": 0.8115,
+      "step": 2623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397094164090978e-05,
+      "loss": 0.7764,
+      "step": 2624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390892754972964e-05,
+      "loss": 0.8169,
+      "step": 2625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038469119529375e-05,
+      "loss": 0.8166,
+      "step": 2626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378489487442005e-05,
+      "loss": 0.8271,
+      "step": 2627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372287633806459e-05,
+      "loss": 0.7671,
+      "step": 2628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366085636775896e-05,
+      "loss": 0.8738,
+      "step": 2629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359883498739151e-05,
+      "loss": 0.8382,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035368122208513e-05,
+      "loss": 0.8091,
+      "step": 2631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0347478809202769e-05,
+      "loss": 0.8049,
+      "step": 2632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034127626248107e-05,
+      "loss": 0.7877,
+      "step": 2633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033507358430909e-05,
+      "loss": 0.8132,
+      "step": 2634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328870777075923e-05,
+      "loss": 0.8537,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0322667843170728e-05,
+      "loss": 0.8035,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0316464784982703e-05,
+      "loss": 0.362,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310261604901093e-05,
+      "loss": 0.8428,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304058305315196e-05,
+      "loss": 0.8112,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297854888614356e-05,
+      "loss": 0.8605,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291651357187957e-05,
+      "loss": 0.7734,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028544771342543e-05,
+      "loss": 0.7781,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279243959716245e-05,
+      "loss": 0.8047,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273040098449925e-05,
+      "loss": 0.755,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266836132016024e-05,
+      "loss": 0.8126,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260632062804143e-05,
+      "loss": 0.884,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0254427893203914e-05,
+      "loss": 0.8067,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024822362560502e-05,
+      "loss": 0.7866,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242019262397171e-05,
+      "loss": 0.8137,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023581480597012e-05,
+      "loss": 0.827,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229610258713656e-05,
+      "loss": 0.7999,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0223405623017595e-05,
+      "loss": 0.8391,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0217200901271798e-05,
+      "loss": 0.3667,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210996095866152e-05,
+      "loss": 0.8584,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204791209190581e-05,
+      "loss": 0.8243,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0198586243635034e-05,
+      "loss": 0.7865,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0192381201589497e-05,
+      "loss": 0.8135,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018617608544398e-05,
+      "loss": 0.8399,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179970897588528e-05,
+      "loss": 0.8498,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0173765640413202e-05,
+      "loss": 0.8084,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0167560316308104e-05,
+      "loss": 0.836,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161354927663358e-05,
+      "loss": 0.8049,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155149476869103e-05,
+      "loss": 0.3898,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014894396631551e-05,
+      "loss": 0.8078,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014273839839278e-05,
+      "loss": 0.8243,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0136532775491114e-05,
+      "loss": 0.7924,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130327100000758e-05,
+      "loss": 0.7829,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012412137431197e-05,
+      "loss": 0.8003,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0117915600815021e-05,
+      "loss": 0.856,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111709781900204e-05,
+      "loss": 0.8079,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105503919957837e-05,
+      "loss": 0.8363,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099298017378242e-05,
+      "loss": 0.7447,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009309207655177e-05,
+      "loss": 0.8433,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008688609986877e-05,
+      "loss": 0.8298,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080680089719624e-05,
+      "loss": 0.805,
+      "step": 2675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074474048494714e-05,
+      "loss": 0.8283,
+      "step": 2676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068267978584434e-05,
+      "loss": 0.7682,
+      "step": 2677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062061882379195e-05,
+      "loss": 0.7564,
+      "step": 2678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055855762269418e-05,
+      "loss": 0.8291,
+      "step": 2679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049649620645525e-05,
+      "loss": 0.3533,
+      "step": 2680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043443459897959e-05,
+      "loss": 0.8583,
+      "step": 2681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037237282417163e-05,
+      "loss": 0.8139,
+      "step": 2682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031031090593576e-05,
+      "loss": 0.8129,
+      "step": 2683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002482488681767e-05,
+      "loss": 0.7769,
+      "step": 2684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001861867347989e-05,
+      "loss": 0.8577,
+      "step": 2685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012412452970708e-05,
+      "loss": 0.7853,
+      "step": 2686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000620622768059e-05,
+      "loss": 0.8346,
+      "step": 2687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 0.8332,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993793772319412e-06,
+      "loss": 0.7631,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987587547029295e-06,
+      "loss": 0.8128,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981381326520112e-06,
+      "loss": 0.7911,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975175113182332e-06,
+      "loss": 0.8133,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968968909406422e-06,
+      "loss": 0.8562,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962762717582842e-06,
+      "loss": 0.8246,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956556540102043e-06,
+      "loss": 0.7829,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950350379354475e-06,
+      "loss": 0.8233,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944144237730587e-06,
+      "loss": 0.3571,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937938117620809e-06,
+      "loss": 0.8301,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931732021415569e-06,
+      "loss": 0.8467,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925525951505288e-06,
+      "loss": 0.7864,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919319910280381e-06,
+      "loss": 0.8342,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91311390013123e-06,
+      "loss": 0.7781,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906907923448232e-06,
+      "loss": 0.841,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900701982621761e-06,
+      "loss": 0.8287,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894496080042165e-06,
+      "loss": 0.8756,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888290218099797e-06,
+      "loss": 0.8011,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88208439918498e-06,
+      "loss": 0.8577,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875878625688033e-06,
+      "loss": 0.7721,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869672899999244e-06,
+      "loss": 0.7995,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863467224508887e-06,
+      "loss": 0.7744,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857261601607226e-06,
+      "loss": 0.3831,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851056033684492e-06,
+      "loss": 0.7747,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844850523130899e-06,
+      "loss": 0.7892,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838645072336647e-06,
+      "loss": 0.792,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.832439683691897e-06,
+      "loss": 0.764,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8262343595868e-06,
+      "loss": 0.8292,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820029102411477e-06,
+      "loss": 0.8422,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813823914556026e-06,
+      "loss": 0.8841,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807618798410507e-06,
+      "loss": 0.8278,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80141375636497e-06,
+      "loss": 0.838,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.795208790809425e-06,
+      "loss": 0.7718,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789003904133851e-06,
+      "loss": 0.8433,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.782799098728205e-06,
+      "loss": 0.7736,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776594376982405e-06,
+      "loss": 0.8331,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77038974128635e-06,
+      "loss": 0.3346,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764185194029882e-06,
+      "loss": 0.7977,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75798073760283e-06,
+      "loss": 0.8308,
+      "step": 2727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751776374394985e-06,
+      "loss": 0.8353,
+      "step": 2728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74557210679609e-06,
+      "loss": 0.7636,
+      "step": 2729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73936793719586e-06,
+      "loss": 0.8644,
+      "step": 2730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733163867983978e-06,
+      "loss": 0.8251,
+      "step": 2731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72695990155008e-06,
+      "loss": 0.8578,
+      "step": 2732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720756040283757e-06,
+      "loss": 0.8559,
+      "step": 2733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714552286574574e-06,
+      "loss": 0.837,
+      "step": 2734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70834864281205e-06,
+      "loss": 0.7442,
+      "step": 2735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702145111385647e-06,
+      "loss": 0.8665,
+      "step": 2736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695941694684805e-06,
+      "loss": 0.7741,
+      "step": 2737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689738395098908e-06,
+      "loss": 0.8159,
+      "step": 2738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683535215017302e-06,
+      "loss": 0.7598,
+      "step": 2739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677332156829276e-06,
+      "loss": 0.8639,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671129222924076e-06,
+      "loss": 0.3704,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664926415690915e-06,
+      "loss": 0.7551,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658723737518932e-06,
+      "loss": 0.7853,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652521190797235e-06,
+      "loss": 0.8456,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646318777914874e-06,
+      "loss": 0.8046,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64011650126085e-06,
+      "loss": 0.8567,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633914363224107e-06,
+      "loss": 0.7372,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627712366193545e-06,
+      "loss": 0.8314,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621510512558e-06,
+      "loss": 0.7938,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615308804706253e-06,
+      "loss": 0.8097,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609107245027038e-06,
+      "loss": 0.8083,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602905835909027e-06,
+      "loss": 0.7728,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596704579740827e-06,
+      "loss": 0.853,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.590503478910996e-06,
+      "loss": 0.838,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584302535808027e-06,
+      "loss": 0.8106,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57810175282036e-06,
+      "loss": 0.7783,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571901132336357e-06,
+      "loss": 0.7948,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565700676744331e-06,
+      "loss": 0.873,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559500388432537e-06,
+      "loss": 0.7624,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553300269789144e-06,
+      "loss": 0.7646,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54710032320227e-06,
+      "loss": 0.7795,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540900551059973e-06,
+      "loss": 0.7365,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534700955750231e-06,
+      "loss": 0.796,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528501539660955e-06,
+      "loss": 0.7481,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522302305179995e-06,
+      "loss": 0.8124,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.516103254695128e-06,
+      "loss": 0.782,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509904390594052e-06,
+      "loss": 0.3806,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503705715264408e-06,
+      "loss": 0.7817,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497507231093749e-06,
+      "loss": 0.8613,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491308940469569e-06,
+      "loss": 0.8391,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485110845779273e-06,
+      "loss": 0.8458,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4789129494102e-06,
+      "loss": 0.8602,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472715253749617e-06,
+      "loss": 0.8646,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466517761184696e-06,
+      "loss": 0.8356,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460320474102546e-06,
+      "loss": 0.7916,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454123394890193e-06,
+      "loss": 0.769,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447926525934586e-06,
+      "loss": 0.3787,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441729869622583e-06,
+      "loss": 0.8041,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43553342834097e-06,
+      "loss": 0.788,
+      "step": 2779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42933720447645e-06,
+      "loss": 0.775,
+      "step": 2780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42314120041563e-06,
+      "loss": 0.8078,
+      "step": 2781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.416945418545049e-06,
+      "loss": 0.8274,
+      "step": 2782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41074986125115e-06,
+      "loss": 0.7744,
+      "step": 2783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404554530920294e-06,
+      "loss": 0.7724,
+      "step": 2784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398359429938753e-06,
+      "loss": 0.8053,
+      "step": 2785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392164560692705e-06,
+      "loss": 0.8792,
+      "step": 2786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385969925568253e-06,
+      "loss": 0.3499,
+      "step": 2787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379775526951393e-06,
+      "loss": 0.872,
+      "step": 2788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37358136722804e-06,
+      "loss": 0.7309,
+      "step": 2789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36738744878402e-06,
+      "loss": 0.8507,
+      "step": 2790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361193774005057e-06,
+      "loss": 0.7959,
+      "step": 2791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.355000345276783e-06,
+      "loss": 0.7515,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348807164984742e-06,
+      "loss": 0.8259,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342614235514378e-06,
+      "loss": 0.8491,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336421559251032e-06,
+      "loss": 0.8114,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330229138579959e-06,
+      "loss": 0.8217,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.324036975886309e-06,
+      "loss": 0.8396,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31784507355513e-06,
+      "loss": 0.8143,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31165343397138e-06,
+      "loss": 0.8296,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305462059519903e-06,
+      "loss": 0.7893,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299270952585454e-06,
+      "loss": 0.3339,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293080115552673e-06,
+      "loss": 0.8153,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2868895508061e-06,
+      "loss": 0.8088,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280699260730178e-06,
+      "loss": 0.7313,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274509247709231e-06,
+      "loss": 0.8066,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268319514127482e-06,
+      "loss": 0.8682,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.262130062369053e-06,
+      "loss": 0.3638,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25594089481795e-06,
+      "loss": 0.877,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249752013858067e-06,
+      "loss": 0.7445,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243563421873195e-06,
+      "loss": 0.7963,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237375121247012e-06,
+      "loss": 0.8666,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231187114363077e-06,
+      "loss": 0.784,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224999403604846e-06,
+      "loss": 0.7927,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218811991355651e-06,
+      "loss": 0.8467,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212624879998724e-06,
+      "loss": 0.3573,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206438071917159e-06,
+      "loss": 0.7581,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20025156949395e-06,
+      "loss": 0.7966,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194065375111973e-06,
+      "loss": 0.8369,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187879491153975e-06,
+      "loss": 0.8044,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18169392000259e-06,
+      "loss": 0.8124,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175508664040335e-06,
+      "loss": 0.8463,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169323725649599e-06,
+      "loss": 0.9189,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163139107212649e-06,
+      "loss": 0.7819,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156954811111633e-06,
+      "loss": 0.8897,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150770839728576e-06,
+      "loss": 0.8648,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144587195445366e-06,
+      "loss": 0.8613,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13840388064378e-06,
+      "loss": 0.8597,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132220897705462e-06,
+      "loss": 0.7855,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12603824901193e-06,
+      "loss": 0.7748,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119855936944565e-06,
+      "loss": 0.7543,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113673963884625e-06,
+      "loss": 0.8386,
+      "step": 2831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107492332213246e-06,
+      "loss": 0.783,
+      "step": 2832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101311044311413e-06,
+      "loss": 0.8269,
+      "step": 2833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.095130102559992e-06,
+      "loss": 0.8345,
+      "step": 2834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.088949509339718e-06,
+      "loss": 0.8077,
+      "step": 2835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08276926703118e-06,
+      "loss": 0.8191,
+      "step": 2836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076589378014838e-06,
+      "loss": 0.3716,
+      "step": 2837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070409844671021e-06,
+      "loss": 0.861,
+      "step": 2838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064230669379919e-06,
+      "loss": 0.7827,
+      "step": 2839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05805185452157e-06,
+      "loss": 0.7873,
+      "step": 2840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051873402475894e-06,
+      "loss": 0.8045,
+      "step": 2841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.045695315622662e-06,
+      "loss": 0.7958,
+      "step": 2842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039517596341497e-06,
+      "loss": 0.751,
+      "step": 2843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033340247011894e-06,
+      "loss": 0.7982,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027163270013193e-06,
+      "loss": 0.7897,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020986667724606e-06,
+      "loss": 0.7661,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014810442525183e-06,
+      "loss": 0.7513,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008634596793838e-06,
+      "loss": 0.8355,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002459132909344e-06,
+      "loss": 0.8299,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996284053250315e-06,
+      "loss": 0.7845,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99010936019522e-06,
+      "loss": 0.8136,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.983935056122392e-06,
+      "loss": 0.822,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977761143409997e-06,
+      "loss": 0.8406,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971587624436058e-06,
+      "loss": 0.8077,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965414501578446e-06,
+      "loss": 0.8113,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959241777214884e-06,
+      "loss": 0.8022,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953069453722926e-06,
+      "loss": 0.7466,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94689753347999e-06,
+      "loss": 0.8502,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940726018863329e-06,
+      "loss": 0.8026,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934554912250046e-06,
+      "loss": 0.7996,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928384216017078e-06,
+      "loss": 0.7528,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922213932541204e-06,
+      "loss": 0.843,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91604406419906e-06,
+      "loss": 0.8165,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909874613367103e-06,
+      "loss": 0.7807,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903705582421633e-06,
+      "loss": 0.8533,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897536973738803e-06,
+      "loss": 0.8438,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891368789694585e-06,
+      "loss": 0.3777,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885201032664793e-06,
+      "loss": 0.8081,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879033705025085e-06,
+      "loss": 0.7705,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872866809150943e-06,
+      "loss": 0.7948,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866700347417684e-06,
+      "loss": 0.8397,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860534322200468e-06,
+      "loss": 0.8328,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854368735874276e-06,
+      "loss": 0.849,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848203590813918e-06,
+      "loss": 0.8446,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842038889394046e-06,
+      "loss": 0.7223,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.835874633989128e-06,
+      "loss": 0.8094,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829710826973476e-06,
+      "loss": 0.8443,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.823547470721215e-06,
+      "loss": 0.7795,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817384567606294e-06,
+      "loss": 0.6924,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81122212000251e-06,
+      "loss": 0.8135,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805060130283458e-06,
+      "loss": 0.8368,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798898600822568e-06,
+      "loss": 0.7924,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792737533993098e-06,
+      "loss": 0.7584,
+      "step": 2883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786576932168123e-06,
+      "loss": 0.7694,
+      "step": 2884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78041679772053e-06,
+      "loss": 0.8595,
+      "step": 2885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774257133023044e-06,
+      "loss": 0.7819,
+      "step": 2886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768097940448194e-06,
+      "loss": 0.8129,
+      "step": 2887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761939222368332e-06,
+      "loss": 0.8586,
+      "step": 2888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755780981155631e-06,
+      "loss": 0.7829,
+      "step": 2889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749623219182075e-06,
+      "loss": 0.8205,
+      "step": 2890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74346593881947e-06,
+      "loss": 0.8807,
+      "step": 2891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737309142439425e-06,
+      "loss": 0.7746,
+      "step": 2892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731152832413373e-06,
+      "loss": 0.3411,
+      "step": 2893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724997011112558e-06,
+      "loss": 0.8107,
+      "step": 2894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718841680908032e-06,
+      "loss": 0.7909,
+      "step": 2895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712686844170657e-06,
+      "loss": 0.8448,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706532503271113e-06,
+      "loss": 0.8454,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.700378660579886e-06,
+      "loss": 0.8593,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694225318467258e-06,
+      "loss": 0.8271,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688072479303335e-06,
+      "loss": 0.8088,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681920145458024e-06,
+      "loss": 0.8112,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675768319301031e-06,
+      "loss": 0.8096,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669617003201875e-06,
+      "loss": 0.7539,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663466199529869e-06,
+      "loss": 0.7904,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657315910654144e-06,
+      "loss": 0.8455,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651166138943617e-06,
+      "loss": 0.8314,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645016886767009e-06,
+      "loss": 0.7906,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638868156492851e-06,
+      "loss": 0.8229,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63271995048946e-06,
+      "loss": 0.8162,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626572271124956e-06,
+      "loss": 0.8149,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620425120767264e-06,
+      "loss": 0.8068,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614278501784096e-06,
+      "loss": 0.8973,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608132416542955e-06,
+      "loss": 0.7945,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601986867411152e-06,
+      "loss": 0.3646,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595841856755785e-06,
+      "loss": 0.3311,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589697386943736e-06,
+      "loss": 0.8229,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583553460341695e-06,
+      "loss": 0.802,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577410079316134e-06,
+      "loss": 0.7842,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571267246233308e-06,
+      "loss": 0.8539,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565124963459277e-06,
+      "loss": 0.7678,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558983233359873e-06,
+      "loss": 0.7919,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552842058300731e-06,
+      "loss": 0.7618,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546701440647258e-06,
+      "loss": 0.8553,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540561382764649e-06,
+      "loss": 0.8241,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534421887017895e-06,
+      "loss": 0.8162,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528282955771756e-06,
+      "loss": 0.7739,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522144591390778e-06,
+      "loss": 0.769,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5160067962393e-06,
+      "loss": 0.8175,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509869572681427e-06,
+      "loss": 0.8278,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503732923081048e-06,
+      "loss": 0.7595,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497596849801836e-06,
+      "loss": 0.7592,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49146135520724e-06,
+      "loss": 0.8184,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48532644166048e-06,
+      "loss": 0.7849,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479192111524559e-06,
+      "loss": 0.7971,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.473058367162252e-06,
+      "loss": 0.7603,
+      "step": 2935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.466925210936117e-06,
+      "loss": 0.8037,
+      "step": 2936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460792645208469e-06,
+      "loss": 0.7848,
+      "step": 2937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454660672341405e-06,
+      "loss": 0.8087,
+      "step": 2938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448529294696803e-06,
+      "loss": 0.8396,
+      "step": 2939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442398514636292e-06,
+      "loss": 0.792,
+      "step": 2940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436268334521283e-06,
+      "loss": 0.8313,
+      "step": 2941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430138756712955e-06,
+      "loss": 0.8468,
+      "step": 2942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424009783572259e-06,
+      "loss": 0.7801,
+      "step": 2943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417881417459894e-06,
+      "loss": 0.8102,
+      "step": 2944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411753660736353e-06,
+      "loss": 0.7885,
+      "step": 2945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405626515761876e-06,
+      "loss": 0.8525,
+      "step": 2946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399499984896467e-06,
+      "loss": 0.8451,
+      "step": 2947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393374070499903e-06,
+      "loss": 0.3505,
+      "step": 2948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.387248774931717e-06,
+      "loss": 0.771,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381124100551209e-06,
+      "loss": 0.7999,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375000049717431e-06,
+      "loss": 0.7481,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3688766247892e-06,
+      "loss": 0.8277,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362753828125099e-06,
+      "loss": 0.8504,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356631662083453e-06,
+      "loss": 0.7816,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350510129022355e-06,
+      "loss": 0.7626,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344389231299659e-06,
+      "loss": 0.8261,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338268971272959e-06,
+      "loss": 0.8476,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332149351299614e-06,
+      "loss": 0.9058,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326030373736737e-06,
+      "loss": 0.7839,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319912040941191e-06,
+      "loss": 0.79,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313794355269586e-06,
+      "loss": 0.797,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30767731907829e-06,
+      "loss": 0.897,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301560934723421e-06,
+      "loss": 0.8271,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295445204560837e-06,
+      "loss": 0.3824,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289330130946153e-06,
+      "loss": 0.815,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283215716234728e-06,
+      "loss": 0.7939,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27710196278167e-06,
+      "loss": 0.836,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270988872941828e-06,
+      "loss": 0.4024,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264876449069791e-06,
+      "loss": 0.7979,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258764693519909e-06,
+      "loss": 0.7375,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252653608646256e-06,
+      "loss": 0.775,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246543196802653e-06,
+      "loss": 0.8544,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240433460342668e-06,
+      "loss": 0.842,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234324401619604e-06,
+      "loss": 0.7812,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.228216022986498e-06,
+      "loss": 0.8456,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222108326796138e-06,
+      "loss": 0.387,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21600131540104e-06,
+      "loss": 0.8667,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209894991153451e-06,
+      "loss": 0.8257,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.203789356405368e-06,
+      "loss": 0.789,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197684413508509e-06,
+      "loss": 0.3668,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19158016481434e-06,
+      "loss": 0.8583,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.185476612674041e-06,
+      "loss": 0.8105,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.179373759438534e-06,
+      "loss": 0.8484,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173271607458478e-06,
+      "loss": 0.8499,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16717015908425e-06,
+      "loss": 0.8196,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161069416665962e-06,
+      "loss": 0.8147,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154969382553453e-06,
+      "loss": 0.7884,
+      "step": 2987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148870059096291e-06,
+      "loss": 0.8243,
+      "step": 2988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142771448643765e-06,
+      "loss": 0.7949,
+      "step": 2989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136673553544897e-06,
+      "loss": 0.8831,
+      "step": 2990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13057637614843e-06,
+      "loss": 0.8253,
+      "step": 2991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.124479918802824e-06,
+      "loss": 0.8097,
+      "step": 2992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118384183856273e-06,
+      "loss": 0.8162,
+      "step": 2993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112289173656684e-06,
+      "loss": 0.787,
+      "step": 2994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106194890551695e-06,
+      "loss": 0.7872,
+      "step": 2995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10010133688865e-06,
+      "loss": 0.8245,
+      "step": 2996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09400851501462e-06,
+      "loss": 0.7921,
+      "step": 2997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.087916427276398e-06,
+      "loss": 0.739,
+      "step": 2998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081825076020487e-06,
+      "loss": 0.8066,
+      "step": 2999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075734463593105e-06,
+      "loss": 0.7558,
+      "step": 3000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069644592340196e-06,
+      "loss": 0.8682,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06355546460741e-06,
+      "loss": 0.8116,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057467082740106e-06,
+      "loss": 0.8574,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051379449083372e-06,
+      "loss": 0.8015,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045292565981995e-06,
+      "loss": 0.8125,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039206435780471e-06,
+      "loss": 0.8156,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033121060823015e-06,
+      "loss": 0.8853,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02703644345355e-06,
+      "loss": 0.7949,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020952586015698e-06,
+      "loss": 0.8147,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014869490852798e-06,
+      "loss": 0.8474,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00878716030789e-06,
+      "loss": 0.8823,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002705596723726e-06,
+      "loss": 0.7824,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996624802442754e-06,
+      "loss": 0.7952,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990544779807127e-06,
+      "loss": 0.8477,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984465531158712e-06,
+      "loss": 0.7849,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978387058839062e-06,
+      "loss": 0.8139,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972309365189437e-06,
+      "loss": 0.8317,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966232452550806e-06,
+      "loss": 0.3454,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960156323263828e-06,
+      "loss": 0.838,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954080979668853e-06,
+      "loss": 0.8214,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948006424105948e-06,
+      "loss": 0.8674,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941932658914863e-06,
+      "loss": 0.847,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93585968643504e-06,
+      "loss": 0.8581,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929787509005628e-06,
+      "loss": 0.803,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923716128965463e-06,
+      "loss": 0.7523,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917645548653076e-06,
+      "loss": 0.8156,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911575770406685e-06,
+      "loss": 0.804,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905506796564202e-06,
+      "loss": 0.8202,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899438629463236e-06,
+      "loss": 0.7882,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893371271441073e-06,
+      "loss": 0.7822,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887304724834693e-06,
+      "loss": 0.7934,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881238991980768e-06,
+      "loss": 0.7269,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875174075215654e-06,
+      "loss": 0.7806,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869109976875384e-06,
+      "loss": 0.8198,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863046699295687e-06,
+      "loss": 0.8315,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856984244811974e-06,
+      "loss": 0.8278,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85092261575933e-06,
+      "loss": 0.797,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844861814472535e-06,
+      "loss": 0.869,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.838801843286044e-06,
+      "loss": 0.8073,
+      "step": 3039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83274270453399e-06,
+      "loss": 0.7481,
+      "step": 3040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826684400550182e-06,
+      "loss": 0.7812,
+      "step": 3041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82062693366812e-06,
+      "loss": 0.7916,
+      "step": 3042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81457030622098e-06,
+      "loss": 0.7638,
+      "step": 3043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.808514520541595e-06,
+      "loss": 0.3244,
+      "step": 3044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802459578962496e-06,
+      "loss": 0.8612,
+      "step": 3045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79640548381589e-06,
+      "loss": 0.8094,
+      "step": 3046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79035223743363e-06,
+      "loss": 0.7754,
+      "step": 3047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78429984214727e-06,
+      "loss": 0.7576,
+      "step": 3048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77824830028803e-06,
+      "loss": 0.7485,
+      "step": 3049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772197614186799e-06,
+      "loss": 0.8376,
+      "step": 3050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766147786174132e-06,
+      "loss": 0.7974,
+      "step": 3051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760098818580253e-06,
+      "loss": 0.8187,
+      "step": 3052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.754050713735072e-06,
+      "loss": 0.8842,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748003473968142e-06,
+      "loss": 0.8223,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741957101608696e-06,
+      "loss": 0.8164,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.735911598985632e-06,
+      "loss": 0.8062,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729866968427523e-06,
+      "loss": 0.9117,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723823212262577e-06,
+      "loss": 0.8266,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717780332818696e-06,
+      "loss": 0.8607,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711738332423433e-06,
+      "loss": 0.7812,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705697213403991e-06,
+      "loss": 0.7825,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699656978087253e-06,
+      "loss": 0.8434,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.693617628799749e-06,
+      "loss": 0.7975,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687579167867673e-06,
+      "loss": 0.7404,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681541597616872e-06,
+      "loss": 0.7724,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675504920372853e-06,
+      "loss": 0.8277,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.669469138460785e-06,
+      "loss": 0.7553,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66343425420548e-06,
+      "loss": 0.7585,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657400269931412e-06,
+      "loss": 0.8345,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651367187962704e-06,
+      "loss": 0.7866,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645335010623145e-06,
+      "loss": 0.8518,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63930374023615e-06,
+      "loss": 0.7865,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63327337912481e-06,
+      "loss": 0.843,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627243929611854e-06,
+      "loss": 0.8153,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621215394019653e-06,
+      "loss": 0.8908,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615187774670243e-06,
+      "loss": 0.8334,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6091610738852965e-06,
+      "loss": 0.7893,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603135293986128e-06,
+      "loss": 0.81,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597110437293711e-06,
+      "loss": 0.8455,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591086506128649e-06,
+      "loss": 0.7821,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585063502811203e-06,
+      "loss": 0.7856,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5790414296612626e-06,
+      "loss": 0.7612,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5730202889983665e-06,
+      "loss": 0.8119,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5670000831416986e-06,
+      "loss": 0.7696,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5609808144100726e-06,
+      "loss": 0.7844,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.554962485121944e-06,
+      "loss": 0.3615,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548945097595416e-06,
+      "loss": 0.804,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.542928654148219e-06,
+      "loss": 0.379,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536913157097718e-06,
+      "loss": 0.8179,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530898608760924e-06,
+      "loss": 0.8521,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524885011454474e-06,
+      "loss": 0.7836,
+      "step": 3091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51887236749464e-06,
+      "loss": 0.8184,
+      "step": 3092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.512860679197329e-06,
+      "loss": 0.8398,
+      "step": 3093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506849948878079e-06,
+      "loss": 0.7677,
+      "step": 3094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500840178852063e-06,
+      "loss": 0.7121,
+      "step": 3095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494831371434074e-06,
+      "loss": 0.8004,
+      "step": 3096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4888235289385405e-06,
+      "loss": 0.7716,
+      "step": 3097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.482816653679525e-06,
+      "loss": 0.8019,
+      "step": 3098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476810747970707e-06,
+      "loss": 0.859,
+      "step": 3099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470805814125396e-06,
+      "loss": 0.8116,
+      "step": 3100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4648018544565315e-06,
+      "loss": 0.7363,
+      "step": 3101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458798871276677e-06,
+      "loss": 0.7948,
+      "step": 3102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4527968668980085e-06,
+      "loss": 0.7199,
+      "step": 3103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446795843632343e-06,
+      "loss": 0.7827,
+      "step": 3104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.440795803791108e-06,
+      "loss": 0.8116,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43479674968535e-06,
+      "loss": 0.8342,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428798683625747e-06,
+      "loss": 0.7702,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422801607922585e-06,
+      "loss": 0.7422,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416805524885782e-06,
+      "loss": 0.7829,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410810436824855e-06,
+      "loss": 0.81,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4048163460489505e-06,
+      "loss": 0.8135,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.398823254866836e-06,
+      "loss": 0.7976,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392831165586877e-06,
+      "loss": 0.7111,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3868400805170635e-06,
+      "loss": 0.7971,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3808500019650044e-06,
+      "loss": 0.8556,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.374860932237911e-06,
+      "loss": 0.732,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368872873642603e-06,
+      "loss": 0.7376,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362885828485525e-06,
+      "loss": 0.8148,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356899799072723e-06,
+      "loss": 0.8481,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350914787709844e-06,
+      "loss": 0.8153,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3449307967021565e-06,
+      "loss": 0.8177,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.338947828354534e-06,
+      "loss": 0.8374,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332965884971441e-06,
+      "loss": 0.7947,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3269849688569695e-06,
+      "loss": 0.8804,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321005082314798e-06,
+      "loss": 0.7207,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3150262276482206e-06,
+      "loss": 0.3421,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309048407160126e-06,
+      "loss": 0.7947,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303071623153004e-06,
+      "loss": 0.7181,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297095877928955e-06,
+      "loss": 0.7758,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2911211737896685e-06,
+      "loss": 0.7481,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285147513036435e-06,
+      "loss": 0.7864,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27917489797015e-06,
+      "loss": 0.8034,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.273203330891303e-06,
+      "loss": 0.808,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26723281409997e-06,
+      "loss": 0.8437,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2612633498958375e-06,
+      "loss": 0.8042,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25529494057818e-06,
+      "loss": 0.8582,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24932758844586e-06,
+      "loss": 0.8476,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243361295797344e-06,
+      "loss": 0.813,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237396064930682e-06,
+      "loss": 0.8259,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231431898143523e-06,
+      "loss": 0.801,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225468797733094e-06,
+      "loss": 0.8222,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219506765996218e-06,
+      "loss": 0.7952,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213545805229316e-06,
+      "loss": 0.7969,
+      "step": 3143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207585917728377e-06,
+      "loss": 0.8183,
+      "step": 3144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2016271057889885e-06,
+      "loss": 0.7844,
+      "step": 3145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195669371706327e-06,
+      "loss": 0.842,
+      "step": 3146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1897127177751455e-06,
+      "loss": 0.7868,
+      "step": 3147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183757146289783e-06,
+      "loss": 0.8088,
+      "step": 3148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177802659544164e-06,
+      "loss": 0.8082,
+      "step": 3149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171849259831797e-06,
+      "loss": 0.7327,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1658969494457605e-06,
+      "loss": 0.8043,
+      "step": 3151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159945730678732e-06,
+      "loss": 0.7231,
+      "step": 3152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15399560582295e-06,
+      "loss": 0.8153,
+      "step": 3153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148046577170247e-06,
+      "loss": 0.8052,
+      "step": 3154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142098647012019e-06,
+      "loss": 0.8338,
+      "step": 3155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136151817639249e-06,
+      "loss": 0.7989,
+      "step": 3156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130206091342498e-06,
+      "loss": 0.7945,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.124261470411891e-06,
+      "loss": 0.8549,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118317957137131e-06,
+      "loss": 0.8276,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112375553807507e-06,
+      "loss": 0.9123,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106434262711862e-06,
+      "loss": 0.7966,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100494086138619e-06,
+      "loss": 0.7619,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094555026375776e-06,
+      "loss": 0.3467,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088617085710896e-06,
+      "loss": 0.844,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082680266431104e-06,
+      "loss": 0.8271,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076744570823111e-06,
+      "loss": 0.7832,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07081000117318e-06,
+      "loss": 0.8065,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06487655976714e-06,
+      "loss": 0.7967,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058944248890398e-06,
+      "loss": 0.8042,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053013070827912e-06,
+      "loss": 0.73,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047083027864219e-06,
+      "loss": 0.8277,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041154122283399e-06,
+      "loss": 0.8191,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035226356369107e-06,
+      "loss": 0.8015,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029299732404561e-06,
+      "loss": 0.8359,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02337425267253e-06,
+      "loss": 0.3436,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017449919455346e-06,
+      "loss": 0.8066,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0115267350349035e-06,
+      "loss": 0.7689,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005604701692652e-06,
+      "loss": 0.7839,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9996838217095906e-06,
+      "loss": 0.7911,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993764097366285e-06,
+      "loss": 0.8119,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.987845530942855e-06,
+      "loss": 0.8737,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9819281247189615e-06,
+      "loss": 0.7387,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976011880973834e-06,
+      "loss": 0.781,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970096801986244e-06,
+      "loss": 0.7636,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.964182890034523e-06,
+      "loss": 0.7616,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.958270147396545e-06,
+      "loss": 0.7269,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952358576349733e-06,
+      "loss": 0.8354,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.946448179171073e-06,
+      "loss": 0.7999,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9405389581370795e-06,
+      "loss": 0.7225,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934630915523823e-06,
+      "loss": 0.7931,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9287240536069255e-06,
+      "loss": 0.8106,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922818374661547e-06,
+      "loss": 0.8158,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916913880962388e-06,
+      "loss": 0.8468,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.911010574783705e-06,
+      "loss": 0.7629,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905108458399289e-06,
+      "loss": 0.7507,
+      "step": 3195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899207534082468e-06,
+      "loss": 0.8326,
+      "step": 3196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.893307804106122e-06,
+      "loss": 0.8071,
+      "step": 3197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887409270742664e-06,
+      "loss": 0.8001,
+      "step": 3198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881511936264052e-06,
+      "loss": 0.7127,
+      "step": 3199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.875615802941772e-06,
+      "loss": 0.8724,
+      "step": 3200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869720873046853e-06,
+      "loss": 0.7846,
+      "step": 3201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.863827148849869e-06,
+      "loss": 0.8132,
+      "step": 3202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857934632620913e-06,
+      "loss": 0.771,
+      "step": 3203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852043326629623e-06,
+      "loss": 0.8087,
+      "step": 3204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846153233145172e-06,
+      "loss": 0.7953,
+      "step": 3205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.840264354436261e-06,
+      "loss": 0.805,
+      "step": 3206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834376692771122e-06,
+      "loss": 0.7717,
+      "step": 3207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828490250417526e-06,
+      "loss": 0.7632,
+      "step": 3208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82260502964277e-06,
+      "loss": 0.7512,
+      "step": 3209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.816721032713676e-06,
+      "loss": 0.8312,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810838261896602e-06,
+      "loss": 0.7854,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8049567194574294e-06,
+      "loss": 0.8128,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799076407661564e-06,
+      "loss": 0.8018,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7931973287739485e-06,
+      "loss": 0.7882,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.787319485059036e-06,
+      "loss": 0.8106,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781442878780819e-06,
+      "loss": 0.8212,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.775567512202799e-06,
+      "loss": 0.8161,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7696933875880065e-06,
+      "loss": 0.7846,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763820507199e-06,
+      "loss": 0.797,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757948873297846e-06,
+      "loss": 0.8329,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752078488146139e-06,
+      "loss": 0.8318,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746209354004991e-06,
+      "loss": 0.837,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740341473135037e-06,
+      "loss": 0.8126,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734474847796415e-06,
+      "loss": 0.7236,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728609480248795e-06,
+      "loss": 0.8403,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722745372751356e-06,
+      "loss": 0.7865,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716882527562787e-06,
+      "loss": 0.7944,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711020946941302e-06,
+      "loss": 0.8276,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705160633144616e-06,
+      "loss": 0.8166,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699301588429968e-06,
+      "loss": 0.8038,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693443815054097e-06,
+      "loss": 0.8372,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687587315273256e-06,
+      "loss": 0.7811,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681732091343216e-06,
+      "loss": 0.8093,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67587814551924e-06,
+      "loss": 0.7705,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6700254800561115e-06,
+      "loss": 0.8515,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664174097208119e-06,
+      "loss": 0.8538,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658323999229056e-06,
+      "loss": 0.8477,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652475188372214e-06,
+      "loss": 0.839,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646627666890402e-06,
+      "loss": 0.8377,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640781437035924e-06,
+      "loss": 0.8443,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6349365010605814e-06,
+      "loss": 0.8009,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629092861215691e-06,
+      "loss": 0.8427,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623250519752065e-06,
+      "loss": 0.7813,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6174094789200025e-06,
+      "loss": 0.7956,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611569740969324e-06,
+      "loss": 0.7724,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.605731308149329e-06,
+      "loss": 0.8206,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599894182708831e-06,
+      "loss": 0.7959,
+      "step": 3247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594058366896124e-06,
+      "loss": 0.7995,
+      "step": 3248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5882238629590025e-06,
+      "loss": 0.8272,
+      "step": 3249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5823906731447675e-06,
+      "loss": 0.7989,
+      "step": 3250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576558799700195e-06,
+      "loss": 0.7884,
+      "step": 3251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5707282448715656e-06,
+      "loss": 0.798,
+      "step": 3252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564899010904653e-06,
+      "loss": 0.7817,
+      "step": 3253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559071100044719e-06,
+      "loss": 0.3234,
+      "step": 3254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553244514536509e-06,
+      "loss": 0.8117,
+      "step": 3255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54741925662427e-06,
+      "loss": 0.8081,
+      "step": 3256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541595328551733e-06,
+      "loss": 0.8112,
+      "step": 3257
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535772732562109e-06,
+      "loss": 0.8206,
+      "step": 3258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5299514708981095e-06,
+      "loss": 0.7493,
+      "step": 3259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52413154580192e-06,
+      "loss": 0.8187,
+      "step": 3260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518312959515225e-06,
+      "loss": 0.7634,
+      "step": 3261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.512495714279176e-06,
+      "loss": 0.7673,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506679812334419e-06,
+      "loss": 0.7902,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500865255921085e-06,
+      "loss": 0.7893,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495052047278778e-06,
+      "loss": 0.7898,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489240188646584e-06,
+      "loss": 0.8306,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483429682263079e-06,
+      "loss": 0.8115,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477620530366309e-06,
+      "loss": 0.7399,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.471812735193797e-06,
+      "loss": 0.36,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46600629898255e-06,
+      "loss": 0.7821,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460201223969049e-06,
+      "loss": 0.8177,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454397512389245e-06,
+      "loss": 0.8007,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448595166478577e-06,
+      "loss": 0.8123,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.442794188471944e-06,
+      "loss": 0.8198,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4369945806037284e-06,
+      "loss": 0.7354,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.431196345107777e-06,
+      "loss": 0.7917,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425399484217409e-06,
+      "loss": 0.7627,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419604000165427e-06,
+      "loss": 0.8154,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413809895184084e-06,
+      "loss": 0.7555,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408017171505111e-06,
+      "loss": 0.8609,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402225831359713e-06,
+      "loss": 0.7526,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3964358769785505e-06,
+      "loss": 0.7764,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3906473105917535e-06,
+      "loss": 0.3675,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384860134428927e-06,
+      "loss": 0.7582,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37907435071913e-06,
+      "loss": 0.7896,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373289961690884e-06,
+      "loss": 0.7875,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367506969572182e-06,
+      "loss": 0.7892,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361725376590476e-06,
+      "loss": 0.783,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355945184972671e-06,
+      "loss": 0.8347,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350166396945143e-06,
+      "loss": 0.7938,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344389014733723e-06,
+      "loss": 0.8124,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338613040563702e-06,
+      "loss": 0.7314,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332838476659825e-06,
+      "loss": 0.813,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327065325246295e-06,
+      "loss": 0.7323,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3212935885467775e-06,
+      "loss": 0.8392,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3155232687843805e-06,
+      "loss": 0.8079,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3097543681816765e-06,
+      "loss": 0.789,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30398688896069e-06,
+      "loss": 0.8008,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298220833342898e-06,
+      "loss": 0.812,
+      "step": 3299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2924562035492195e-06,
+      "loss": 0.7933,
+      "step": 3300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286693001800039e-06,
+      "loss": 0.7497,
+      "step": 3301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280931230315186e-06,
+      "loss": 0.8017,
+      "step": 3302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275170891313928e-06,
+      "loss": 0.7683,
+      "step": 3303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269411987014999e-06,
+      "loss": 0.8348,
+      "step": 3304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263654519636566e-06,
+      "loss": 0.7451,
+      "step": 3305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.257898491396255e-06,
+      "loss": 0.8167,
+      "step": 3306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252143904511123e-06,
+      "loss": 0.8492,
+      "step": 3307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24639076119768e-06,
+      "loss": 0.3351,
+      "step": 3308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240639063671887e-06,
+      "loss": 0.8306,
+      "step": 3309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234888814149133e-06,
+      "loss": 0.826,
+      "step": 3310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22914001484426e-06,
+      "loss": 0.8229,
+      "step": 3311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223392667971547e-06,
+      "loss": 0.8509,
+      "step": 3312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217646775744719e-06,
+      "loss": 0.8244,
+      "step": 3313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21190234037693e-06,
+      "loss": 0.8802,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206159364080784e-06,
+      "loss": 0.8302,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200417849068319e-06,
+      "loss": 0.7438,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194677797551003e-06,
+      "loss": 0.7659,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188939211739755e-06,
+      "loss": 0.868,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183202093844916e-06,
+      "loss": 0.861,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177466446076272e-06,
+      "loss": 0.7347,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171732270643034e-06,
+      "loss": 0.8027,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165999569753847e-06,
+      "loss": 0.794,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1602683456167975e-06,
+      "loss": 0.803,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154538600439393e-06,
+      "loss": 0.8582,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14881033642857e-06,
+      "loss": 0.8088,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143083555790709e-06,
+      "loss": 0.7546,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.137358260731601e-06,
+      "loss": 0.8268,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131634453456474e-06,
+      "loss": 0.7577,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125912136169987e-06,
+      "loss": 0.8301,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120191311076219e-06,
+      "loss": 0.7339,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114471980378669e-06,
+      "loss": 0.7545,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108754146280275e-06,
+      "loss": 0.7497,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.103037810983388e-06,
+      "loss": 0.3408,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097322976689777e-06,
+      "loss": 0.7622,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091609645600649e-06,
+      "loss": 0.8334,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085897819916618e-06,
+      "loss": 0.7952,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080187501837728e-06,
+      "loss": 0.8366,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074478693563433e-06,
+      "loss": 0.8128,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068771397292609e-06,
+      "loss": 0.8437,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063065615223556e-06,
+      "loss": 0.7793,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.05736134955398e-06,
+      "loss": 0.7419,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0516586024810075e-06,
+      "loss": 0.3609,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045957376201186e-06,
+      "loss": 0.7969,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04025767291047e-06,
+      "loss": 0.8021,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034559494804224e-06,
+      "loss": 0.7669,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028862844077238e-06,
+      "loss": 0.7705,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023167722923704e-06,
+      "loss": 0.7519,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017474133537222e-06,
+      "loss": 0.7871,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011782078110814e-06,
+      "loss": 0.7555,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006091558836897e-06,
+      "loss": 0.3536,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0004025779073134e-06,
+      "loss": 0.8179,
+      "step": 3351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994715137513294e-06,
+      "loss": 0.7426,
+      "step": 3352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9890292398454876e-06,
+      "loss": 0.7812,
+      "step": 3353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983344887093949e-06,
+      "loss": 0.8116,
+      "step": 3354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977662081448134e-06,
+      "loss": 0.3543,
+      "step": 3355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.971980825096901e-06,
+      "loss": 0.8189,
+      "step": 3356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966301120228519e-06,
+      "loss": 0.7858,
+      "step": 3357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960622969030656e-06,
+      "loss": 0.7598,
+      "step": 3358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.954946373690373e-06,
+      "loss": 0.8372,
+      "step": 3359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949271336394147e-06,
+      "loss": 0.7621,
+      "step": 3360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.943597859327847e-06,
+      "loss": 0.3461,
+      "step": 3361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.937925944676733e-06,
+      "loss": 0.7469,
+      "step": 3362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93225559462548e-06,
+      "loss": 0.7464,
+      "step": 3363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9265868113581525e-06,
+      "loss": 0.8057,
+      "step": 3364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9209195970582015e-06,
+      "loss": 0.8037,
+      "step": 3365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915253953908491e-06,
+      "loss": 0.8232,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909589884091267e-06,
+      "loss": 0.3407,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9039273897881824e-06,
+      "loss": 0.7238,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898266473180265e-06,
+      "loss": 0.8298,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89260713644795e-06,
+      "loss": 0.3324,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886949381771063e-06,
+      "loss": 0.7363,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881293211328809e-06,
+      "loss": 0.8098,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8756386272997935e-06,
+      "loss": 0.7593,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869985631862007e-06,
+      "loss": 0.8026,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8643342271928335e-06,
+      "loss": 0.8576,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858684415469031e-06,
+      "loss": 0.7382,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853036198866759e-06,
+      "loss": 0.7416,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8473895795615554e-06,
+      "loss": 0.7881,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.84174455972834e-06,
+      "loss": 0.7447,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836101141541421e-06,
+      "loss": 0.7809,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830459327174491e-06,
+      "loss": 0.7997,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824819118800629e-06,
+      "loss": 0.3259,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.819180518592271e-06,
+      "loss": 0.7856,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813543528721267e-06,
+      "loss": 0.7357,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807908151358832e-06,
+      "loss": 0.7915,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802274388675545e-06,
+      "loss": 0.7597,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79664224284139e-06,
+      "loss": 0.761,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791011716025714e-06,
+      "loss": 0.8061,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785382810397238e-06,
+      "loss": 0.8182,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.779755528124065e-06,
+      "loss": 0.8786,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77412987137367e-06,
+      "loss": 0.8424,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7685058423129055e-06,
+      "loss": 0.7839,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762883443107983e-06,
+      "loss": 0.8111,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757262675924508e-06,
+      "loss": 0.8145,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751643542927442e-06,
+      "loss": 0.75,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.746026046281121e-06,
+      "loss": 0.823,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740410188149253e-06,
+      "loss": 0.8187,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734795970694913e-06,
+      "loss": 0.7482,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729183396080541e-06,
+      "loss": 0.7958,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7235724664679505e-06,
+      "loss": 0.8144,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717963184018316e-06,
+      "loss": 0.7558,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.712355550892177e-06,
+      "loss": 0.7672,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7067495692494525e-06,
+      "loss": 0.8662,
+      "step": 3403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701145241249402e-06,
+      "loss": 0.795,
+      "step": 3404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6955425690506565e-06,
+      "loss": 0.7664,
+      "step": 3405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689941554811227e-06,
+      "loss": 0.767,
+      "step": 3406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684342200688457e-06,
+      "loss": 0.7726,
+      "step": 3407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.678744508839067e-06,
+      "loss": 0.7251,
+      "step": 3408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673148481419144e-06,
+      "loss": 0.8225,
+      "step": 3409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667554120584116e-06,
+      "loss": 0.7932,
+      "step": 3410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.661961428488774e-06,
+      "loss": 0.784,
+      "step": 3411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656370407287282e-06,
+      "loss": 0.8039,
+      "step": 3412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650781059133146e-06,
+      "loss": 0.7314,
+      "step": 3413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645193386179222e-06,
+      "loss": 0.776,
+      "step": 3414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639607390577731e-06,
+      "loss": 0.7887,
+      "step": 3415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634023074480255e-06,
+      "loss": 0.833,
+      "step": 3416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62844044003771e-06,
+      "loss": 0.84,
+      "step": 3417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622859489400373e-06,
+      "loss": 0.8322,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617280224717882e-06,
+      "loss": 0.7509,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611702648139218e-06,
+      "loss": 0.8063,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606126761812697e-06,
+      "loss": 0.8024,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6005525678860115e-06,
+      "loss": 0.8246,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5949800685061885e-06,
+      "loss": 0.8208,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589409265819589e-06,
+      "loss": 0.7713,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5838401619719475e-06,
+      "loss": 0.7636,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5782727591083275e-06,
+      "loss": 0.7936,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572707059373137e-06,
+      "loss": 0.7856,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5671430649101335e-06,
+      "loss": 0.7731,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.561580777862415e-06,
+      "loss": 0.8222,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556020200372423e-06,
+      "loss": 0.8733,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.550461334581941e-06,
+      "loss": 0.8017,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544904182632088e-06,
+      "loss": 0.7824,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539348746663327e-06,
+      "loss": 0.807,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5337950288154696e-06,
+      "loss": 0.7761,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528243031227647e-06,
+      "loss": 0.7875,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522692756038337e-06,
+      "loss": 0.8029,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517144205385356e-06,
+      "loss": 0.7944,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5115973814058554e-06,
+      "loss": 0.8008,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5060522862363176e-06,
+      "loss": 0.8107,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500508922012558e-06,
+      "loss": 0.7653,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494967290869742e-06,
+      "loss": 0.7595,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489427394942342e-06,
+      "loss": 0.8556,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483889236364175e-06,
+      "loss": 0.7504,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.478352817268399e-06,
+      "loss": 0.8418,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.472818139787479e-06,
+      "loss": 0.7764,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.467285206053223e-06,
+      "loss": 0.8169,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461754018196777e-06,
+      "loss": 0.3582,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4562245783485926e-06,
+      "loss": 0.8584,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450696888638456e-06,
+      "loss": 0.7817,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445170951195492e-06,
+      "loss": 0.7706,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439646768148142e-06,
+      "loss": 0.828,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434124341624154e-06,
+      "loss": 0.8545,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428603673750632e-06,
+      "loss": 0.8329,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423084766653985e-06,
+      "loss": 0.7669,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417567622459931e-06,
+      "loss": 0.345,
+      "step": 3455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4120522432935375e-06,
+      "loss": 0.8068,
+      "step": 3456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406538631279174e-06,
+      "loss": 0.81,
+      "step": 3457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401026788540532e-06,
+      "loss": 0.787,
+      "step": 3458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395516717200622e-06,
+      "loss": 0.8183,
+      "step": 3459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390008419381775e-06,
+      "loss": 0.7772,
+      "step": 3460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384501897205634e-06,
+      "loss": 0.8045,
+      "step": 3461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378997152793163e-06,
+      "loss": 0.7465,
+      "step": 3462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373494188264638e-06,
+      "loss": 0.7843,
+      "step": 3463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36799300573965e-06,
+      "loss": 0.8097,
+      "step": 3464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362493607337105e-06,
+      "loss": 0.3301,
+      "step": 3465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35699599517522e-06,
+      "loss": 0.7888,
+      "step": 3466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351500171371527e-06,
+      "loss": 0.804,
+      "step": 3467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346006138042865e-06,
+      "loss": 0.7787,
+      "step": 3468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340513897305386e-06,
+      "loss": 0.8107,
+      "step": 3469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33502345127455e-06,
+      "loss": 0.8246,
+      "step": 3470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.329534802065125e-06,
+      "loss": 0.8378,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324047951791198e-06,
+      "loss": 0.7675,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318562902566145e-06,
+      "loss": 0.7679,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313079656502652e-06,
+      "loss": 0.8012,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307598215712732e-06,
+      "loss": 0.765,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3021185823076715e-06,
+      "loss": 0.7765,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296640758398078e-06,
+      "loss": 0.829,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291164746093864e-06,
+      "loss": 0.8068,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285690547504243e-06,
+      "loss": 0.7386,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280218164737714e-06,
+      "loss": 0.7848,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2747475999020995e-06,
+      "loss": 0.8249,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269278855104512e-06,
+      "loss": 0.7856,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263811932451352e-06,
+      "loss": 0.8759,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.258346834048338e-06,
+      "loss": 0.7196,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252883562000479e-06,
+      "loss": 0.3334,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2474221184120645e-06,
+      "loss": 0.7915,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241962505386706e-06,
+      "loss": 0.8477,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.236504725027294e-06,
+      "loss": 0.7821,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231048779436014e-06,
+      "loss": 0.7663,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225594670714349e-06,
+      "loss": 0.7341,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220142400963072e-06,
+      "loss": 0.811,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214691972282246e-06,
+      "loss": 0.7949,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2092433867712325e-06,
+      "loss": 0.3338,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203796646528671e-06,
+      "loss": 0.822,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1983517536525015e-06,
+      "loss": 0.7577,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192908710239948e-06,
+      "loss": 0.7548,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.187467518387519e-06,
+      "loss": 0.8322,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182028180191014e-06,
+      "loss": 0.7832,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.17659069774552e-06,
+      "loss": 0.8343,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171155073145402e-06,
+      "loss": 0.795,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1657213084843175e-06,
+      "loss": 0.7763,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160289405855199e-06,
+      "loss": 0.7569,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154859367350278e-06,
+      "loss": 0.8756,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149431195061046e-06,
+      "loss": 0.8103,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144004891078285e-06,
+      "loss": 0.7407,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138580457492073e-06,
+      "loss": 0.774,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133157896391741e-06,
+      "loss": 0.779,
+      "step": 3507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12773720986591e-06,
+      "loss": 0.8109,
+      "step": 3508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1223184000024885e-06,
+      "loss": 0.7645,
+      "step": 3509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116901468888655e-06,
+      "loss": 0.8067,
+      "step": 3510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111486418610851e-06,
+      "loss": 0.8048,
+      "step": 3511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106073251254817e-06,
+      "loss": 0.3423,
+      "step": 3512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100661968905558e-06,
+      "loss": 0.8469,
+      "step": 3513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09525257364734e-06,
+      "loss": 0.7815,
+      "step": 3514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089845067563726e-06,
+      "loss": 0.8387,
+      "step": 3515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084439452737534e-06,
+      "loss": 0.779,
+      "step": 3516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079035731250861e-06,
+      "loss": 0.7893,
+      "step": 3517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073633905185071e-06,
+      "loss": 0.7853,
+      "step": 3518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068233976620802e-06,
+      "loss": 0.807,
+      "step": 3519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0628359476379565e-06,
+      "loss": 0.8292,
+      "step": 3520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057439820315708e-06,
+      "loss": 0.8192,
+      "step": 3521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0520455967325e-06,
+      "loss": 0.8505,
+      "step": 3522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0466532789660364e-06,
+      "loss": 0.825,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041262869093293e-06,
+      "loss": 0.7945,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035874369190505e-06,
+      "loss": 0.7315,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0304877813331795e-06,
+      "loss": 0.7345,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025103107596081e-06,
+      "loss": 0.7699,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0197203500532385e-06,
+      "loss": 0.8013,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014339510777941e-06,
+      "loss": 0.8012,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008960591842744e-06,
+      "loss": 0.8063,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003583595319459e-06,
+      "loss": 0.8192,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998208523279158e-06,
+      "loss": 0.7283,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992835377792167e-06,
+      "loss": 0.7532,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987464160928087e-06,
+      "loss": 0.7416,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9820948747557515e-06,
+      "loss": 0.8419,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976727521343264e-06,
+      "loss": 0.8269,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971362102757994e-06,
+      "loss": 0.8274,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96599862106654e-06,
+      "loss": 0.8504,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.960637078334772e-06,
+      "loss": 0.8311,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955277476627815e-06,
+      "loss": 0.7773,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94991981801004e-06,
+      "loss": 0.8016,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9445641045450634e-06,
+      "loss": 0.7206,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939210338295767e-06,
+      "loss": 0.819,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9338585213242765e-06,
+      "loss": 0.7793,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928508655691955e-06,
+      "loss": 0.8285,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923160743459434e-06,
+      "loss": 0.8106,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91781478668658e-06,
+      "loss": 0.7765,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91247078743251e-06,
+      "loss": 0.8205,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.907128747755587e-06,
+      "loss": 0.8031,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9017886697134166e-06,
+      "loss": 0.7353,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896450555362852e-06,
+      "loss": 0.8187,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891114406759989e-06,
+      "loss": 0.7887,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885780225960165e-06,
+      "loss": 0.7859,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880448015017962e-06,
+      "loss": 0.7997,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875117775987202e-06,
+      "loss": 0.769,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869789510920947e-06,
+      "loss": 0.8181,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864463221871497e-06,
+      "loss": 0.7563,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8591389108903965e-06,
+      "loss": 0.8123,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853816580028422e-06,
+      "loss": 0.8041,
+      "step": 3559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84849623133559e-06,
+      "loss": 0.8291,
+      "step": 3560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.843177866861149e-06,
+      "loss": 0.7772,
+      "step": 3561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837861488653603e-06,
+      "loss": 0.7789,
+      "step": 3562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832547098760658e-06,
+      "loss": 0.7392,
+      "step": 3563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8272346992292765e-06,
+      "loss": 0.8056,
+      "step": 3564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821924292105659e-06,
+      "loss": 0.8433,
+      "step": 3565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.816615879435217e-06,
+      "loss": 0.8042,
+      "step": 3566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811309463262607e-06,
+      "loss": 0.7681,
+      "step": 3567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806005045631727e-06,
+      "loss": 0.7557,
+      "step": 3568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800702628585683e-06,
+      "loss": 0.8335,
+      "step": 3569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.795402214166818e-06,
+      "loss": 0.7441,
+      "step": 3570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790103804416717e-06,
+      "loss": 0.7825,
+      "step": 3571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784807401376182e-06,
+      "loss": 0.8279,
+      "step": 3572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779513007085231e-06,
+      "loss": 0.8185,
+      "step": 3573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774220623583132e-06,
+      "loss": 0.8054,
+      "step": 3574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768930252908365e-06,
+      "loss": 0.784,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763641897098626e-06,
+      "loss": 0.8695,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758355558190856e-06,
+      "loss": 0.7412,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753071238221204e-06,
+      "loss": 0.8239,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747788939225047e-06,
+      "loss": 0.8711,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742508663236979e-06,
+      "loss": 0.7635,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737230412290822e-06,
+      "loss": 0.7944,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731954188419611e-06,
+      "loss": 0.7386,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726679993655604e-06,
+      "loss": 0.7684,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721407830030276e-06,
+      "loss": 0.7931,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716137699574323e-06,
+      "loss": 0.3502,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710869604317653e-06,
+      "loss": 0.8086,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705603546289392e-06,
+      "loss": 0.7566,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700339527517881e-06,
+      "loss": 0.7909,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695077550030678e-06,
+      "loss": 0.7558,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689817615854552e-06,
+      "loss": 0.3416,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684559727015486e-06,
+      "loss": 0.352,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67930388553867e-06,
+      "loss": 0.8138,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674050093448523e-06,
+      "loss": 0.7793,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668798352768651e-06,
+      "loss": 0.7949,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663548665521881e-06,
+      "loss": 0.7927,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658301033730258e-06,
+      "loss": 0.8135,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653055459415018e-06,
+      "loss": 0.3513,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647811944596611e-06,
+      "loss": 0.7839,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642570491294703e-06,
+      "loss": 0.8191,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637331101528161e-06,
+      "loss": 0.8176,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63209377731504e-06,
+      "loss": 0.8146,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626858520672627e-06,
+      "loss": 0.7342,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621625333617399e-06,
+      "loss": 0.7785,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616394218165027e-06,
+      "loss": 0.7369,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611165176330402e-06,
+      "loss": 0.7819,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605938210127607e-06,
+      "loss": 0.8426,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600713321569926e-06,
+      "loss": 0.7465,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595490512669841e-06,
+      "loss": 0.8006,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590269785439039e-06,
+      "loss": 0.7598,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585051141888397e-06,
+      "loss": 0.8391,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579834584027997e-06,
+      "loss": 0.3508,
+      "step": 3611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574620113867113e-06,
+      "loss": 0.8049,
+      "step": 3612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5694077334142176e-06,
+      "loss": 0.7533,
+      "step": 3613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564197444676976e-06,
+      "loss": 0.7801,
+      "step": 3614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558989249662248e-06,
+      "loss": 0.8072,
+      "step": 3615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55378315037609e-06,
+      "loss": 0.7861,
+      "step": 3616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548579148823748e-06,
+      "loss": 0.7947,
+      "step": 3617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543377247009663e-06,
+      "loss": 0.8533,
+      "step": 3618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538177446937462e-06,
+      "loss": 0.8118,
+      "step": 3619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532979750609968e-06,
+      "loss": 0.7872,
+      "step": 3620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5277841600291915e-06,
+      "loss": 0.7657,
+      "step": 3621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522590677196331e-06,
+      "loss": 0.7814,
+      "step": 3622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517399304111773e-06,
+      "loss": 0.8545,
+      "step": 3623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512210042775099e-06,
+      "loss": 0.7861,
+      "step": 3624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507022895185062e-06,
+      "loss": 0.7824,
+      "step": 3625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501837863339611e-06,
+      "loss": 0.7986,
+      "step": 3626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496654949235887e-06,
+      "loss": 0.7789,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491474154870197e-06,
+      "loss": 0.3644,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48629548223804e-06,
+      "loss": 0.8014,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481118933334109e-06,
+      "loss": 0.8049,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475944510152266e-06,
+      "loss": 0.777,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470772214685549e-06,
+      "loss": 0.7812,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465602048926194e-06,
+      "loss": 0.3525,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460434014865609e-06,
+      "loss": 0.8366,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455268114494366e-06,
+      "loss": 0.8428,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450104349802243e-06,
+      "loss": 0.3453,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444942722778175e-06,
+      "loss": 0.7684,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4397832354102796e-06,
+      "loss": 0.7535,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434625889685851e-06,
+      "loss": 0.8125,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4294706875913575e-06,
+      "loss": 0.7754,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424317631112441e-06,
+      "loss": 0.8184,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419166722233919e-06,
+      "loss": 0.7274,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414017962939783e-06,
+      "loss": 0.7931,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40887135521319e-06,
+      "loss": 0.7841,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403726901036475e-06,
+      "loss": 0.8177,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398584602391139e-06,
+      "loss": 0.8521,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.393444461257858e-06,
+      "loss": 0.7749,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388306479616473e-06,
+      "loss": 0.865,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383170659445992e-06,
+      "loss": 0.7691,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378037002724596e-06,
+      "loss": 0.7972,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372905511429625e-06,
+      "loss": 0.7471,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3677761875375915e-06,
+      "loss": 0.7972,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36264903302417e-06,
+      "loss": 0.7843,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357524049864198e-06,
+      "loss": 0.7604,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352401240031689e-06,
+      "loss": 0.3417,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3472806054997974e-06,
+      "loss": 0.7839,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342162148240853e-06,
+      "loss": 0.766,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337045870226355e-06,
+      "loss": 0.7591,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331931773426945e-06,
+      "loss": 0.7731,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326819859812431e-06,
+      "loss": 0.7872,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3217101313517915e-06,
+      "loss": 0.7933,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316602590013153e-06,
+      "loss": 0.7793,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311497237763791e-06,
+      "loss": 0.7908,
+      "step": 3663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3063940765701574e-06,
+      "loss": 0.7925,
+      "step": 3664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30129310839785e-06,
+      "loss": 0.7383,
+      "step": 3665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296194335211613e-06,
+      "loss": 0.7948,
+      "step": 3666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2910977589753635e-06,
+      "loss": 0.8542,
+      "step": 3667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286003381652161e-06,
+      "loss": 0.7461,
+      "step": 3668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28091120520422e-06,
+      "loss": 0.7917,
+      "step": 3669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275821231592908e-06,
+      "loss": 0.7761,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270733462778742e-06,
+      "loss": 0.3367,
+      "step": 3671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265647900721393e-06,
+      "loss": 0.7685,
+      "step": 3672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2605645473796806e-06,
+      "loss": 0.833,
+      "step": 3673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2554834047115734e-06,
+      "loss": 0.8061,
+      "step": 3674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250404474674188e-06,
+      "loss": 0.8772,
+      "step": 3675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24532775922379e-06,
+      "loss": 0.8008,
+      "step": 3676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24025326031579e-06,
+      "loss": 0.8001,
+      "step": 3677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2351809799047475e-06,
+      "loss": 0.8308,
+      "step": 3678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230110919944365e-06,
+      "loss": 0.8581,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225043082387494e-06,
+      "loss": 0.7891,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219977469186124e-06,
+      "loss": 0.7725,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214914082291387e-06,
+      "loss": 0.8184,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2098529236535746e-06,
+      "loss": 0.7721,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204793995222093e-06,
+      "loss": 0.7877,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199737298945506e-06,
+      "loss": 0.7833,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194682836771522e-06,
+      "loss": 0.788,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1896306106469735e-06,
+      "loss": 0.9042,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184580622517839e-06,
+      "loss": 0.7692,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179532874329248e-06,
+      "loss": 0.8119,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174487368025442e-06,
+      "loss": 0.7765,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169444105549812e-06,
+      "loss": 0.7457,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164403088844895e-06,
+      "loss": 0.7526,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159364319852349e-06,
+      "loss": 0.7882,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1543278005129625e-06,
+      "loss": 0.8415,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149293532766675e-06,
+      "loss": 0.7659,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144261518552548e-06,
+      "loss": 0.3213,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139231759808765e-06,
+      "loss": 0.8067,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134204258472662e-06,
+      "loss": 0.3169,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129179016480693e-06,
+      "loss": 0.7812,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124156035768443e-06,
+      "loss": 0.7853,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119135318270626e-06,
+      "loss": 0.373,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114116865921086e-06,
+      "loss": 0.8244,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109100680652793e-06,
+      "loss": 0.7973,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1040867643978455e-06,
+      "loss": 0.7859,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099075119087467e-06,
+      "loss": 0.7843,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094065746652004e-06,
+      "loss": 0.7991,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0890586490209325e-06,
+      "loss": 0.8275,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084053828122848e-06,
+      "loss": 0.3277,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079051285885471e-06,
+      "loss": 0.8046,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0740510242356445e-06,
+      "loss": 0.3369,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069053045099332e-06,
+      "loss": 0.6846,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0640573504016176e-06,
+      "loss": 0.785,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059063942066704e-06,
+      "loss": 0.8239,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054072822017927e-06,
+      "loss": 0.8244,
+      "step": 3714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049083992177717e-06,
+      "loss": 0.7418,
+      "step": 3715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044097454467636e-06,
+      "loss": 0.7689,
+      "step": 3716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039113210808373e-06,
+      "loss": 0.7879,
+      "step": 3717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034131263119709e-06,
+      "loss": 0.8269,
+      "step": 3718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029151613320558e-06,
+      "loss": 0.3202,
+      "step": 3719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024174263328949e-06,
+      "loss": 0.786,
+      "step": 3720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019199215062023e-06,
+      "loss": 0.785,
+      "step": 3721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014226470436021e-06,
+      "loss": 0.8154,
+      "step": 3722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00925603136632e-06,
+      "loss": 0.7349,
+      "step": 3723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004287899767394e-06,
+      "loss": 0.7739,
+      "step": 3724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999322077552825e-06,
+      "loss": 0.8077,
+      "step": 3725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9943585666353205e-06,
+      "loss": 0.8073,
+      "step": 3726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9893973689266854e-06,
+      "loss": 0.8144,
+      "step": 3727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984438486337837e-06,
+      "loss": 0.7589,
+      "step": 3728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979481920778803e-06,
+      "loss": 0.7898,
+      "step": 3729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974527674158715e-06,
+      "loss": 0.7626,
+      "step": 3730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969575748385816e-06,
+      "loss": 0.8054,
+      "step": 3731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.964626145367449e-06,
+      "loss": 0.7712,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95967886701007e-06,
+      "loss": 0.8678,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954733915219231e-06,
+      "loss": 0.8024,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949791291899594e-06,
+      "loss": 0.8391,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944850998954923e-06,
+      "loss": 0.7553,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939913038288084e-06,
+      "loss": 0.7907,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934977411801042e-06,
+      "loss": 0.7648,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930044121394868e-06,
+      "loss": 0.7866,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925113168969729e-06,
+      "loss": 0.8034,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920184556424894e-06,
+      "loss": 0.3268,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9152582856587305e-06,
+      "loss": 0.832,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910334358568701e-06,
+      "loss": 0.767,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905412777051365e-06,
+      "loss": 0.7911,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900493543002393e-06,
+      "loss": 0.7376,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895576658316528e-06,
+      "loss": 0.7792,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890662124887619e-06,
+      "loss": 0.8032,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88574994460862e-06,
+      "loss": 0.7897,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880840119371559e-06,
+      "loss": 0.7051,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875932651067567e-06,
+      "loss": 0.8475,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8710275415868724e-06,
+      "loss": 0.7446,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866124792818791e-06,
+      "loss": 0.7838,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861224406651715e-06,
+      "loss": 0.8044,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85632638497315e-06,
+      "loss": 0.8244,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85143072966968e-06,
+      "loss": 0.747,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84653744262697e-06,
+      "loss": 0.8037,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8416465257297865e-06,
+      "loss": 0.7898,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836757980861976e-06,
+      "loss": 0.7459,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.831871809906472e-06,
+      "loss": 0.7431,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826988014745293e-06,
+      "loss": 0.8053,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822106597259545e-06,
+      "loss": 0.7731,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817227559329418e-06,
+      "loss": 0.8139,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8123509028341755e-06,
+      "loss": 0.3326,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8074766296521813e-06,
+      "loss": 0.7489,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802604741660869e-06,
+      "loss": 0.8064,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7977352407367573e-06,
+      "loss": 0.8179,
+      "step": 3766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.792868128755444e-06,
+      "loss": 0.7812,
+      "step": 3767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7880034075916082e-06,
+      "loss": 0.8836,
+      "step": 3768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.783141079119007e-06,
+      "loss": 0.8222,
+      "step": 3769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7782811452104763e-06,
+      "loss": 0.7505,
+      "step": 3770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7734236077379304e-06,
+      "loss": 0.8164,
+      "step": 3771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76856846857236e-06,
+      "loss": 0.7396,
+      "step": 3772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7637157295838313e-06,
+      "loss": 0.8333,
+      "step": 3773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.758865392641485e-06,
+      "loss": 0.8184,
+      "step": 3774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7540174596135393e-06,
+      "loss": 0.7436,
+      "step": 3775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7491719323672916e-06,
+      "loss": 0.829,
+      "step": 3776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.744328812769098e-06,
+      "loss": 0.7992,
+      "step": 3777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7394881026843956e-06,
+      "loss": 0.7883,
+      "step": 3778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7346498039777044e-06,
+      "loss": 0.7692,
+      "step": 3779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7298139185125927e-06,
+      "loss": 0.7911,
+      "step": 3780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7249804481517173e-06,
+      "loss": 0.8178,
+      "step": 3781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.720149394756792e-06,
+      "loss": 0.7847,
+      "step": 3782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.715320760188621e-06,
+      "loss": 0.7351,
+      "step": 3783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7104945463070484e-06,
+      "loss": 0.7814,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7056707549710012e-06,
+      "loss": 0.7625,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.700849388038482e-06,
+      "loss": 0.7898,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.696030447366541e-06,
+      "loss": 0.7885,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6912139348112994e-06,
+      "loss": 0.8025,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6863998522279555e-06,
+      "loss": 0.839,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6815882014707605e-06,
+      "loss": 0.8193,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6767789843930213e-06,
+      "loss": 0.7489,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.671972202847126e-06,
+      "loss": 0.8171,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6671678586845173e-06,
+      "loss": 0.8428,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6623659537556867e-06,
+      "loss": 0.7689,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.657566489910205e-06,
+      "loss": 0.3424,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.652769468996693e-06,
+      "loss": 0.8182,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.647974892862831e-06,
+      "loss": 0.8638,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6431827633553587e-06,
+      "loss": 0.7217,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638393082320074e-06,
+      "loss": 0.8449,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63360585160183e-06,
+      "loss": 0.7701,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6288210730445382e-06,
+      "loss": 0.775,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6240387484911642e-06,
+      "loss": 0.7518,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619258879783728e-06,
+      "loss": 0.3507,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6144814687633046e-06,
+      "loss": 0.7459,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609706517270023e-06,
+      "loss": 0.7927,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.604934027143062e-06,
+      "loss": 0.7664,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6001640002206573e-06,
+      "loss": 0.7893,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5953964383400896e-06,
+      "loss": 0.7765,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590631343337695e-06,
+      "loss": 0.8299,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.585868717048855e-06,
+      "loss": 0.8682,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581108561308012e-06,
+      "loss": 0.8587,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5763508779486377e-06,
+      "loss": 0.7294,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5715956688032627e-06,
+      "loss": 0.7797,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.566842935703473e-06,
+      "loss": 0.8364,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5620926804798805e-06,
+      "loss": 0.7605,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5573449049621554e-06,
+      "loss": 0.8254,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5525996109790195e-06,
+      "loss": 0.8395,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5478568003582216e-06,
+      "loss": 0.3569,
+      "step": 3818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5431164749265624e-06,
+      "loss": 0.7982,
+      "step": 3819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5383786365098904e-06,
+      "loss": 0.8153,
+      "step": 3820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.533643286933094e-06,
+      "loss": 0.8276,
+      "step": 3821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.528910428020089e-06,
+      "loss": 0.7841,
+      "step": 3822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5241800615938536e-06,
+      "loss": 0.7774,
+      "step": 3823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5194521894763955e-06,
+      "loss": 0.3682,
+      "step": 3824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5147268134887526e-06,
+      "loss": 0.8157,
+      "step": 3825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5100039354510186e-06,
+      "loss": 0.806,
+      "step": 3826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.505283557182315e-06,
+      "loss": 0.7778,
+      "step": 3827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5005656805008027e-06,
+      "loss": 0.8325,
+      "step": 3828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4958503072236782e-06,
+      "loss": 0.8003,
+      "step": 3829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4911374391671737e-06,
+      "loss": 0.7749,
+      "step": 3830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.486427078146559e-06,
+      "loss": 0.7756,
+      "step": 3831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4817192259761346e-06,
+      "loss": 0.7284,
+      "step": 3832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.477013884469238e-06,
+      "loss": 0.8332,
+      "step": 3833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472311055438237e-06,
+      "loss": 0.7471,
+      "step": 3834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4676107406945315e-06,
+      "loss": 0.8276,
+      "step": 3835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4629129420485575e-06,
+      "loss": 0.7847,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458217661309775e-06,
+      "loss": 0.8023,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453524900286681e-06,
+      "loss": 0.7712,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448834660786796e-06,
+      "loss": 0.764,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444146944616674e-06,
+      "loss": 0.7539,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.439461753581892e-06,
+      "loss": 0.8327,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4347790894870647e-06,
+      "loss": 0.8072,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4300989541358186e-06,
+      "loss": 0.2794,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4254213493308143e-06,
+      "loss": 0.8055,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4207462768737453e-06,
+      "loss": 0.8566,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416073738565313e-06,
+      "loss": 0.8028,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411403736205251e-06,
+      "loss": 0.3681,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4067362715923237e-06,
+      "loss": 0.795,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4020713465243105e-06,
+      "loss": 0.7482,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3974089627980024e-06,
+      "loss": 0.8049,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3927491222092335e-06,
+      "loss": 0.8401,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3880918265528453e-06,
+      "loss": 0.8423,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.383437077622691e-06,
+      "loss": 0.7311,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3787848772116636e-06,
+      "loss": 0.7676,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3741352271116644e-06,
+      "loss": 0.8336,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3694881291135983e-06,
+      "loss": 0.7463,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3648435850074136e-06,
+      "loss": 0.327,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360201596582057e-06,
+      "loss": 0.7804,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3555621656254944e-06,
+      "loss": 0.8123,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3509252939247094e-06,
+      "loss": 0.7673,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3462909832656986e-06,
+      "loss": 0.8022,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3416592354334697e-06,
+      "loss": 0.7925,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.337030052212047e-06,
+      "loss": 0.7511,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.332403435384465e-06,
+      "loss": 0.7785,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.327779386732769e-06,
+      "loss": 0.3295,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3231579080380174e-06,
+      "loss": 0.7855,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3185390010802767e-06,
+      "loss": 0.8096,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3139226676386225e-06,
+      "loss": 0.3304,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3093089094911434e-06,
+      "loss": 0.8325,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3046977284149294e-06,
+      "loss": 0.7125,
+      "step": 3870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3000891261860844e-06,
+      "loss": 0.7335,
+      "step": 3871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2954831045797108e-06,
+      "loss": 0.8455,
+      "step": 3872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2908796653699314e-06,
+      "loss": 0.7618,
+      "step": 3873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2862788103298573e-06,
+      "loss": 0.8395,
+      "step": 3874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.281680541231611e-06,
+      "loss": 0.7498,
+      "step": 3875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27708485984633e-06,
+      "loss": 0.7828,
+      "step": 3876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.272491767944135e-06,
+      "loss": 0.7681,
+      "step": 3877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267901267294159e-06,
+      "loss": 0.8254,
+      "step": 3878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263313359664545e-06,
+      "loss": 0.7909,
+      "step": 3879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2587280468224293e-06,
+      "loss": 0.7671,
+      "step": 3880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254145330533939e-06,
+      "loss": 0.8074,
+      "step": 3881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2495652125642207e-06,
+      "loss": 0.8138,
+      "step": 3882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2449876946774115e-06,
+      "loss": 0.784,
+      "step": 3883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2404127786366345e-06,
+      "loss": 0.7974,
+      "step": 3884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235840466204034e-06,
+      "loss": 0.3729,
+      "step": 3885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2312707591407356e-06,
+      "loss": 0.8217,
+      "step": 3886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2267036592068656e-06,
+      "loss": 0.3321,
+      "step": 3887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2221391681615455e-06,
+      "loss": 0.7524,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217577287762894e-06,
+      "loss": 0.7989,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2130180197680207e-06,
+      "loss": 0.8043,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.208461365933032e-06,
+      "loss": 0.7415,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.203907328013027e-06,
+      "loss": 0.7959,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199355907762095e-06,
+      "loss": 0.8342,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1948071069333207e-06,
+      "loss": 0.8307,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1902609272787755e-06,
+      "loss": 0.7872,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185717370549525e-06,
+      "loss": 0.7425,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1811764384956223e-06,
+      "loss": 0.8073,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.176638132866111e-06,
+      "loss": 0.7978,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.172102455409022e-06,
+      "loss": 0.7715,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1675694078713738e-06,
+      "loss": 0.7587,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1630389919991733e-06,
+      "loss": 0.8222,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1585112095374117e-06,
+      "loss": 0.8096,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1539860622300646e-06,
+      "loss": 0.7873,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1494635518201046e-06,
+      "loss": 0.7487,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144943680049469e-06,
+      "loss": 0.7806,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1404264486590885e-06,
+      "loss": 0.7336,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1359118593888883e-06,
+      "loss": 0.3831,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131399913977754e-06,
+      "loss": 0.7874,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1268906141635626e-06,
+      "loss": 0.7631,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1223839616831817e-06,
+      "loss": 0.7841,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1178799582724496e-06,
+      "loss": 0.8247,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.113378605666174e-06,
+      "loss": 0.7638,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1088799055981656e-06,
+      "loss": 0.7738,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104383859801199e-06,
+      "loss": 0.7963,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0998904700070186e-06,
+      "loss": 0.7968,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095399737946364e-06,
+      "loss": 0.8251,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0909116653489413e-06,
+      "loss": 0.8361,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086426253943431e-06,
+      "loss": 0.8029,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081943505457492e-06,
+      "loss": 0.8202,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.077463421617758e-06,
+      "loss": 0.8045,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0729860041498318e-06,
+      "loss": 0.7647,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068511254778295e-06,
+      "loss": 0.7666,
+      "step": 3922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0640391752266962e-06,
+      "loss": 0.3607,
+      "step": 3923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0595697672175596e-06,
+      "loss": 0.881,
+      "step": 3924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0551030324723786e-06,
+      "loss": 0.3569,
+      "step": 3925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050638972711618e-06,
+      "loss": 0.7358,
+      "step": 3926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.046177589654711e-06,
+      "loss": 0.7748,
+      "step": 3927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0417188850200597e-06,
+      "loss": 0.8647,
+      "step": 3928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0372628605250364e-06,
+      "loss": 0.825,
+      "step": 3929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0328095178859796e-06,
+      "loss": 0.8032,
+      "step": 3930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.028358858818191e-06,
+      "loss": 0.7985,
+      "step": 3931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.023910885035953e-06,
+      "loss": 0.7164,
+      "step": 3932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0194655982524946e-06,
+      "loss": 0.7984,
+      "step": 3933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0150230001800173e-06,
+      "loss": 0.8119,
+      "step": 3934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0105830925296998e-06,
+      "loss": 0.7921,
+      "step": 3935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0061458770116613e-06,
+      "loss": 0.8047,
+      "step": 3936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0017113553349975e-06,
+      "loss": 0.7555,
+      "step": 3937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997279529207774e-06,
+      "loss": 0.6577,
+      "step": 3938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9928504003369995e-06,
+      "loss": 0.7448,
+      "step": 3939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9884239704286545e-06,
+      "loss": 0.7684,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9840002411876822e-06,
+      "loss": 0.7662,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9795792143179846e-06,
+      "loss": 0.7999,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9751608915224106e-06,
+      "loss": 0.8345,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9707452745027864e-06,
+      "loss": 0.8155,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9663323649598873e-06,
+      "loss": 0.8248,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.961922164593435e-06,
+      "loss": 0.8251,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9575146751021276e-06,
+      "loss": 0.7617,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9531098981836082e-06,
+      "loss": 0.7637,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9487078355344744e-06,
+      "loss": 0.7492,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9443084888502815e-06,
+      "loss": 0.829,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9399118598255373e-06,
+      "loss": 0.7448,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9355179501537024e-06,
+      "loss": 0.8557,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9311267615271922e-06,
+      "loss": 0.757,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9267382956373702e-06,
+      "loss": 0.8275,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9223525541745555e-06,
+      "loss": 0.3217,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.917969538828016e-06,
+      "loss": 0.777,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9135892512859677e-06,
+      "loss": 0.7355,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9092116932355772e-06,
+      "loss": 0.848,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904836866362963e-06,
+      "loss": 0.7852,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9004647723531866e-06,
+      "loss": 0.791,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8960954128902596e-06,
+      "loss": 0.7552,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8917287896571365e-06,
+      "loss": 0.7876,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887364904335731e-06,
+      "loss": 0.7226,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8830037586068826e-06,
+      "loss": 0.7327,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8786453541503846e-06,
+      "loss": 0.746,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874289692644987e-06,
+      "loss": 0.7851,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.869936775768359e-06,
+      "loss": 0.7642,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.865586605197128e-06,
+      "loss": 0.7633,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8612391826068664e-06,
+      "loss": 0.755,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8568945096720813e-06,
+      "loss": 0.8526,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.852552588066213e-06,
+      "loss": 0.7837,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8482134194616607e-06,
+      "loss": 0.7491,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843877005529754e-06,
+      "loss": 0.7337,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.83954334794075e-06,
+      "loss": 0.8195,
+      "step": 3974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8352124483638653e-06,
+      "loss": 0.8082,
+      "step": 3975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8308843084672442e-06,
+      "loss": 0.7605,
+      "step": 3976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82655892991796e-06,
+      "loss": 0.7995,
+      "step": 3977
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8222363143820365e-06,
+      "loss": 0.7785,
+      "step": 3978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.817916463524425e-06,
+      "loss": 0.3071,
+      "step": 3979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8135993790090154e-06,
+      "loss": 0.8074,
+      "step": 3980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809285062498629e-06,
+      "loss": 0.8292,
+      "step": 3981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804973515655024e-06,
+      "loss": 0.8192,
+      "step": 3982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8006647401388886e-06,
+      "loss": 0.2919,
+      "step": 3983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7963587376098465e-06,
+      "loss": 0.7469,
+      "step": 3984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7920555097264525e-06,
+      "loss": 0.742,
+      "step": 3985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.78775505814619e-06,
+      "loss": 0.8408,
+      "step": 3986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7834573845254786e-06,
+      "loss": 0.7384,
+      "step": 3987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7791624905196624e-06,
+      "loss": 0.8043,
+      "step": 3988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774870377783018e-06,
+      "loss": 0.7467,
+      "step": 3989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7705810479687488e-06,
+      "loss": 0.7425,
+      "step": 3990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.766294502728989e-06,
+      "loss": 0.3315,
+      "step": 3991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762010743714798e-06,
+      "loss": 0.8142,
+      "step": 3992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7577297725761597e-06,
+      "loss": 0.8597,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.753451590961995e-06,
+      "loss": 0.7657,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.749176200520136e-06,
+      "loss": 0.786,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7449036028973432e-06,
+      "loss": 0.343,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740633799739315e-06,
+      "loss": 0.3378,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7363667926906546e-06,
+      "loss": 0.732,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7321025833948946e-06,
+      "loss": 0.7723,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.727841173494499e-06,
+      "loss": 0.726,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7235825646308455e-06,
+      "loss": 0.8209,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7193267584442262e-06,
+      "loss": 0.7906,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.715073756573869e-06,
+      "loss": 0.7919,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.710823560657915e-06,
+      "loss": 0.7818,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.706576172333413e-06,
+      "loss": 0.8306,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7023315932363515e-06,
+      "loss": 0.7203,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.698089825001622e-06,
+      "loss": 0.7965,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.693850869263038e-06,
+      "loss": 0.7833,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.689614727653329e-06,
+      "loss": 0.7695,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6853814018041425e-06,
+      "loss": 0.7716,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6811508933460383e-06,
+      "loss": 0.7513,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6769232039084925e-06,
+      "loss": 0.8634,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6726983351198965e-06,
+      "loss": 0.8475,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.668476288607554e-06,
+      "loss": 0.7785,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.664257065997682e-06,
+      "loss": 0.8188,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.660040668915409e-06,
+      "loss": 0.8155,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6558270989847756e-06,
+      "loss": 0.7708,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6516163578287346e-06,
+      "loss": 0.7315,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6474084470691485e-06,
+      "loss": 0.706,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.643203368326789e-06,
+      "loss": 0.8131,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6390011232213365e-06,
+      "loss": 0.7816,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6348017133713832e-06,
+      "loss": 0.7098,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6306051403944246e-06,
+      "loss": 0.823,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6264114059068645e-06,
+      "loss": 0.816,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622220511524023e-06,
+      "loss": 0.3383,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.618032458860109e-06,
+      "loss": 0.8023,
+      "step": 4026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6138472495282475e-06,
+      "loss": 0.8011,
+      "step": 4027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6096648851404725e-06,
+      "loss": 0.8181,
+      "step": 4028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.605485367307711e-06,
+      "loss": 0.7768,
+      "step": 4029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.601308697639795e-06,
+      "loss": 0.7361,
+      "step": 4030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597134877745473e-06,
+      "loss": 0.7624,
+      "step": 4031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5929639092323846e-06,
+      "loss": 0.8337,
+      "step": 4032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588795793707064e-06,
+      "loss": 0.3453,
+      "step": 4033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5846305327749653e-06,
+      "loss": 0.7723,
+      "step": 4034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580468128040432e-06,
+      "loss": 0.7253,
+      "step": 4035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576308581106699e-06,
+      "loss": 0.8011,
+      "step": 4036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5721518935759214e-06,
+      "loss": 0.7884,
+      "step": 4037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5679980670491356e-06,
+      "loss": 0.7477,
+      "step": 4038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5638471031262848e-06,
+      "loss": 0.7957,
+      "step": 4039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5596990034062054e-06,
+      "loss": 0.7406,
+      "step": 4040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5555537694866315e-06,
+      "loss": 0.7746,
+      "step": 4041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5514114029641944e-06,
+      "loss": 0.6868,
+      "step": 4042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547271905434421e-06,
+      "loss": 0.731,
+      "step": 4043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.543135278491732e-06,
+      "loss": 0.7948,
+      "step": 4044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.539001523729443e-06,
+      "loss": 0.7736,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534870642739764e-06,
+      "loss": 0.781,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5307426371137965e-06,
+      "loss": 0.8005,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5266175084415355e-06,
+      "loss": 0.8293,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5224952583118677e-06,
+      "loss": 0.7626,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5183758883125718e-06,
+      "loss": 0.7639,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5142594000303156e-06,
+      "loss": 0.7745,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5101457950506556e-06,
+      "loss": 0.7418,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5060350749580487e-06,
+      "loss": 0.7325,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.501927241335823e-06,
+      "loss": 0.7363,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4978222957662037e-06,
+      "loss": 0.7916,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493720239830312e-06,
+      "loss": 0.7908,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4896210751081396e-06,
+      "loss": 0.331,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4855248031785715e-06,
+      "loss": 0.7767,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4814314256193895e-06,
+      "loss": 0.7817,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4773409440072416e-06,
+      "loss": 0.8743,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47325335991767e-06,
+      "loss": 0.7902,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4691686749251053e-06,
+      "loss": 0.8304,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465086890602857e-06,
+      "loss": 0.7561,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461008008523108e-06,
+      "loss": 0.815,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45693203025694e-06,
+      "loss": 0.851,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45285895737431e-06,
+      "loss": 0.3278,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4487887914440446e-06,
+      "loss": 0.7882,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444721534033869e-06,
+      "loss": 0.8008,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440657186710378e-06,
+      "loss": 0.7887,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436595751039045e-06,
+      "loss": 0.7035,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4325372285842253e-06,
+      "loss": 0.8158,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42848162090915e-06,
+      "loss": 0.7905,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4244289295759293e-06,
+      "loss": 0.7966,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420379156145548e-06,
+      "loss": 0.3177,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4163323021778694e-06,
+      "loss": 0.7747,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4122883692316303e-06,
+      "loss": 0.7523,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4082473588644417e-06,
+      "loss": 0.7518,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4042092726327924e-06,
+      "loss": 0.7293,
+      "step": 4078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400174112092042e-06,
+      "loss": 0.7358,
+      "step": 4079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3961418787964243e-06,
+      "loss": 0.7854,
+      "step": 4080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3921125742990446e-06,
+      "loss": 0.7927,
+      "step": 4081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388086200151881e-06,
+      "loss": 0.3369,
+      "step": 4082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.384062757905782e-06,
+      "loss": 0.7679,
+      "step": 4083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3800422491104725e-06,
+      "loss": 0.7679,
+      "step": 4084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3760246753145356e-06,
+      "loss": 0.7601,
+      "step": 4085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372010038065432e-06,
+      "loss": 0.7585,
+      "step": 4086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3679983389094975e-06,
+      "loss": 0.7744,
+      "step": 4087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3639895793919198e-06,
+      "loss": 0.8027,
+      "step": 4088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3599837610567653e-06,
+      "loss": 0.8278,
+      "step": 4089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35598088544697e-06,
+      "loss": 0.7307,
+      "step": 4090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3519809541043327e-06,
+      "loss": 0.7162,
+      "step": 4091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347983968569508e-06,
+      "loss": 0.7832,
+      "step": 4092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3439899303820356e-06,
+      "loss": 0.7999,
+      "step": 4093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3399988410803097e-06,
+      "loss": 0.7085,
+      "step": 4094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336010702201579e-06,
+      "loss": 0.7682,
+      "step": 4095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332025515281975e-06,
+      "loss": 0.7918,
+      "step": 4096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3280432818564814e-06,
+      "loss": 0.3335,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324064003458939e-06,
+      "loss": 0.8384,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320087681622064e-06,
+      "loss": 0.8233,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.316114317877425e-06,
+      "loss": 0.7714,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3121439137554524e-06,
+      "loss": 0.7741,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308176470785438e-06,
+      "loss": 0.8092,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.304211990495533e-06,
+      "loss": 0.7599,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3002504744127443e-06,
+      "loss": 0.7732,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296291924062941e-06,
+      "loss": 0.7638,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.29233634097085e-06,
+      "loss": 0.7252,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2883837266600517e-06,
+      "loss": 0.7962,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284434082652985e-06,
+      "loss": 0.8101,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2804874104709464e-06,
+      "loss": 0.8685,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2765437116340838e-06,
+      "loss": 0.76,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2726029876614044e-06,
+      "loss": 0.8282,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2686652400707663e-06,
+      "loss": 0.8158,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2647304703788832e-06,
+      "loss": 0.8025,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260798680101316e-06,
+      "loss": 0.7987,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2568698707524928e-06,
+      "loss": 0.8164,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.252944043845674e-06,
+      "loss": 0.7372,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249021200892982e-06,
+      "loss": 0.8053,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.245101343405397e-06,
+      "loss": 0.7318,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.241184472892731e-06,
+      "loss": 0.8453,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2372705908636572e-06,
+      "loss": 0.7949,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2333596988257e-06,
+      "loss": 0.7951,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2294517982852293e-06,
+      "loss": 0.8173,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225546890747453e-06,
+      "loss": 0.7105,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.221644977716443e-06,
+      "loss": 0.7872,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2177460606951084e-06,
+      "loss": 0.7253,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2138501411851986e-06,
+      "loss": 0.7617,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209957220687323e-06,
+      "loss": 0.813,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206067300700926e-06,
+      "loss": 0.7977,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.202180382724299e-06,
+      "loss": 0.7548,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1982964682545716e-06,
+      "loss": 0.7154,
+      "step": 4130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194415558787727e-06,
+      "loss": 0.8409,
+      "step": 4131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1905376558185866e-06,
+      "loss": 0.7705,
+      "step": 4132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186662760840803e-06,
+      "loss": 0.7693,
+      "step": 4133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1827908753468898e-06,
+      "loss": 0.7622,
+      "step": 4134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1789220008281888e-06,
+      "loss": 0.8078,
+      "step": 4135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1750561387748823e-06,
+      "loss": 0.8083,
+      "step": 4136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1711932906759967e-06,
+      "loss": 0.7979,
+      "step": 4137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1673334580193962e-06,
+      "loss": 0.735,
+      "step": 4138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1634766422917807e-06,
+      "loss": 0.7725,
+      "step": 4139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1596228449786914e-06,
+      "loss": 0.748,
+      "step": 4140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1557720675645044e-06,
+      "loss": 0.8137,
+      "step": 4141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1519243115324363e-06,
+      "loss": 0.7738,
+      "step": 4142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1480795783645347e-06,
+      "loss": 0.333,
+      "step": 4143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1442378695416875e-06,
+      "loss": 0.7858,
+      "step": 4144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1403991865436123e-06,
+      "loss": 0.766,
+      "step": 4145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136563530848874e-06,
+      "loss": 0.8049,
+      "step": 4146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1327309039348532e-06,
+      "loss": 0.7867,
+      "step": 4147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1289013072777754e-06,
+      "loss": 0.8125,
+      "step": 4148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1250747423526976e-06,
+      "loss": 0.7754,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.121251210633508e-06,
+      "loss": 0.8057,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.117430713592925e-06,
+      "loss": 0.8266,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113613252702499e-06,
+      "loss": 0.7644,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109798829432618e-06,
+      "loss": 0.744,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.105987445252485e-06,
+      "loss": 0.8062,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102179101630143e-06,
+      "loss": 0.8119,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0983738000324692e-06,
+      "loss": 0.7679,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.094571541925152e-06,
+      "loss": 0.7459,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09077232877272e-06,
+      "loss": 0.3456,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0869761620385296e-06,
+      "loss": 0.8164,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.083183043184761e-06,
+      "loss": 0.746,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0793929736724118e-06,
+      "loss": 0.8136,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0756059549613217e-06,
+      "loss": 0.7226,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0718219885101454e-06,
+      "loss": 0.7716,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.068041075776357e-06,
+      "loss": 0.7631,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064263218216267e-06,
+      "loss": 0.7762,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0604884172850014e-06,
+      "loss": 0.3314,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05671667443651e-06,
+      "loss": 0.7615,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052947991123565e-06,
+      "loss": 0.7737,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.049182368797761e-06,
+      "loss": 0.782,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04541980890951e-06,
+      "loss": 0.7697,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04166031290805e-06,
+      "loss": 0.8005,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0379038822414365e-06,
+      "loss": 0.722,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0341505183565414e-06,
+      "loss": 0.753,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0304002226990582e-06,
+      "loss": 0.8274,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0266529967134997e-06,
+      "loss": 0.7802,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022908841843194e-06,
+      "loss": 0.6974,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0191677595302885e-06,
+      "loss": 0.3396,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0154297512157438e-06,
+      "loss": 0.832,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0116948183393394e-06,
+      "loss": 0.759,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.007962962339669e-06,
+      "loss": 0.7143,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0042341846541425e-06,
+      "loss": 0.8452,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0005084867189817e-06,
+      "loss": 0.7916,
+      "step": 4182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.996785869969222e-06,
+      "loss": 0.6552,
+      "step": 4183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9930663358387214e-06,
+      "loss": 0.7487,
+      "step": 4184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9893498857601347e-06,
+      "loss": 0.7688,
+      "step": 4185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9856365211649363e-06,
+      "loss": 0.759,
+      "step": 4186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981926243483422e-06,
+      "loss": 0.7673,
+      "step": 4187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978219054144679e-06,
+      "loss": 0.7472,
+      "step": 4188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9745149545766174e-06,
+      "loss": 0.7856,
+      "step": 4189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9708139462059605e-06,
+      "loss": 0.8016,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967116030458234e-06,
+      "loss": 0.7412,
+      "step": 4191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9634212087577662e-06,
+      "loss": 0.7262,
+      "step": 4192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95972948252771e-06,
+      "loss": 0.8336,
+      "step": 4193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9560408531900177e-06,
+      "loss": 0.7398,
+      "step": 4194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95235532216544e-06,
+      "loss": 0.8285,
+      "step": 4195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948672890873551e-06,
+      "loss": 0.8222,
+      "step": 4196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94499356073272e-06,
+      "loss": 0.7589,
+      "step": 4197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941317333160122e-06,
+      "loss": 0.8136,
+      "step": 4198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937644209571743e-06,
+      "loss": 0.8038,
+      "step": 4199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9339741913823673e-06,
+      "loss": 0.8299,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9303072800055867e-06,
+      "loss": 0.7236,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9266434768537957e-06,
+      "loss": 0.6927,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9229827833381885e-06,
+      "loss": 0.7393,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919325200868767e-06,
+      "loss": 0.8269,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.915670730854331e-06,
+      "loss": 0.7796,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912019374702482e-06,
+      "loss": 0.8114,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908371133819622e-06,
+      "loss": 0.76,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9047260096109555e-06,
+      "loss": 0.8031,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9010840034804822e-06,
+      "loss": 0.7731,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.897445116831006e-06,
+      "loss": 0.7901,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.893809351064123e-06,
+      "loss": 0.348,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8901767075802369e-06,
+      "loss": 0.7808,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8865471877785368e-06,
+      "loss": 0.782,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8829207930570137e-06,
+      "loss": 0.3368,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8792975248124645e-06,
+      "loss": 0.313,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8756773844404641e-06,
+      "loss": 0.6915,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8720603733353916e-06,
+      "loss": 0.7837,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8684464928904257e-06,
+      "loss": 0.3136,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.864835744497535e-06,
+      "loss": 0.7331,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8612281295474733e-06,
+      "loss": 0.8017,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.857623649429804e-06,
+      "loss": 0.7833,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.854022305532872e-06,
+      "loss": 0.7902,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8504240992438093e-06,
+      "loss": 0.8497,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846829031948556e-06,
+      "loss": 0.3247,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8432371050318332e-06,
+      "loss": 0.3599,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.839648319877143e-06,
+      "loss": 0.7771,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8360626778667977e-06,
+      "loss": 0.737,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8324801803818847e-06,
+      "loss": 0.8365,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8289008288022846e-06,
+      "loss": 0.7776,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8253246245066658e-06,
+      "loss": 0.8534,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.821751568872484e-06,
+      "loss": 0.7884,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8181816632759819e-06,
+      "loss": 0.808,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8146149090921906e-06,
+      "loss": 0.8466,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8110513076949266e-06,
+      "loss": 0.7369,
+      "step": 4234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8074908604567908e-06,
+      "loss": 0.8062,
+      "step": 4235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80393356874917e-06,
+      "loss": 0.7402,
+      "step": 4236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8003794339422365e-06,
+      "loss": 0.7911,
+      "step": 4237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7968284574049456e-06,
+      "loss": 0.7342,
+      "step": 4238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.793280640505035e-06,
+      "loss": 0.7987,
+      "step": 4239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7897359846090301e-06,
+      "loss": 0.8273,
+      "step": 4240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7861944910822316e-06,
+      "loss": 0.7934,
+      "step": 4241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782656161288726e-06,
+      "loss": 0.8063,
+      "step": 4242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.779120996591388e-06,
+      "loss": 0.3133,
+      "step": 4243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7755889983518582e-06,
+      "loss": 0.8242,
+      "step": 4244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.772060167930566e-06,
+      "loss": 0.7689,
+      "step": 4245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7685345066867287e-06,
+      "loss": 0.7387,
+      "step": 4246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7650120159783268e-06,
+      "loss": 0.7263,
+      "step": 4247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761492697162127e-06,
+      "loss": 0.8191,
+      "step": 4248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7579765515936808e-06,
+      "loss": 0.7201,
+      "step": 4249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7544635806273091e-06,
+      "loss": 0.8046,
+      "step": 4250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7509537856161074e-06,
+      "loss": 0.7587,
+      "step": 4251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7474471679119575e-06,
+      "loss": 0.798,
+      "step": 4252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7439437288655147e-06,
+      "loss": 0.8389,
+      "step": 4253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7404434698262007e-06,
+      "loss": 0.7959,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736946392142226e-06,
+      "loss": 0.6858,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.733452497160566e-06,
+      "loss": 0.7333,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7299617862269759e-06,
+      "loss": 0.81,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7264742606859797e-06,
+      "loss": 0.7783,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7229899218808777e-06,
+      "loss": 0.8017,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7195087711537428e-06,
+      "loss": 0.7949,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.716030809845418e-06,
+      "loss": 0.7587,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7125560392955199e-06,
+      "loss": 0.7885,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709084460842435e-06,
+      "loss": 0.7734,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056160758233188e-06,
+      "loss": 0.7832,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021508855740999e-06,
+      "loss": 0.7401,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6986888914294753e-06,
+      "loss": 0.7608,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6952300947229083e-06,
+      "loss": 0.8317,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6917744967866355e-06,
+      "loss": 0.8375,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6883220989516569e-06,
+      "loss": 0.7936,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6848729025477418e-06,
+      "loss": 0.722,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6814269089034274e-06,
+      "loss": 0.7536,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6779841193460144e-06,
+      "loss": 0.8041,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.674544535201571e-06,
+      "loss": 0.3476,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6711081577949373e-06,
+      "loss": 0.778,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6676749884497044e-06,
+      "loss": 0.7703,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.664245028488236e-06,
+      "loss": 0.7914,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6608182792316652e-06,
+      "loss": 0.7851,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6573947419998748e-06,
+      "loss": 0.7969,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6539744181115192e-06,
+      "loss": 0.8474,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6505573088840166e-06,
+      "loss": 0.7736,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6471434156335452e-06,
+      "loss": 0.7762,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6437327396750358e-06,
+      "loss": 0.3188,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6403252823221948e-06,
+      "loss": 0.8065,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6369210448874818e-06,
+      "loss": 0.7924,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6335200286821073e-06,
+      "loss": 0.737,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6301222350160595e-06,
+      "loss": 0.8399,
+      "step": 4286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6267276651980711e-06,
+      "loss": 0.7781,
+      "step": 4287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6233363205356402e-06,
+      "loss": 0.8411,
+      "step": 4288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6199482023350167e-06,
+      "loss": 0.7327,
+      "step": 4289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6165633119012125e-06,
+      "loss": 0.7832,
+      "step": 4290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6131816505379949e-06,
+      "loss": 0.8036,
+      "step": 4291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6098032195478873e-06,
+      "loss": 0.8418,
+      "step": 4292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6064280202321702e-06,
+      "loss": 0.7581,
+      "step": 4293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6030560538908757e-06,
+      "loss": 0.7399,
+      "step": 4294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5996873218227949e-06,
+      "loss": 0.7794,
+      "step": 4295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5963218253254709e-06,
+      "loss": 0.7912,
+      "step": 4296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5929595656952012e-06,
+      "loss": 0.8574,
+      "step": 4297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5896005442270345e-06,
+      "loss": 0.7712,
+      "step": 4298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.586244762214776e-06,
+      "loss": 0.7468,
+      "step": 4299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5828922209509811e-06,
+      "loss": 0.7259,
+      "step": 4300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5795429217269553e-06,
+      "loss": 0.7039,
+      "step": 4301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5761968658327588e-06,
+      "loss": 0.7743,
+      "step": 4302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5728540545572003e-06,
+      "loss": 0.774,
+      "step": 4303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5695144891878367e-06,
+      "loss": 0.7883,
+      "step": 4304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5661781710109846e-06,
+      "loss": 0.7711,
+      "step": 4305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5628451013116951e-06,
+      "loss": 0.76,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5595152813737756e-06,
+      "loss": 0.7919,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5561887124797903e-06,
+      "loss": 0.3382,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.552865395911033e-06,
+      "loss": 0.7303,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5495453329475563e-06,
+      "loss": 0.7492,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5462285248681607e-06,
+      "loss": 0.8353,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5429149729503913e-06,
+      "loss": 0.8074,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539604678470531e-06,
+      "loss": 0.79,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5362976427036202e-06,
+      "loss": 0.7753,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5329938669234401e-06,
+      "loss": 0.7801,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.529693352402507e-06,
+      "loss": 0.7665,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5263961004120952e-06,
+      "loss": 0.8465,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5231021122222156e-06,
+      "loss": 0.7824,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5198113891016208e-06,
+      "loss": 0.7386,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5165239323178082e-06,
+      "loss": 0.3621,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513239743137015e-06,
+      "loss": 0.3362,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5099588228242235e-06,
+      "loss": 0.7223,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.506681172643152e-06,
+      "loss": 0.7018,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5034067938562625e-06,
+      "loss": 0.7598,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5001356877247563e-06,
+      "loss": 0.7691,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4968678555085747e-06,
+      "loss": 0.8127,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4936032984663974e-06,
+      "loss": 0.7535,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4903420178556405e-06,
+      "loss": 0.7675,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4870840149324617e-06,
+      "loss": 0.7725,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4838292909517548e-06,
+      "loss": 0.8286,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4805778471671506e-06,
+      "loss": 0.7525,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4773296848310138e-06,
+      "loss": 0.8115,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4740848051944534e-06,
+      "loss": 0.8153,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4708432095073021e-06,
+      "loss": 0.7369,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4676048990181346e-06,
+      "loss": 0.7433,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4643698749742663e-06,
+      "loss": 0.76,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4611381386217327e-06,
+      "loss": 0.7526,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4579096912053114e-06,
+      "loss": 0.8094,
+      "step": 4338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4546845339685146e-06,
+      "loss": 0.762,
+      "step": 4339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451462668153586e-06,
+      "loss": 0.7706,
+      "step": 4340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448244095001493e-06,
+      "loss": 0.8431,
+      "step": 4341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4450288157519499e-06,
+      "loss": 0.8177,
+      "step": 4342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4418168316433923e-06,
+      "loss": 0.7714,
+      "step": 4343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.438608143912984e-06,
+      "loss": 0.6799,
+      "step": 4344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4354027537966285e-06,
+      "loss": 0.7425,
+      "step": 4345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4322006625289563e-06,
+      "loss": 0.7464,
+      "step": 4346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4290018713433174e-06,
+      "loss": 0.7693,
+      "step": 4347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.425806381471806e-06,
+      "loss": 0.7567,
+      "step": 4348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4226141941452354e-06,
+      "loss": 0.7366,
+      "step": 4349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4194253105931478e-06,
+      "loss": 0.8299,
+      "step": 4350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416239732043815e-06,
+      "loss": 0.8251,
+      "step": 4351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4130574597242342e-06,
+      "loss": 0.7702,
+      "step": 4352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4098784948601308e-06,
+      "loss": 0.7912,
+      "step": 4353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.406702838675953e-06,
+      "loss": 0.7144,
+      "step": 4354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.403530492394879e-06,
+      "loss": 0.839,
+      "step": 4355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400361457238808e-06,
+      "loss": 0.7558,
+      "step": 4356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3971957344283671e-06,
+      "loss": 0.7528,
+      "step": 4357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3940333251829053e-06,
+      "loss": 0.8269,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390874230720497e-06,
+      "loss": 0.751,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.387718452257938e-06,
+      "loss": 0.7677,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3845659910107478e-06,
+      "loss": 0.8127,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.38141684819317e-06,
+      "loss": 0.7958,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.378271025018164e-06,
+      "loss": 0.7959,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.375128522697423e-06,
+      "loss": 0.755,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3719893424413455e-06,
+      "loss": 0.7653,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3688534854590585e-06,
+      "loss": 0.3316,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3657209529584147e-06,
+      "loss": 0.8068,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3625917461459726e-06,
+      "loss": 0.8015,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3594658662270187e-06,
+      "loss": 0.7996,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3563433144055594e-06,
+      "loss": 0.8109,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3532240918843175e-06,
+      "loss": 0.7401,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3501081998647236e-06,
+      "loss": 0.7851,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.346995639546943e-06,
+      "loss": 0.7554,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3438864121298489e-06,
+      "loss": 0.7877,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3407805188110224e-06,
+      "loss": 0.7964,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3376779607867752e-06,
+      "loss": 0.8603,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.334578739252127e-06,
+      "loss": 0.7526,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.331482855400813e-06,
+      "loss": 0.7546,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3283903104252815e-06,
+      "loss": 0.8084,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3253011055166987e-06,
+      "loss": 0.8251,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3222152418649404e-06,
+      "loss": 0.7954,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3191327206585968e-06,
+      "loss": 0.7364,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3160535430849709e-06,
+      "loss": 0.7504,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3129777103300777e-06,
+      "loss": 0.79,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3099052235786435e-06,
+      "loss": 0.8042,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3068360840141081e-06,
+      "loss": 0.7145,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303770292818618e-06,
+      "loss": 0.7901,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.300707851173033e-06,
+      "loss": 0.7627,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2976487602569222e-06,
+      "loss": 0.7436,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2945930212485625e-06,
+      "loss": 0.8211,
+      "step": 4390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2915406353249428e-06,
+      "loss": 0.7602,
+      "step": 4391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2884916036617567e-06,
+      "loss": 0.7727,
+      "step": 4392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2854459274334097e-06,
+      "loss": 0.745,
+      "step": 4393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2824036078130098e-06,
+      "loss": 0.7827,
+      "step": 4394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279364645972383e-06,
+      "loss": 0.7602,
+      "step": 4395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2763290430820453e-06,
+      "loss": 0.7802,
+      "step": 4396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.27329680031123e-06,
+      "loss": 0.7048,
+      "step": 4397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2702679188278811e-06,
+      "loss": 0.7661,
+      "step": 4398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2672423997986327e-06,
+      "loss": 0.8219,
+      "step": 4399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2642202443888329e-06,
+      "loss": 0.8044,
+      "step": 4400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2612014537625371e-06,
+      "loss": 0.7132,
+      "step": 4401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2581860290825022e-06,
+      "loss": 0.7641,
+      "step": 4402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2551739715101785e-06,
+      "loss": 0.8024,
+      "step": 4403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2521652822057372e-06,
+      "loss": 0.7743,
+      "step": 4404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2491599623280426e-06,
+      "loss": 0.7918,
+      "step": 4405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.246158013034654e-06,
+      "loss": 0.7382,
+      "step": 4406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2431594354818478e-06,
+      "loss": 0.7962,
+      "step": 4407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2401642308245909e-06,
+      "loss": 0.7253,
+      "step": 4408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2371724002165563e-06,
+      "loss": 0.7373,
+      "step": 4409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2341839448101122e-06,
+      "loss": 0.741,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2311988657563335e-06,
+      "loss": 0.7887,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2282171642049879e-06,
+      "loss": 0.8055,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2252388413045468e-06,
+      "loss": 0.774,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2222638982021794e-06,
+      "loss": 0.7683,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2192923360437504e-06,
+      "loss": 0.7333,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216324155973827e-06,
+      "loss": 0.7805,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2133593591356686e-06,
+      "loss": 0.745,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2103979466712345e-06,
+      "loss": 0.7518,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2074399197211805e-06,
+      "loss": 0.7524,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2044852794248575e-06,
+      "loss": 0.7761,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2015340269203113e-06,
+      "loss": 0.7398,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.198586163344282e-06,
+      "loss": 0.8166,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1956416898322132e-06,
+      "loss": 0.7766,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.192700607518229e-06,
+      "loss": 0.794,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1897629175351532e-06,
+      "loss": 0.779,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1868286210145096e-06,
+      "loss": 0.7829,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1838977190865042e-06,
+      "loss": 0.7882,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1809702128800392e-06,
+      "loss": 0.3461,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1780461035227164e-06,
+      "loss": 0.3408,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1751253921408169e-06,
+      "loss": 0.8018,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1722080798593183e-06,
+      "loss": 0.822,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1692941678018942e-06,
+      "loss": 0.7724,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1663836570909037e-06,
+      "loss": 0.7525,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1634765488473887e-06,
+      "loss": 0.7619,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1605728441910956e-06,
+      "loss": 0.7445,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1576725442404514e-06,
+      "loss": 0.7711,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1547756501125661e-06,
+      "loss": 0.7643,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1518821629232503e-06,
+      "loss": 0.8022,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148992083786995e-06,
+      "loss": 0.7905,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.146105413816977e-06,
+      "loss": 0.8111,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1432221541250655e-06,
+      "loss": 0.7988,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.140342305821812e-06,
+      "loss": 0.778,
+      "step": 4442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1374658700164553e-06,
+      "loss": 0.7598,
+      "step": 4443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.134592847816921e-06,
+      "loss": 0.8009,
+      "step": 4444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1317232403298172e-06,
+      "loss": 0.7645,
+      "step": 4445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.128857048660439e-06,
+      "loss": 0.7807,
+      "step": 4446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1259942739127637e-06,
+      "loss": 0.7601,
+      "step": 4447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1231349171894557e-06,
+      "loss": 0.7372,
+      "step": 4448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.120278979591861e-06,
+      "loss": 0.8205,
+      "step": 4449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.117426462220007e-06,
+      "loss": 0.7491,
+      "step": 4450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1145773661726066e-06,
+      "loss": 0.7821,
+      "step": 4451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1117316925470544e-06,
+      "loss": 0.7712,
+      "step": 4452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1088894424394226e-06,
+      "loss": 0.8182,
+      "step": 4453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.106050616944475e-06,
+      "loss": 0.7781,
+      "step": 4454
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1032152171556433e-06,
+      "loss": 0.7811,
+      "step": 4455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1003832441650453e-06,
+      "loss": 0.8119,
+      "step": 4456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0975546990634866e-06,
+      "loss": 0.7881,
+      "step": 4457
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0947295829404404e-06,
+      "loss": 0.3246,
+      "step": 4458
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0919078968840613e-06,
+      "loss": 0.8186,
+      "step": 4459
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0890896419811914e-06,
+      "loss": 0.7102,
+      "step": 4460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0862748193173467e-06,
+      "loss": 0.328,
+      "step": 4461
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083463429976711e-06,
+      "loss": 0.3503,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0806554750421628e-06,
+      "loss": 0.6829,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0778509555952488e-06,
+      "loss": 0.7515,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0750498727161862e-06,
+      "loss": 0.7996,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0722522274838842e-06,
+      "loss": 0.3343,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0694580209759165e-06,
+      "loss": 0.8002,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0666672542685307e-06,
+      "loss": 0.8081,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0638799284366596e-06,
+      "loss": 0.7913,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061096044553902e-06,
+      "loss": 0.8257,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0583156036925346e-06,
+      "loss": 0.715,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0555386069235075e-06,
+      "loss": 0.8179,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0527650553164425e-06,
+      "loss": 0.786,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.04999494993964e-06,
+      "loss": 0.7618,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0472282918600617e-06,
+      "loss": 0.762,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0444650821433532e-06,
+      "loss": 0.7978,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0417053218538287e-06,
+      "loss": 0.7556,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0389490120544687e-06,
+      "loss": 0.8034,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0361961538069309e-06,
+      "loss": 0.7489,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0334467481715404e-06,
+      "loss": 0.7574,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0307007962072923e-06,
+      "loss": 0.7939,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.027958298971854e-06,
+      "loss": 0.7522,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.025219257521558e-06,
+      "loss": 0.7814,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0224836729114085e-06,
+      "loss": 0.7275,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0197515461950824e-06,
+      "loss": 0.7392,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0170228784249149e-06,
+      "loss": 0.8125,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0142976706519136e-06,
+      "loss": 0.7105,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0115759239257606e-06,
+      "loss": 0.7459,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0088576392947924e-06,
+      "loss": 0.7421,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0061428178060173e-06,
+      "loss": 0.8116,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0034314605051143e-06,
+      "loss": 0.691,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0007235684364248e-06,
+      "loss": 0.8135,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.980191426429486e-07,
+      "loss": 0.7254,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.95318184166363e-07,
+      "loss": 0.8077,
+      "step": 4494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.926206940470018e-07,
+      "loss": 0.7703,
+      "step": 4495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.89926673323861e-07,
+      "loss": 0.7168,
+      "step": 4496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.87236123034605e-07,
+      "loss": 0.778,
+      "step": 4497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.845490442155626e-07,
+      "loss": 0.8058,
+      "step": 4498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.818654379017245e-07,
+      "loss": 0.7854,
+      "step": 4499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.791853051267352e-07,
+      "loss": 0.8292,
+      "step": 4500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.765086469229158e-07,
+      "loss": 0.7361,
+      "step": 4501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.73835464321241e-07,
+      "loss": 0.8037,
+      "step": 4502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.711657583513412e-07,
+      "loss": 0.7879,
+      "step": 4503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.684995300415212e-07,
+      "loss": 0.7731,
+      "step": 4504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.658367804187385e-07,
+      "loss": 0.8253,
+      "step": 4505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.631775105086061e-07,
+      "loss": 0.7898,
+      "step": 4506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.60521721335408e-07,
+      "loss": 0.7455,
+      "step": 4507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.57869413922079e-07,
+      "loss": 0.8127,
+      "step": 4508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.552205892902155e-07,
+      "loss": 0.7747,
+      "step": 4509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.525752484600737e-07,
+      "loss": 0.7606,
+      "step": 4510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.499333924505638e-07,
+      "loss": 0.3363,
+      "step": 4511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.472950222792588e-07,
+      "loss": 0.7256,
+      "step": 4512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.446601389623866e-07,
+      "loss": 0.7705,
+      "step": 4513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420287435148301e-07,
+      "loss": 0.8177,
+      "step": 4514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.394008369501317e-07,
+      "loss": 0.7736,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.367764202804929e-07,
+      "loss": 0.7305,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.341554945167631e-07,
+      "loss": 0.7461,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.315380606684521e-07,
+      "loss": 0.7572,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.289241197437238e-07,
+      "loss": 0.7638,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.263136727493982e-07,
+      "loss": 0.7973,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.237067206909456e-07,
+      "loss": 0.7281,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.211032645724927e-07,
+      "loss": 0.7907,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.185033053968239e-07,
+      "loss": 0.8168,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.159068441653685e-07,
+      "loss": 0.7539,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.133138818782117e-07,
+      "loss": 0.7321,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.107244195340959e-07,
+      "loss": 0.7984,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.081384581304076e-07,
+      "loss": 0.8456,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.055559986631878e-07,
+      "loss": 0.783,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.029770421271323e-07,
+      "loss": 0.7238,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.004015895155838e-07,
+      "loss": 0.7933,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.978296418205335e-07,
+      "loss": 0.8145,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.952612000326289e-07,
+      "loss": 0.7628,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.926962651411641e-07,
+      "loss": 0.8059,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.901348381340758e-07,
+      "loss": 0.8345,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.875769199979623e-07,
+      "loss": 0.832,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.850225117180599e-07,
+      "loss": 0.7768,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824716142782597e-07,
+      "loss": 0.7941,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.799242286610943e-07,
+      "loss": 0.8093,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.7738035584775e-07,
+      "loss": 0.8224,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.748399968180554e-07,
+      "loss": 0.3656,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.723031525504877e-07,
+      "loss": 0.7862,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.697698240221697e-07,
+      "loss": 0.773,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.672400122088709e-07,
+      "loss": 0.815,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.647137180850063e-07,
+      "loss": 0.7875,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.62190942623633e-07,
+      "loss": 0.7779,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.596716867964561e-07,
+      "loss": 0.721,
+      "step": 4546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.571559515738248e-07,
+      "loss": 0.3159,
+      "step": 4547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.546437379247319e-07,
+      "loss": 0.7797,
+      "step": 4548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.521350468168121e-07,
+      "loss": 0.7277,
+      "step": 4549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.496298792163449e-07,
+      "loss": 0.8295,
+      "step": 4550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.471282360882526e-07,
+      "loss": 0.7599,
+      "step": 4551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.446301183961004e-07,
+      "loss": 0.8108,
+      "step": 4552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.421355271020915e-07,
+      "loss": 0.3248,
+      "step": 4553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.396444631670809e-07,
+      "loss": 0.8332,
+      "step": 4554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.371569275505531e-07,
+      "loss": 0.8091,
+      "step": 4555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.346729212106386e-07,
+      "loss": 0.7678,
+      "step": 4556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.32192445104113e-07,
+      "loss": 0.7676,
+      "step": 4557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.297155001863832e-07,
+      "loss": 0.7563,
+      "step": 4558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.272420874115017e-07,
+      "loss": 0.8248,
+      "step": 4559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.247722077321618e-07,
+      "loss": 0.7575,
+      "step": 4560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.223058620996938e-07,
+      "loss": 0.7951,
+      "step": 4561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198430514640621e-07,
+      "loss": 0.7096,
+      "step": 4562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.173837767738779e-07,
+      "loss": 0.7193,
+      "step": 4563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.149280389763881e-07,
+      "loss": 0.7839,
+      "step": 4564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124758390174692e-07,
+      "loss": 0.7941,
+      "step": 4565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.100271778416469e-07,
+      "loss": 0.7317,
+      "step": 4566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.075820563920777e-07,
+      "loss": 0.7541,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.051404756105552e-07,
+      "loss": 0.7806,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.027024364375091e-07,
+      "loss": 0.7539,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.002679398120062e-07,
+      "loss": 0.7514,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.978369866717472e-07,
+      "loss": 0.7679,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.95409577953069e-07,
+      "loss": 0.7794,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.929857145909437e-07,
+      "loss": 0.7851,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.905653975189764e-07,
+      "loss": 0.7998,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.881486276694073e-07,
+      "loss": 0.7708,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.857354059731104e-07,
+      "loss": 0.8202,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.833257333595923e-07,
+      "loss": 0.8004,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.809196107569927e-07,
+      "loss": 0.7702,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.785170390920859e-07,
+      "loss": 0.3401,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.761180192902751e-07,
+      "loss": 0.7829,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.737225522755954e-07,
+      "loss": 0.7892,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.713306389707209e-07,
+      "loss": 0.7489,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.689422802969459e-07,
+      "loss": 0.8401,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.665574771742024e-07,
+      "loss": 0.7873,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.641762305210532e-07,
+      "loss": 0.7738,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.61798541254688e-07,
+      "loss": 0.7714,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.594244102909243e-07,
+      "loss": 0.7716,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.570538385442195e-07,
+      "loss": 0.3261,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.546868269276475e-07,
+      "loss": 0.7877,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.52323376352917e-07,
+      "loss": 0.8031,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.499634877303674e-07,
+      "loss": 0.7444,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.476071619689617e-07,
+      "loss": 0.7377,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.452543999762907e-07,
+      "loss": 0.8574,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429052026585759e-07,
+      "loss": 0.7568,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.405595709206648e-07,
+      "loss": 0.8333,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.382175056660257e-07,
+      "loss": 0.8037,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.358790077967637e-07,
+      "loss": 0.7974,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.335440782136027e-07,
+      "loss": 0.7824,
+      "step": 4598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.312127178158912e-07,
+      "loss": 0.8195,
+      "step": 4599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.288849275016085e-07,
+      "loss": 0.8065,
+      "step": 4600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.265607081673543e-07,
+      "loss": 0.8272,
+      "step": 4601
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.242400607083555e-07,
+      "loss": 0.828,
+      "step": 4602
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.219229860184606e-07,
+      "loss": 0.7774,
+      "step": 4603
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.196094849901436e-07,
+      "loss": 0.7938,
+      "step": 4604
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.172995585145026e-07,
+      "loss": 0.8123,
+      "step": 4605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.149932074812582e-07,
+      "loss": 0.7949,
+      "step": 4606
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.12690432778752e-07,
+      "loss": 0.7539,
+      "step": 4607
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103912352939524e-07,
+      "loss": 0.8203,
+      "step": 4608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080956159124464e-07,
+      "loss": 0.7453,
+      "step": 4609
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.058035755184422e-07,
+      "loss": 0.7667,
+      "step": 4610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.035151149947739e-07,
+      "loss": 0.7742,
+      "step": 4611
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.012302352228906e-07,
+      "loss": 0.8338,
+      "step": 4612
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.989489370828706e-07,
+      "loss": 0.7984,
+      "step": 4613
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.966712214534022e-07,
+      "loss": 0.7453,
+      "step": 4614
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.943970892118002e-07,
+      "loss": 0.8068,
+      "step": 4615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.921265412340028e-07,
+      "loss": 0.7966,
+      "step": 4616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.898595783945572e-07,
+      "loss": 0.7408,
+      "step": 4617
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.875962015666371e-07,
+      "loss": 0.8086,
+      "step": 4618
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.853364116220351e-07,
+      "loss": 0.7907,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.830802094311606e-07,
+      "loss": 0.8313,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.808275958630383e-07,
+      "loss": 0.7732,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.785785717853155e-07,
+      "loss": 0.8196,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.763331380642557e-07,
+      "loss": 0.7934,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.740912955647338e-07,
+      "loss": 0.8149,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.718530451502514e-07,
+      "loss": 0.7916,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.696183876829199e-07,
+      "loss": 0.7984,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.673873240234674e-07,
+      "loss": 0.7775,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.651598550312399e-07,
+      "loss": 0.7845,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.629359815641956e-07,
+      "loss": 0.824,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.607157044789115e-07,
+      "loss": 0.7654,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.584990246305767e-07,
+      "loss": 0.6762,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.562859428729962e-07,
+      "loss": 0.8119,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.540764600585881e-07,
+      "loss": 0.7914,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.518705770383848e-07,
+      "loss": 0.7831,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.496682946620326e-07,
+      "loss": 0.7636,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.474696137777903e-07,
+      "loss": 0.748,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.452745352325295e-07,
+      "loss": 0.7241,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.430830598717341e-07,
+      "loss": 0.8187,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.408951885395021e-07,
+      "loss": 0.7248,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.387109220785393e-07,
+      "loss": 0.7921,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.365302613301671e-07,
+      "loss": 0.8138,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.343532071343161e-07,
+      "loss": 0.729,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.321797603295255e-07,
+      "loss": 0.7202,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.300099217529532e-07,
+      "loss": 0.7767,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.278436922403575e-07,
+      "loss": 0.7448,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.256810726261098e-07,
+      "loss": 0.7724,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.235220637431971e-07,
+      "loss": 0.7628,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.213666664232054e-07,
+      "loss": 0.7263,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.192148814963361e-07,
+      "loss": 0.7588,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.170667097913996e-07,
+      "loss": 0.7955,
+      "step": 4650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.149221521358139e-07,
+      "loss": 0.7408,
+      "step": 4651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.127812093555985e-07,
+      "loss": 0.811,
+      "step": 4652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.106438822753924e-07,
+      "loss": 0.7787,
+      "step": 4653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.085101717184327e-07,
+      "loss": 0.7646,
+      "step": 4654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.063800785065643e-07,
+      "loss": 0.8229,
+      "step": 4655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.042536034602442e-07,
+      "loss": 0.8184,
+      "step": 4656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.021307473985306e-07,
+      "loss": 0.7444,
+      "step": 4657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.000115111390902e-07,
+      "loss": 0.7932,
+      "step": 4658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.978958954981962e-07,
+      "loss": 0.7357,
+      "step": 4659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.957839012907229e-07,
+      "loss": 0.8053,
+      "step": 4660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.936755293301555e-07,
+      "loss": 0.7652,
+      "step": 4661
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.915707804285797e-07,
+      "loss": 0.7798,
+      "step": 4662
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.894696553966872e-07,
+      "loss": 0.7325,
+      "step": 4663
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.87372155043775e-07,
+      "loss": 0.834,
+      "step": 4664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.852782801777424e-07,
+      "loss": 0.7838,
+      "step": 4665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.831880316050919e-07,
+      "loss": 0.3324,
+      "step": 4666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.811014101309309e-07,
+      "loss": 0.8248,
+      "step": 4667
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.790184165589696e-07,
+      "loss": 0.7533,
+      "step": 4668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.769390516915186e-07,
+      "loss": 0.7737,
+      "step": 4669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.748633163294937e-07,
+      "loss": 0.8145,
+      "step": 4670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.727912112724099e-07,
+      "loss": 0.809,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.707227373183865e-07,
+      "loss": 0.8112,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.686578952641442e-07,
+      "loss": 0.7972,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.665966859049998e-07,
+      "loss": 0.7529,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.645391100348818e-07,
+      "loss": 0.7469,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.62485168446305e-07,
+      "loss": 0.7721,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.604348619303934e-07,
+      "loss": 0.7543,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.583881912768729e-07,
+      "loss": 0.7734,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.563451572740608e-07,
+      "loss": 0.7373,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.543057607088775e-07,
+      "loss": 0.8205,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.52270002366847e-07,
+      "loss": 0.8095,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.502378830320865e-07,
+      "loss": 0.3351,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.482094034873087e-07,
+      "loss": 0.7422,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.46184564513832e-07,
+      "loss": 0.8445,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.441633668915703e-07,
+      "loss": 0.8318,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.421458113990296e-07,
+      "loss": 0.3074,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.401318988133198e-07,
+      "loss": 0.8158,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.381216299101455e-07,
+      "loss": 0.823,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.361150054638053e-07,
+      "loss": 0.7926,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.341120262471977e-07,
+      "loss": 0.7652,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.321126930318144e-07,
+      "loss": 0.778,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.301170065877448e-07,
+      "loss": 0.3257,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28124967683672e-07,
+      "loss": 0.3052,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.261365770868753e-07,
+      "loss": 0.7736,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24151835563228e-07,
+      "loss": 0.7748,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.221707438771983e-07,
+      "loss": 0.7712,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.201933027918493e-07,
+      "loss": 0.7452,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.182195130688372e-07,
+      "loss": 0.765,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.162493754684116e-07,
+      "loss": 0.7569,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.142828907494146e-07,
+      "loss": 0.7371,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.123200596692835e-07,
+      "loss": 0.8192,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.103608829840456e-07,
+      "loss": 0.754,
+      "step": 4702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.084053614483253e-07,
+      "loss": 0.7542,
+      "step": 4703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.064534958153344e-07,
+      "loss": 0.7752,
+      "step": 4704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.045052868368749e-07,
+      "loss": 0.803,
+      "step": 4705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.025607352633499e-07,
+      "loss": 0.7833,
+      "step": 4706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.006198418437436e-07,
+      "loss": 0.7908,
+      "step": 4707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.986826073256334e-07,
+      "loss": 0.7058,
+      "step": 4708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.967490324551949e-07,
+      "loss": 0.7702,
+      "step": 4709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.948191179771833e-07,
+      "loss": 0.7189,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.928928646349496e-07,
+      "loss": 0.7716,
+      "step": 4711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.909702731704347e-07,
+      "loss": 0.7633,
+      "step": 4712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.890513443241706e-07,
+      "loss": 0.7925,
+      "step": 4713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.871360788352708e-07,
+      "loss": 0.7363,
+      "step": 4714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.852244774414472e-07,
+      "loss": 0.7476,
+      "step": 4715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833165408789975e-07,
+      "loss": 0.8667,
+      "step": 4716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.814122698828017e-07,
+      "loss": 0.7453,
+      "step": 4717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795116651863363e-07,
+      "loss": 0.7743,
+      "step": 4718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.776147275216624e-07,
+      "loss": 0.7811,
+      "step": 4719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.757214576194291e-07,
+      "loss": 0.7676,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.738318562088695e-07,
+      "loss": 0.3711,
+      "step": 4721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7194592401780926e-07,
+      "loss": 0.8081,
+      "step": 4722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7006366177265504e-07,
+      "loss": 0.7772,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6818507019840584e-07,
+      "loss": 0.7634,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.663101500186418e-07,
+      "loss": 0.8168,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6443890195553065e-07,
+      "loss": 0.7778,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.625713267298271e-07,
+      "loss": 0.7454,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.607074250608701e-07,
+      "loss": 0.749,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.588471976665831e-07,
+      "loss": 0.7661,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5699064526347423e-07,
+      "loss": 0.3149,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5513776856663803e-07,
+      "loss": 0.7174,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5328856828975144e-07,
+      "loss": 0.7638,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.514430451450735e-07,
+      "loss": 0.7795,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.496011998434546e-07,
+      "loss": 0.8187,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4776303309431836e-07,
+      "loss": 0.72,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.459285456056772e-07,
+      "loss": 0.7635,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.440977380841294e-07,
+      "loss": 0.7817,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.422706112348474e-07,
+      "loss": 0.7524,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.404471657615894e-07,
+      "loss": 0.7664,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3862740236670254e-07,
+      "loss": 0.831,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.368113217511072e-07,
+      "loss": 0.7563,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.349989246143049e-07,
+      "loss": 0.7867,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3319021165438515e-07,
+      "loss": 0.6756,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3138518356801494e-07,
+      "loss": 0.7805,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2958384105043824e-07,
+      "loss": 0.3186,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.277861847954867e-07,
+      "loss": 0.7365,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.259922154955665e-07,
+      "loss": 0.8076,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2420193384166585e-07,
+      "loss": 0.7818,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.224153405233533e-07,
+      "loss": 0.8201,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2063243622877394e-07,
+      "loss": 0.7683,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.188532216446539e-07,
+      "loss": 0.7897,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1707769745629954e-07,
+      "loss": 0.7152,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.153058643475927e-07,
+      "loss": 0.7267,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1353772300099625e-07,
+      "loss": 0.8392,
+      "step": 4754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1177327409755e-07,
+      "loss": 0.7859,
+      "step": 4755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100125183168702e-07,
+      "loss": 0.82,
+      "step": 4756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0825545633715216e-07,
+      "loss": 0.7514,
+      "step": 4757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0650208883516787e-07,
+      "loss": 0.7566,
+      "step": 4758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0475241648626817e-07,
+      "loss": 0.7744,
+      "step": 4759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.030064399643774e-07,
+      "loss": 0.817,
+      "step": 4760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.012641599419975e-07,
+      "loss": 0.7321,
+      "step": 4761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9952557709020957e-07,
+      "loss": 0.7949,
+      "step": 4762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9779069207866563e-07,
+      "loss": 0.7522,
+      "step": 4763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960595055755967e-07,
+      "loss": 0.7611,
+      "step": 4764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.943320182478105e-07,
+      "loss": 0.8162,
+      "step": 4765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.926082307606849e-07,
+      "loss": 0.7682,
+      "step": 4766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.908881437781764e-07,
+      "loss": 0.7935,
+      "step": 4767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.891717579628185e-07,
+      "loss": 0.725,
+      "step": 4768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8745907397571226e-07,
+      "loss": 0.7292,
+      "step": 4769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.857500924765367e-07,
+      "loss": 0.8194,
+      "step": 4770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8404481412354646e-07,
+      "loss": 0.7417,
+      "step": 4771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8234323957357044e-07,
+      "loss": 0.7688,
+      "step": 4772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8064536948200335e-07,
+      "loss": 0.8371,
+      "step": 4773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7895120450282207e-07,
+      "loss": 0.8196,
+      "step": 4774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7726074528857235e-07,
+      "loss": 0.7286,
+      "step": 4775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.755739924903712e-07,
+      "loss": 0.3429,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7389094675790995e-07,
+      "loss": 0.7904,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.722116087394534e-07,
+      "loss": 0.763,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7053597908183525e-07,
+      "loss": 0.7742,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.688640584304626e-07,
+      "loss": 0.8063,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.671958474293147e-07,
+      "loss": 0.6911,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6553134672093873e-07,
+      "loss": 0.7861,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6387055694645624e-07,
+      "loss": 0.7192,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.622134787455567e-07,
+      "loss": 0.7914,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6056011275650284e-07,
+      "loss": 0.7513,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5891045961612525e-07,
+      "loss": 0.7675,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.572645199598257e-07,
+      "loss": 0.7564,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.556222944215748e-07,
+      "loss": 0.7638,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5398378363391327e-07,
+      "loss": 0.8027,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5234898822795074e-07,
+      "loss": 0.7784,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.507179088333645e-07,
+      "loss": 0.3599,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.490905460784022e-07,
+      "loss": 0.7947,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474669005898812e-07,
+      "loss": 0.8474,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4584697299318236e-07,
+      "loss": 0.8216,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4423076391225885e-07,
+      "loss": 0.3172,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4261827396963267e-07,
+      "loss": 0.7102,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.410095037863881e-07,
+      "loss": 0.6846,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3940445398217837e-07,
+      "loss": 0.7777,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3780312517522896e-07,
+      "loss": 0.8208,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3620551798232423e-07,
+      "loss": 0.8044,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3461163301881874e-07,
+      "loss": 0.8084,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3302147089863566e-07,
+      "loss": 0.8226,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.314350322342619e-07,
+      "loss": 0.783,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.298523176367463e-07,
+      "loss": 0.7669,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.282733277157124e-07,
+      "loss": 0.7777,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.266980630793426e-07,
+      "loss": 0.7724,
+      "step": 4806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2512652433438266e-07,
+      "loss": 0.854,
+      "step": 4807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.235587120861505e-07,
+      "loss": 0.7664,
+      "step": 4808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.219946269385221e-07,
+      "loss": 0.7372,
+      "step": 4809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2043426949393995e-07,
+      "loss": 0.8259,
+      "step": 4810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.188776403534133e-07,
+      "loss": 0.7922,
+      "step": 4811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1732474011651026e-07,
+      "loss": 0.7805,
+      "step": 4812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.15775569381368e-07,
+      "loss": 0.7745,
+      "step": 4813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.142301287446836e-07,
+      "loss": 0.7812,
+      "step": 4814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1268841880171987e-07,
+      "loss": 0.7768,
+      "step": 4815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1115044014629835e-07,
+      "loss": 0.7733,
+      "step": 4816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0961619337080974e-07,
+      "loss": 0.7985,
+      "step": 4817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0808567906620123e-07,
+      "loss": 0.7978,
+      "step": 4818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.065588978219869e-07,
+      "loss": 0.775,
+      "step": 4819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.050358502262407e-07,
+      "loss": 0.7817,
+      "step": 4820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.035165368655979e-07,
+      "loss": 0.7946,
+      "step": 4821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.020009583252581e-07,
+      "loss": 0.771,
+      "step": 4822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0048911518897774e-07,
+      "loss": 0.8207,
+      "step": 4823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9898100803908316e-07,
+      "loss": 0.7846,
+      "step": 4824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.974766374564497e-07,
+      "loss": 0.759,
+      "step": 4825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9597600402052286e-07,
+      "loss": 0.8118,
+      "step": 4826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.944791083093068e-07,
+      "loss": 0.7586,
+      "step": 4827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9298595089936265e-07,
+      "loss": 0.7564,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9149653236581257e-07,
+      "loss": 0.7924,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9001085328234426e-07,
+      "loss": 0.7274,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8852891422119777e-07,
+      "loss": 0.3443,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8705071575317436e-07,
+      "loss": 0.7577,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8557625844763846e-07,
+      "loss": 0.7256,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8410554287251037e-07,
+      "loss": 0.7843,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.826385695942657e-07,
+      "loss": 0.8499,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.81175339177947e-07,
+      "loss": 0.7839,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7971585218715014e-07,
+      "loss": 0.3378,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7826010918402646e-07,
+      "loss": 0.7602,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.768081107292919e-07,
+      "loss": 0.8229,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.753598573822136e-07,
+      "loss": 0.7495,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7391534970062195e-07,
+      "loss": 0.7989,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.724745882408997e-07,
+      "loss": 0.8075,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.710375735579884e-07,
+      "loss": 0.7982,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.69604306205391e-07,
+      "loss": 0.7936,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.681747867351558e-07,
+      "loss": 0.7535,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6674901569790137e-07,
+      "loss": 0.3146,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6532699364279155e-07,
+      "loss": 0.7215,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639087211175528e-07,
+      "loss": 0.8211,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6249419866846257e-07,
+      "loss": 0.7555,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6108342684035837e-07,
+      "loss": 0.8352,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.596764061766299e-07,
+      "loss": 0.7551,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5827313721922377e-07,
+      "loss": 0.8181,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5687362050863974e-07,
+      "loss": 0.7387,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.554778565839333e-07,
+      "loss": 0.7807,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.540858459827167e-07,
+      "loss": 0.7401,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5269758924115205e-07,
+      "loss": 0.8317,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.513130868939584e-07,
+      "loss": 0.7847,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.499323394744091e-07,
+      "loss": 0.3255,
+      "step": 4858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4855534751432985e-07,
+      "loss": 0.3595,
+      "step": 4859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4718211154409866e-07,
+      "loss": 0.7451,
+      "step": 4860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.458126320926513e-07,
+      "loss": 0.7564,
+      "step": 4861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4444690968747243e-07,
+      "loss": 0.7661,
+      "step": 4862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.430849448546013e-07,
+      "loss": 0.8008,
+      "step": 4863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4172673811862724e-07,
+      "loss": 0.7022,
+      "step": 4864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4037229000269835e-07,
+      "loss": 0.6855,
+      "step": 4865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3902160102850845e-07,
+      "loss": 0.7546,
+      "step": 4866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3767467171630587e-07,
+      "loss": 0.7702,
+      "step": 4867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3633150258489223e-07,
+      "loss": 0.7531,
+      "step": 4868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3499209415161928e-07,
+      "loss": 0.6682,
+      "step": 4869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3365644693238987e-07,
+      "loss": 0.7601,
+      "step": 4870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3232456144165916e-07,
+      "loss": 0.7438,
+      "step": 4871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3099643819243346e-07,
+      "loss": 0.734,
+      "step": 4872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2967207769626797e-07,
+      "loss": 0.3259,
+      "step": 4873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2835148046327138e-07,
+      "loss": 0.7771,
+      "step": 4874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2703464700210343e-07,
+      "loss": 0.757,
+      "step": 4875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.257215778199673e-07,
+      "loss": 0.7336,
+      "step": 4876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2441227342262616e-07,
+      "loss": 0.7782,
+      "step": 4877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2310673431438668e-07,
+      "loss": 0.7741,
+      "step": 4878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2180496099810655e-07,
+      "loss": 0.7032,
+      "step": 4879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2050695397519252e-07,
+      "loss": 0.7819,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.192127137456035e-07,
+      "loss": 0.8156,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1792224080784408e-07,
+      "loss": 0.8444,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1663553565896888e-07,
+      "loss": 0.835,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1535259879458258e-07,
+      "loss": 0.7873,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1407343070883546e-07,
+      "loss": 0.7961,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1279803189443117e-07,
+      "loss": 0.3334,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.115264028426156e-07,
+      "loss": 0.8361,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.102585440431859e-07,
+      "loss": 0.7687,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.089944559844892e-07,
+      "loss": 0.7953,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.077341391534138e-07,
+      "loss": 0.7138,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0647759403540247e-07,
+      "loss": 0.7641,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.052248211144392e-07,
+      "loss": 0.7537,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0397582087306244e-07,
+      "loss": 0.7901,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0273059379234739e-07,
+      "loss": 0.803,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.014891403519248e-07,
+      "loss": 0.7134,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0025146102996772e-07,
+      "loss": 0.7483,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9901755630319708e-07,
+      "loss": 0.7812,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.977874266468771e-07,
+      "loss": 0.8194,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9656107253482325e-07,
+      "loss": 0.8161,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.953384944393921e-07,
+      "loss": 0.7173,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.941196928314859e-07,
+      "loss": 0.8019,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9290466818055575e-07,
+      "loss": 0.7663,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.916934209545962e-07,
+      "loss": 0.7749,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9048595162014293e-07,
+      "loss": 0.7736,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8928226064228505e-07,
+      "loss": 0.8634,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8808234848464723e-07,
+      "loss": 0.8045,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8688621560940644e-07,
+      "loss": 0.7846,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.856938624772775e-07,
+      "loss": 0.7805,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8450528954752412e-07,
+      "loss": 0.7456,
+      "step": 4909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8332049727795008e-07,
+      "loss": 0.8081,
+      "step": 4910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8213948612490706e-07,
+      "loss": 0.7943,
+      "step": 4911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.809622565432878e-07,
+      "loss": 0.7869,
+      "step": 4912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.797888089865274e-07,
+      "loss": 0.754,
+      "step": 4913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7861914390660762e-07,
+      "loss": 0.6977,
+      "step": 4914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7745326175405143e-07,
+      "loss": 0.7436,
+      "step": 4915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7629116297792292e-07,
+      "loss": 0.8285,
+      "step": 4916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7513284802583296e-07,
+      "loss": 0.7196,
+      "step": 4917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7397831734393022e-07,
+      "loss": 0.7906,
+      "step": 4918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.728275713769112e-07,
+      "loss": 0.7724,
+      "step": 4919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.716806105680102e-07,
+      "loss": 0.3282,
+      "step": 4920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7053743535900502e-07,
+      "loss": 0.3378,
+      "step": 4921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.693980461902167e-07,
+      "loss": 0.6573,
+      "step": 4922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6826244350050424e-07,
+      "loss": 0.8427,
+      "step": 4923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.671306277272755e-07,
+      "loss": 0.78,
+      "step": 4924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6600259930646955e-07,
+      "loss": 0.8199,
+      "step": 4925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.648783586725744e-07,
+      "loss": 0.7331,
+      "step": 4926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.637579062586192e-07,
+      "loss": 0.3212,
+      "step": 4927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6264124249616764e-07,
+      "loss": 0.7947,
+      "step": 4928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6152836781532898e-07,
+      "loss": 0.7852,
+      "step": 4929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604192826447548e-07,
+      "loss": 0.7754,
+      "step": 4930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5931398741163228e-07,
+      "loss": 0.7259,
+      "step": 4931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5821248254168976e-07,
+      "loss": 0.8011,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.571147684591978e-07,
+      "loss": 0.7429,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5602084558696827e-07,
+      "loss": 0.7721,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5493071434634631e-07,
+      "loss": 0.7939,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5384437515722384e-07,
+      "loss": 0.7574,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5276182843802722e-07,
+      "loss": 0.8238,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5168307460572517e-07,
+      "loss": 0.7658,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.506081140758231e-07,
+      "loss": 0.8184,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4953694726236646e-07,
+      "loss": 0.8232,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4846957457793965e-07,
+      "loss": 0.7689,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4740599643366605e-07,
+      "loss": 0.7232,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4634621323920571e-07,
+      "loss": 0.7008,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4529022540275884e-07,
+      "loss": 0.7614,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4423803333106334e-07,
+      "loss": 0.78,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4318963742939396e-07,
+      "loss": 0.801,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4214503810156543e-07,
+      "loss": 0.781,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4110423574992705e-07,
+      "loss": 0.8106,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4006723077536922e-07,
+      "loss": 0.7944,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3903402357731688e-07,
+      "loss": 0.8135,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.380046145537328e-07,
+      "loss": 0.3709,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3697900410111987e-07,
+      "loss": 0.7149,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.359571926145109e-07,
+      "loss": 0.8558,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3493918048748224e-07,
+      "loss": 0.7957,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3392496811214573e-07,
+      "loss": 0.761,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3291455587914558e-07,
+      "loss": 0.7935,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3190794417766606e-07,
+      "loss": 0.7181,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3090513339542698e-07,
+      "loss": 0.7258,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2990612391868163e-07,
+      "loss": 0.798,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.289109161322233e-07,
+      "loss": 0.7937,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2791951041937755e-07,
+      "loss": 0.6997,
+      "step": 4961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.26931907162009e-07,
+      "loss": 0.8657,
+      "step": 4962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2594810674051217e-07,
+      "loss": 0.7878,
+      "step": 4963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2496810953382177e-07,
+      "loss": 0.732,
+      "step": 4964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2399191591940585e-07,
+      "loss": 0.797,
+      "step": 4965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2301952627326696e-07,
+      "loss": 0.7893,
+      "step": 4966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2205094096994442e-07,
+      "loss": 0.7869,
+      "step": 4967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2108616038250976e-07,
+      "loss": 0.7911,
+      "step": 4968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2012518488256908e-07,
+      "loss": 0.779,
+      "step": 4969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.191680148402663e-07,
+      "loss": 0.767,
+      "step": 4970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.182146506242754e-07,
+      "loss": 0.7403,
+      "step": 4971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1726509260180596e-07,
+      "loss": 0.8144,
+      "step": 4972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1631934113860322e-07,
+      "loss": 0.7889,
+      "step": 4973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1537739659894353e-07,
+      "loss": 0.7305,
+      "step": 4974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1443925934563894e-07,
+      "loss": 0.7212,
+      "step": 4975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1350492974003368e-07,
+      "loss": 0.8039,
+      "step": 4976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1257440814200548e-07,
+      "loss": 0.7889,
+      "step": 4977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1164769490996762e-07,
+      "loss": 0.7803,
+      "step": 4978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1072479040086348e-07,
+      "loss": 0.8104,
+      "step": 4979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.098056949701709e-07,
+      "loss": 0.3466,
+      "step": 4980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0889040897190006e-07,
+      "loss": 0.7839,
+      "step": 4981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0797893275859339e-07,
+      "loss": 0.7231,
+      "step": 4982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0707126668132894e-07,
+      "loss": 0.79,
+      "step": 4983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0616741108971373e-07,
+      "loss": 0.358,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0526736633188595e-07,
+      "loss": 0.8336,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.043711327545216e-07,
+      "loss": 0.7818,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0347871070282345e-07,
+      "loss": 0.7849,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0259010052052876e-07,
+      "loss": 0.747,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0170530254990485e-07,
+      "loss": 0.3442,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0082431713175467e-07,
+      "loss": 0.7524,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.994714460540455e-08,
+      "loss": 0.8238,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.907378530872202e-08,
+      "loss": 0.7862,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.820423957810022e-08,
+      "loss": 0.8025,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.733850774846231e-08,
+      "loss": 0.7676,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.647659015326716e-08,
+      "loss": 0.7912,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.561848712450029e-08,
+      "loss": 0.7606,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.476419899267953e-08,
+      "loss": 0.7666,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.39137260868539e-08,
+      "loss": 0.7375,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.306706873460136e-08,
+      "loss": 0.8133,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.222422726203218e-08,
+      "loss": 0.7625,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.138520199378664e-08,
+      "loss": 0.7692,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.054999325303182e-08,
+      "loss": 0.7617,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.971860136146926e-08,
+      "loss": 0.7858,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.889102663932836e-08,
+      "loss": 0.7728,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.806726940536747e-08,
+      "loss": 0.7814,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.724732997687501e-08,
+      "loss": 0.7662,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.643120866967059e-08,
+      "loss": 0.7594,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.561890579810051e-08,
+      "loss": 0.7753,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.481042167504228e-08,
+      "loss": 0.7755,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.400575661190235e-08,
+      "loss": 0.7661,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.320491091861393e-08,
+      "loss": 0.7216,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.24078849036436e-08,
+      "loss": 0.8107,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.161467887398022e-08,
+      "loss": 0.2998,
+      "step": 5013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.082529313514942e-08,
+      "loss": 0.3242,
+      "step": 5014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.003972799119908e-08,
+      "loss": 0.8067,
+      "step": 5015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.925798374470606e-08,
+      "loss": 0.7485,
+      "step": 5016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.848006069677839e-08,
+      "loss": 0.7963,
+      "step": 5017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.770595914705081e-08,
+      "loss": 0.8086,
+      "step": 5018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.693567939368484e-08,
+      "loss": 0.7561,
+      "step": 5019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.616922173337315e-08,
+      "loss": 0.7859,
+      "step": 5020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.540658646133292e-08,
+      "loss": 0.3227,
+      "step": 5021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.464777387130917e-08,
+      "loss": 0.7702,
+      "step": 5022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.38927842555781e-08,
+      "loss": 0.7474,
+      "step": 5023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.314161790494157e-08,
+      "loss": 0.7475,
+      "step": 5024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.239427510872476e-08,
+      "loss": 0.7066,
+      "step": 5025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.165075615478745e-08,
+      "loss": 0.666,
+      "step": 5026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.091106132951054e-08,
+      "loss": 0.3065,
+      "step": 5027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.0175190917805e-08,
+      "loss": 0.7792,
+      "step": 5028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.944314520310747e-08,
+      "loss": 0.7768,
+      "step": 5029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.871492446738348e-08,
+      "loss": 0.8432,
+      "step": 5030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.799052899112201e-08,
+      "loss": 0.8043,
+      "step": 5031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.726995905334099e-08,
+      "loss": 0.7984,
+      "step": 5032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.655321493158396e-08,
+      "loss": 0.7632,
+      "step": 5033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.584029690192228e-08,
+      "loss": 0.7826,
+      "step": 5034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.513120523895189e-08,
+      "loss": 0.7807,
+      "step": 5035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.442594021579429e-08,
+      "loss": 0.7878,
+      "step": 5036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.372450210409997e-08,
+      "loss": 0.7654,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.302689117404281e-08,
+      "loss": 0.755,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.233310769432455e-08,
+      "loss": 0.7495,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.164315193216919e-08,
+      "loss": 0.7274,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.095702415333193e-08,
+      "loss": 0.7822,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.027472462208806e-08,
+      "loss": 0.7653,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.959625360124177e-08,
+      "loss": 0.8172,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.892161135212071e-08,
+      "loss": 0.7744,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8250798134580345e-08,
+      "loss": 0.7891,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7583814206998436e-08,
+      "loss": 0.7815,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.692065982627837e-08,
+      "loss": 0.7933,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6261335247850265e-08,
+      "loss": 0.7684,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.560584072566766e-08,
+      "loss": 0.7961,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.495417651220747e-08,
+      "loss": 0.7948,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.430634285847558e-08,
+      "loss": 0.3472,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.366234001399795e-08,
+      "loss": 0.788,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.302216822682726e-08,
+      "loss": 0.8172,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.238582774354073e-08,
+      "loss": 0.8116,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.175331880923895e-08,
+      "loss": 0.3073,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.112464166754705e-08,
+      "loss": 0.7766,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0499796560614656e-08,
+      "loss": 0.7702,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.98787837291137e-08,
+      "loss": 0.8255,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.926160341224284e-08,
+      "loss": 0.7068,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.864825584772303e-08,
+      "loss": 0.7915,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.803874127179753e-08,
+      "loss": 0.7807,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.74330599192363e-08,
+      "loss": 0.7463,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6831212023330516e-08,
+      "loss": 0.7203,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6233197815894747e-08,
+      "loss": 0.7774,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.563901752726918e-08,
+      "loss": 0.8,
+      "step": 5065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.504867138631519e-08,
+      "loss": 0.7641,
+      "step": 5066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4462159620418665e-08,
+      "loss": 0.7701,
+      "step": 5067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.387948245548668e-08,
+      "loss": 0.7936,
+      "step": 5068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3300640115950806e-08,
+      "loss": 0.7913,
+      "step": 5069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.272563282476494e-08,
+      "loss": 0.8571,
+      "step": 5070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.215446080340746e-08,
+      "loss": 0.7703,
+      "step": 5071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.158712427187683e-08,
+      "loss": 0.777,
+      "step": 5072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.102362344869826e-08,
+      "loss": 0.7736,
+      "step": 5073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.046395855091256e-08,
+      "loss": 0.77,
+      "step": 5074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.990812979409064e-08,
+      "loss": 0.767,
+      "step": 5075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.935613739232236e-08,
+      "loss": 0.7652,
+      "step": 5076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.880798155821763e-08,
+      "loss": 0.822,
+      "step": 5077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8263662502913136e-08,
+      "loss": 0.797,
+      "step": 5078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.772318043606671e-08,
+      "loss": 0.7455,
+      "step": 5079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.718653556585517e-08,
+      "loss": 0.7331,
+      "step": 5080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6653728098978716e-08,
+      "loss": 0.3629,
+      "step": 5081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.612475824066319e-08,
+      "loss": 0.8189,
+      "step": 5082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.559962619465118e-08,
+      "loss": 0.7876,
+      "step": 5083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5078332163209774e-08,
+      "loss": 0.8335,
+      "step": 5084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.456087634712724e-08,
+      "loss": 0.7123,
+      "step": 5085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.404725894571415e-08,
+      "loss": 0.7924,
+      "step": 5086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.353748015680003e-08,
+      "loss": 0.7872,
+      "step": 5087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.303154017673893e-08,
+      "loss": 0.8558,
+      "step": 5088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.252943920040497e-08,
+      "loss": 0.7786,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.203117742119344e-08,
+      "loss": 0.7363,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.153675503102194e-08,
+      "loss": 0.7295,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.104617222032813e-08,
+      "loss": 0.8367,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0559429178070867e-08,
+      "loss": 0.7863,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0076526091730174e-08,
+      "loss": 0.3406,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.959746314730616e-08,
+      "loss": 0.7453,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9122240529323443e-08,
+      "loss": 0.8215,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.865085842082227e-08,
+      "loss": 0.8184,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8183317003367406e-08,
+      "loss": 0.8018,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7719616457042576e-08,
+      "loss": 0.7809,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.725975696045269e-08,
+      "loss": 0.3482,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6803738690722724e-08,
+      "loss": 0.8004,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.635156182349885e-08,
+      "loss": 0.8298,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5903226532946192e-08,
+      "loss": 0.7679,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5458732991753278e-08,
+      "loss": 0.3267,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.501808137112649e-08,
+      "loss": 0.7066,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4581271840790067e-08,
+      "loss": 0.7961,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4148304568994974e-08,
+      "loss": 0.7376,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.371917972250559e-08,
+      "loss": 0.764,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3293897466609705e-08,
+      "loss": 0.8006,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2872457965115168e-08,
+      "loss": 0.7066,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.245486138034769e-08,
+      "loss": 0.7885,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2041107873155278e-08,
+      "loss": 0.775,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1631197602903775e-08,
+      "loss": 0.7673,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1225130727480224e-08,
+      "loss": 0.8096,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0822907403289516e-08,
+      "loss": 0.7591,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.042452778525772e-08,
+      "loss": 0.7464,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.002999202682987e-08,
+      "loss": 0.7282,
+      "step": 5117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9639300279968855e-08,
+      "loss": 0.8267,
+      "step": 5118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.925245269515985e-08,
+      "loss": 0.7783,
+      "step": 5119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8869449421405895e-08,
+      "loss": 0.7776,
+      "step": 5120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8490290606230088e-08,
+      "loss": 0.7535,
+      "step": 5121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.811497639567228e-08,
+      "loss": 0.7602,
+      "step": 5122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.77435069342935e-08,
+      "loss": 0.7566,
+      "step": 5123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7375882365175956e-08,
+      "loss": 0.7895,
+      "step": 5124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7012102829915278e-08,
+      "loss": 0.7738,
+      "step": 5125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6652168468631602e-08,
+      "loss": 0.8186,
+      "step": 5126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6296079419960697e-08,
+      "loss": 0.7837,
+      "step": 5127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5943835821059516e-08,
+      "loss": 0.7832,
+      "step": 5128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5595437807601756e-08,
+      "loss": 0.8002,
+      "step": 5129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5250885513781177e-08,
+      "loss": 0.7967,
+      "step": 5130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4910179072309406e-08,
+      "loss": 0.7906,
+      "step": 5131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.457331861441702e-08,
+      "loss": 0.7614,
+      "step": 5132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4240304269854676e-08,
+      "loss": 0.8222,
+      "step": 5133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3911136166889772e-08,
+      "loss": 0.3199,
+      "step": 5134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3585814432308664e-08,
+      "loss": 0.7212,
+      "step": 5135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3264339191416675e-08,
+      "loss": 0.7317,
+      "step": 5136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2946710568035869e-08,
+      "loss": 0.7392,
+      "step": 5137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2632928684510603e-08,
+      "loss": 0.3179,
+      "step": 5138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2322993661699756e-08,
+      "loss": 0.3527,
+      "step": 5139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2016905618982277e-08,
+      "loss": 0.7811,
+      "step": 5140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1714664674253861e-08,
+      "loss": 0.7772,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1416270943930275e-08,
+      "loss": 0.7571,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1121724542945133e-08,
+      "loss": 0.8243,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.08310255847488e-08,
+      "loss": 0.7993,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0544174181310596e-08,
+      "loss": 0.8125,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0261170443117696e-08,
+      "loss": 0.758,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.982014479176239e-09,
+      "loss": 0.808,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.706706397007993e-09,
+      "loss": 0.7441,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.435246302655909e-09,
+      "loss": 0.8294,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.167634300678573e-09,
+      "loss": 0.7945,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.903870494151312e-09,
+      "loss": 0.3634,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.643954984670633e-09,
+      "loss": 0.7625,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.387887872347567e-09,
+      "loss": 0.6942,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.135669255813217e-09,
+      "loss": 0.7971,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.887299232214318e-09,
+      "loss": 0.7592,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.642777897216568e-09,
+      "loss": 0.7837,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.4021053450035185e-09,
+      "loss": 0.703,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.165281668274348e-09,
+      "loss": 0.7518,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.932306958248314e-09,
+      "loss": 0.803,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7031813046603004e-09,
+      "loss": 0.7368,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.477904795764156e-09,
+      "loss": 0.807,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.256477518328252e-09,
+      "loss": 0.7013,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.038899557642142e-09,
+      "loss": 0.746,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.825170997509899e-09,
+      "loss": 0.7177,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.615291920254562e-09,
+      "loss": 0.7842,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.409262406715909e-09,
+      "loss": 0.7758,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.207082536249353e-09,
+      "loss": 0.668,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.008752386730375e-09,
+      "loss": 0.7956,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.814272034551204e-09,
+      "loss": 0.7951,
+      "step": 5169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.623641554618585e-09,
+      "loss": 0.7706,
+      "step": 5170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.43686102035934e-09,
+      "loss": 0.7608,
+      "step": 5171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.253930503714809e-09,
+      "loss": 0.8451,
+      "step": 5172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0748500751475184e-09,
+      "loss": 0.8141,
+      "step": 5173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.899619803631183e-09,
+      "loss": 0.7544,
+      "step": 5174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7282397566618114e-09,
+      "loss": 0.3136,
+      "step": 5175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5607100002488236e-09,
+      "loss": 0.7355,
+      "step": 5176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3970305989217134e-09,
+      "loss": 0.751,
+      "step": 5177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2372016157244946e-09,
+      "loss": 0.742,
+      "step": 5178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0812231122179238e-09,
+      "loss": 0.8104,
+      "step": 5179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9290951484828302e-09,
+      "loss": 0.8256,
+      "step": 5180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.780817783113454e-09,
+      "loss": 0.7952,
+      "step": 5181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.636391073221889e-09,
+      "loss": 0.8016,
+      "step": 5182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4958150744380795e-09,
+      "loss": 0.7651,
+      "step": 5183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3590898409076025e-09,
+      "loss": 0.7702,
+      "step": 5184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2262154252938873e-09,
+      "loss": 0.7254,
+      "step": 5185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0971918787759948e-09,
+      "loss": 0.7785,
+      "step": 5186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9720192510497283e-09,
+      "loss": 0.8563,
+      "step": 5187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.850697590329853e-09,
+      "loss": 0.7675,
+      "step": 5188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7332269433445458e-09,
+      "loss": 0.7637,
+      "step": 5189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.619607355339836e-09,
+      "loss": 0.7081,
+      "step": 5190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5098388700807154e-09,
+      "loss": 0.8411,
+      "step": 5191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4039215298466968e-09,
+      "loss": 0.8319,
+      "step": 5192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3018553754329255e-09,
+      "loss": 0.8089,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.203640446153509e-09,
+      "loss": 0.812,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1092767798381866e-09,
+      "loss": 0.8024,
+      "step": 5195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0187644128334395e-09,
+      "loss": 0.7191,
+      "step": 5196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.321033800013813e-10,
+      "loss": 0.7833,
+      "step": 5197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.492937147208668e-10,
+      "loss": 0.7554,
+      "step": 5198
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.703354488897141e-10,
+      "loss": 0.3576,
+      "step": 5199
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.952286129191521e-10,
+      "loss": 0.8356,
+      "step": 5200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.239732357393724e-10,
+      "loss": 0.787,
+      "step": 5201
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.565693447950882e-10,
+      "loss": 0.7039,
+      "step": 5202
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.930169660488649e-10,
+      "loss": 0.8347,
+      "step": 5203
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.333161239788997e-10,
+      "loss": 0.7925,
+      "step": 5204
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7746684158124216e-10,
+      "loss": 0.7394,
+      "step": 5205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2546914036646335e-10,
+      "loss": 0.7741,
+      "step": 5206
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7732304036298673e-10,
+      "loss": 0.7825,
+      "step": 5207
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.330285601159776e-10,
+      "loss": 0.6968,
+      "step": 5208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9258571668512306e-10,
+      "loss": 0.8186,
+      "step": 5209
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5599452564907246e-10,
+      "loss": 0.8255,
+      "step": 5210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.232550011021072e-10,
+      "loss": 0.8293,
+      "step": 5211
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.436715565303012e-11,
+      "loss": 0.6935,
+      "step": 5212
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.933100042960661e-11,
+      "loss": 0.7641,
+      "step": 5213
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8146545076344084e-11,
+      "loss": 0.7934,
+      "step": 5214
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.081379775005111e-11,
+      "loss": 0.7425,
+      "step": 5215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7332765128719174e-11,
+      "loss": 0.7728,
+      "step": 5216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.703452404861367e-12,
+      "loss": 0.8243,
+      "step": 5217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9258632866225867e-12,
+      "loss": 0.7949,
+      "step": 5218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6442,
+      "step": 5219
+    },
+    {
+      "epoch": 1.0,
+      "step": 5219,
+      "total_flos": 1.0199943414939648e+16,
+      "train_loss": 0.8014171527340669,
+      "train_runtime": 71013.2852,
+      "train_samples_per_second": 9.408,
+      "train_steps_per_second": 0.073
+    }
+  ],
+  "max_steps": 5219,
+  "num_train_epochs": 1,
+  "total_flos": 1.0199943414939648e+16,
+  "trial_name": null,
+  "trial_params": null
+}